that does not require online interaction with PRNs. Through a .... number of PR interferers at receiver v, and the probability density function (pdf) of the distance ...
Channel Access Mechanisms and Protocols for Opportunistic Cognitive Radio Networks
Item type
text; Electronic Dissertation
Authors
Bany Salameh, Haythem Ahmad Mohammed
Publisher
The University of Arizona.
Rights
Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
Downloaded
25-Feb-2018 22:20:09
Link to item
http://hdl.handle.net/10150/193865
CHANNEL ACCESS MECHANISMS AND PROTOCOLS FOR OPPORTUNISTIC COGNITIVE RADIO NETWORKS by Haythem Ahmad Bany Salameh
A Dissertation Submitted to the Faculty of the DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING In Partial Fulfillment of the Requirements For the Degree of DOCTOR OF PHILOSOPHY In the Graduate College THE UNIVERSITY OF ARIZONA
2009
2 THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE
As members of the Final Examination Committee, we certify that we have read the dissertation prepared by Haythem Ahmad Bany Salameh entitled Channel Access Mechanisms and Protocols for Opportunistic Cognitive Radio Networks and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of Doctor of Philosophy. Date: 17 April 2009 Dr. Marwan Krunz
Date: 17 April 2009 Dr. Srinivasan Ramasubramanian
Date: 17 April 2009 Dr. Loukas Lazos
Date: 17 April 2009
Date: 17 April 2009
Final approval and acceptance of this dissertation is contingent upon the candidate’s submission of the final copies of the dissertation to the Graduate College. I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement. Date: 17 April 2009 Dissertation Director: Dr. Marwan Krunz
3
STATEMENT BY AUTHOR
This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library. Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.
SIGNED: Haythem Ahmad Bany Salameh
4 ACKNOWLEDGEMENTS
I would like to thank many people, without whom this dissertation would not be possible. First of all, I wish to express my gratitude to my dissertation advisor, Dr. Marwan Krunz, for his guidance, support, and assistance in developing and refining my research. His technical advice and encouragement have provided me with the skills needed to carry me through the completion of this dissertation. Special thanks to Dr. Ossama Younis for long discussions about various aspects of my research, a major part of my dissertations is a result of our collaboration. I would also like to extend my thanks to Professor Srinivasan Ramasubramanian and Professor Loukas Lazos who generously agreed to participate in my dissertation committee, and provided me with valuable feedback. I also wish to thank all my colleagues at the Wireless and Advance Networking group at The University of Arizona. In addition, I would like to thank Dr. Tao Shu for helping me in the initial stage of my mathematical research. Far too many friends to mention individually have assisted in so many ways during my work. A huge thank-you goes to all of my friends. Finally, I am forever indebted to my parents, brother, and sisters for their unbounded support and encouragement and for always being there for me. My father’s great advice, my mother’s prayers, and their love and dedication provided the foundations for this work.
5
DEDICATION
I would like to dedicate this dissertation to my parents. Their constant love and caring are the reasons for where I am and what I am. My gratitude and my love to them are beyond words. I would also like to dedicate this dissertation to my home country Jordan, which I will serve until the last day of my life.
6
TABLE OF CONTENTS
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
CHAPTER 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 13 1.1 1.2
Motivation and General Scope . . . . . . . . . . . . . . . . . . . . . . 13 Main Contributions and Dissertation Overview . . . . . . . . . . . . . 16
CHAPTER 2 Dynamic Spectrum Access Protocol Without Power Mask Constraints . . . . . . . . . . . . . . . . . . . . 20 2.1
2.2 2.3 2.4
2.5 2.6
2.7
2.8
Introduction . . . . . . . . . . . . . . . . . . . 2.1.1 Goals and Contributions . . . . . . . . 2.1.2 Organization . . . . . . . . . . . . . . Related Work . . . . . . . . . . . . . . . . . . System Model . . . . . . . . . . . . . . . . . . Interference Analysis . . . . . . . . . . . . . . 2.4.1 Wireless Channel Model . . . . . . . . 2.4.2 PR-to-CR Interference . . . . . . . . . 2.4.3 PR-to-PR Interference . . . . . . . . . 2.4.4 Model Verification . . . . . . . . . . . Guaranteeing Outage Probability for PR users The COMAC Protocol . . . . . . . . . . . . . 2.6.1 Transmission Regions for a CR User . 2.6.2 Spectrum Access . . . . . . . . . . . . 2.6.3 Channel Assignment . . . . . . . . . . Performance Evaluation . . . . . . . . . . . . 2.7.1 Simulation Setup . . . . . . . . . . . . 2.7.2 Single-hop Scenarios . . . . . . . . . . 2.7.3 Multi-hop Scenarios . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
20 20 22 22 24 26 27 28 31 33 35 38 38 41 42 43 45 46 48 49
CHAPTER 3 Distance- and Traffic-Aware Channel Assignment in Cognitive Radio Networks . . . . . . . . . . . . 58 3.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
7 TABLE OF CONTENTS – Continued
3.2 3.3
3.4
3.5
3.6
3.7
3.8
3.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2 Goals and Contributions . . . . . . . . . . . . . . . . 3.1.3 Organization . . . . . . . . . . . . . . . . . . . . . . Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Network Model . . . . . . . . . . . . . . . . . . . . . 3.3.2 Analysis of the Average SINR . . . . . . . . . . . . . 3.3.3 Carrier Frequency and Distance Effects on Path Loss Optimal Channel Assignment Problem . . . . . . . . . . . . 3.4.1 CRN Transmission Requirements . . . . . . . . . . . 3.4.2 Maximizing the Utilization of Local Spectrum . . . . Distance-Dependent Channel Assignment Algorithm . . . . . 3.5.1 Spectrum Assignment for Known Traffic Profiles . . . 3.5.2 Spectrum Assignment for Unknown Traffic Profiles . 3.5.3 Complexity . . . . . . . . . . . . . . . . . . . . . . . 3.5.4 Examples . . . . . . . . . . . . . . . . . . . . . . . . DDMAC Protocol . . . . . . . . . . . . . . . . . . . . . . . . 3.6.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . . 3.6.2 Channel Access in DDMAC . . . . . . . . . . . . . . 3.6.3 Spatial Reuse and DDMAC . . . . . . . . . . . . . . 3.6.4 Worst-Case Scenarios for DDMAC . . . . . . . . . . Protocol Evaluation . . . . . . . . . . . . . . . . . . . . . . . 3.7.1 Single-hop Scenarios . . . . . . . . . . . . . . . . . . 3.7.2 Multi-hop Scenarios . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
58 60 62 62 64 64 67 67 69 69 69 70 71 72 75 75 76 77 78 81 82 84 85 87 91
CHAPTER 4 Cooperative Adaptive Spectrum Sharing for Throughput Enhancement in Cognitive Radio Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.1
4.2
4.3
Introduction . . . . . . . . . . . . . . . . . . . 4.1.1 Goals and Contributions . . . . . . . . 4.1.2 Organization . . . . . . . . . . . . . . Problem Formulation and Design Constraints 4.2.1 Network Model . . . . . . . . . . . . . 4.2.2 Feasibility Constraints . . . . . . . . . 4.2.3 Problem Statement . . . . . . . . . . . 4.2.4 Problem Formulation . . . . . . . . . . Optimal Channel Assignment . . . . . . . . . 4.3.1 Proposed Algorithm . . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
96 96 97 97 97 98 99 100 102 102
8 TABLE OF CONTENTS – Continued
4.4
4.5 4.6
4.7 4.8
4.3.2 Channel Access Protocol for Single-hop CRNs Distributed Channel Assignment for Multi-hop CRNs 4.4.1 Challenges . . . . . . . . . . . . . . . . . . . . 4.4.2 Channel Assignment . . . . . . . . . . . . . . 4.4.3 Channel Access Protocol . . . . . . . . . . . . Throughput Analysis . . . . . . . . . . . . . . . . . . Performance Evaluation . . . . . . . . . . . . . . . . 4.6.1 Simulation Setup . . . . . . . . . . . . . . . . 4.6.2 Single-hop Network . . . . . . . . . . . . . . . 4.6.3 Multi-hop Network . . . . . . . . . . . . . . . Related Work . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
104 106 107 107 108 111 112 113 114 115 116 118
CHAPTER 5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . 122 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
9
LIST OF FIGURES
1.1 1.2 2.1 2.2 2.3 2.4 2.5
2.6 2.7 2.8 2.9 2.10 2.11 2.12 2.13
A CRN environment containing one CRN of PDAs and 2 PRNs of WiMAx devices and cell phones. . . . . . . . . . . . . . . . . . . . . 14 Frequency-dependent power mask for a CR user. . . . . . . . . . . . 15 Opportunistic access environment containing one CRN and 3 PRNs. Illustrating the aggregate interference from PRs at receiver v. . . . Mean and variance verification when n = 4. . . . . . . . . . . . . . Interference model verification when n = 2. . . . . . . . . . . . . . . Scenarios in which a CR transmitter C can/cannot reuse the channels assigned to A. Solid circles indicate data-transmission ranges, while dashed circles indicate control-transmission ranges. . . . . . . . . . Performance of a PRN. . . . . . . . . . . . . . . . . . . . . . . . . . Channel usage for the CRN. . . . . . . . . . . . . . . . . . . . . . . Performance of the CRN. . . . . . . . . . . . . . . . . . . . . . . . . Impact of selecting p∗ . . . . . . . . . . . . . . . . . . . . . . . . . . Impact of activity profile on performance. . . . . . . . . . . . . . . Performance under skewed and uniform user deployments. . . . . . Illustration of channel-aware routing. . . . . . . . . . . . . . . . . End-to-end throughput of min-hop routing and CAR as a function of the packet generation rate. . . . . . . . . . . . . . . . . . . . . . .
Scenarios in which two CR transmissions can/cannot proceed simultaneously. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Example of an opportunistic CRN that coexists with two PRNs. . 3.3 Operating spectrum in the hybrid network. . . . . . . . . . . . . . 3.4 Path loss vs. carrier frequency for two path loss exponents (Da = 5 cm, Gt (f ) = Gr (f ) = 1). . . . . . . . . . . . . . . . . . . . . . . . 3.5 Four regions around a CR transmitter for assigning channels. . . . 3.6 Time diagram of pmf’s updating process. . . . . . . . . . . . . . . 3.7 Example that illustrates the channel assignment process in a dynamic CRN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8 Formats of DDMAC control packets. . . . . . . . . . . . . . . . . . 3.9 RTS-PCA-EPCA-DATA-ACK packet exchange. . . . . . . . . . . . 3.10 Scenarios in which a CR transmitter C can/cannot reuse the channels assigned to A. Solid circles indicate data-transmission ranges, whereas dashed circles indicate control-transmission ranges. . . . . .
24 . 26 . 34 . 35
. . . . . . . .
43 51 52 53 54 55 56 56
. 57
3.1
. 60 . 64 . 64 . 68 . 72 . 73 . 76 . 79 . 81
. 82
10 LIST OF FIGURES – Continued 3.11 Illustration of two worst-case scenarios in DDMAC. . . . . . . . . . 3.12 Throughput vs. λ for a small-scale network (comparison with the optimal scheme). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.13 Performance of a CRN. . . . . . . . . . . . . . . . . . . . . . . . . . 3.14 Per-user throughput and fairness Performance. . . . . . . . . . . . . 3.15 Performance of DDMAC. . . . . . . . . . . . . . . . . . . . . . . . . 3.16 Impact of inaccurate distance estimation on DDMAC. . . . . . . . . 3.17 End-to-end throughput vs. λ . . . . . . . . . . . . . . . . . . . . .
. 83 . . . . . .
86 88 92 93 94 95
4.1 4.2 4.3 4.4 4.5 4.6
. . . . .
103 105 107 110 112
Bipartite graph with M = N = 3. . . . . . . . . . . . . . . . . . . Basic operation of AW-MAC . . . . . . . . . . . . . . . . . . . . . . Basic operation of 2-radio AW-MAC. . . . . . . . . . . . . . . . . . Time diagram of control packet exchange in WFC-MAC. . . . . . . Basic operation of the distributed spectrum access scheme. . . . . Maximum achievable throughput (in packet/ (Tdata + M Tctrl )) vs. total number of frequency channels (control-packet size = 120 bits). 4.7 CRN performance in single-hop scenarios. . . . . . . . . . . . . . . 4.8 Fairness index in single-hop scenarios (2-radio AW-MAC depicted similar behavior as AW-MAC). . . . . . . . . . . . . . . . . . . . . 4.9 CR channel usage in single-hop scenarios (2-radio AW-MAC depicted similar behavior as AW-MAC). . . . . . . . . . . . . . . . . . . . . 4.10 CRN performance in multi-hop scenarios. . . . . . . . . . . . . . . .
. 113 . 119 . 120 . 120 . 121
11
LIST OF TABLES
2.1 2.2
Simulation parameters used to verify interference analysis. . . . . . . 33 Mean and Variance of the PR-to-CR Interference for n = 4. . . . . . . 34
3.1 3.2
Summary Of Notations Used In This Chapter. . . . . . . . . . . . . . 66 Performance of DDMAC at the optimal α as a function of Twin . . . . 90
12 ABSTRACT
High traffic load over the unlicensed portion of the radio spectrum (a.k.a., ISM bands) along with inefficient usage of the licensed spectrum gave impetus for a new paradigm in spectrum allocation, whose main purpose is to improve spectrum efficiency through opportunistic access. Cognitive radios (CRs) have been proposed as a key enabling technology for such paradigm. Operating a CR network (CRN) without impacting the performance of licensed (primary) users requires new protocols for information exchange as well as mathematical tools to optimize the controllable parameters of the CRN. In this dissertation, we target the design of such protocols. First, we develop a distributed CRN MAC (COMAC) protocol that enables unlicensed users to dynamically utilize the spectrum while limiting the interference they inflict on primary (PR) users. The main novelty in COMAC lies in not assuming a predefined CR-to-PR power mask and not requiring coordination with PR users. Second, we propose a novel distance-dependent MAC protocol for CRNs in which each CR is equipped with multiple transceivers. Our protocol (called DDMAC) attempts to maximize the CRN throughput by following a novel probabilistic channel assignment mechanism. This mechanism exploits the dependence between the signal’s attenuation model and the transmission distance while considering the traffic profile. We show that through its distance- and traffic-aware, DDMAC significantly improves network throughput. Finally, we address the problem of assigning channels to CR transmissions, assuming one transceiver per CR. The main goal of our design is to maximize the CRN throughput with respect to both spectrum assignment and transmission power. Specifically, we present centralized and distributed solutions that leverage the unique capabilities of CRs. Compared with previously proposed protocols, our schemes are shown to significantly improve network throughput.
13
CHAPTER 1 Introduction
1.1 Motivation and General Scope The tremendous growth of wireless applications and services is straining the effectiveness of conventional static spectrum planning policies. Recent field studies conducted by the FCC and other agencies revealed vast temporal and geographical variations in the utilization of the licensed spectrum, ranging from 15% to 85% [6, 9, 27]. Such measurements prompted regulators to push for a more efficient and adaptive spectrum allocation policy. As a result, the FCC is currently considering revising its regulations to allow for opportunistic (on demand) access to the spectrum. Cognitive radio (CR) is a technology that promises to offer such an opportunistic capability without affecting primary radio (PR) users. CRs are mainly characterized by their cognitive capability and reconfigurability. The cognitive capability provides spectrum awareness, whereas reconfigurability enables a CR user to dynamically adapt its operating parameters to the surrounding RF environment. More specifically, the CR can be programmed to transmit and receive over widely-separated frequency bands, adapt its transmit power, and determine its optimal transmission strategy [25, 39, 41, 49, 22]. A typical CR network (CRN) environment consists of a number of primary radio networks (PRNs) that are licensed to operate over orthogonal spectrum bands1 and one (secondary) CRN. All networks co-exist within the same geographical space. Figure 1.1 shows a conceptual view of a multi-hop CRN environment, where a group of PDAs use CRs to exploit the under-utilized spectrum in a WiMAX network (PRN 1) and a network of cell phones (PRN 2). Figure 1.1 also shows the operating spectrum of such a hybrid environment. PR users that belong to a given 1
The terms band and channel will be used interchangeably in this dissertation.
14 PRN are licensed to operate over a given portion of the spectrum. CR users form an opportunistic network. They can dynamically access the entire spectrum that is available to all PRNs. Specifically, a CRN is expected to operate over a set of widely-separated non-contiguous frequency bands. Another important characteristic of a CRN is that users must operate using a regulated transmission power to avoid degrading the performance of PR users. The above peculiar characteristics of CRNs distinguish them from traditional multi-channel wireless networks, and call for new protocols for information exchange as well as mathematical tools or algorithms that use this information to efficiently utilize the available spectrum and improve the overall spectrum efficiency. In this dissertation, we develop novel spectrum access/sharing protocols and algorithms that attempt to effectively address the unique resource constraints and the dynamic operating environment of CRNs.
WIMAX router WIMAX router WIMAX router
WIMAX router WIMAX router
WIMAX router
WIMAX router
PSD PRN 1
PRN 2
f CRN
Figure 1.1: A CRN environment containing one CRN of PDAs and 2 PRNs of WiMAx devices and cell phones. Coexistence between CR and PR users in the same area poses a new challenge regarding protecting the PR communications. In a mixed CRN/PRNs environment, three types of interference need to be accounted for: CR-to-CR interference, PR-to-CR interference, and CR-to-PR interference. The latter is the most critical, because of its direct influence on PRNs’ performance. CR transmission powers over the various PR bands need to be regulated such that PR receptions are not nega-
15
CR Power Mask
Power
Pmask = ( P
(1 )
mask
, P (2 ) mask ,....., P ( N ) mask )
………. f1
f2
fN
Frequency
Figure 1.2: Frequency-dependent power mask for a CR user.
tively affected by CR transmissions. To address this issue, a frequency-dependent power mask on the CR transmissions is often adopted (see Figure 1.2). This mask reflects the maximum permissible transmission power vector of a CR user. Most existing spectrum-sharing protocols for CRNs focus on identifying and avoiding interference with PR transmissions (e.g., [69, 29, 32]). In these protocols, CR users can access only “white spaces”, which refer to frequency bands with no PR activity. This type of opportunistic spectrum sharing requires a CR user to perform sensing before attempting to transmit. Accordingly, the CR user identifies whether or not a given PR band is idle. If so, the CR user can transmit. Under the assumption of perfect sensing, this approach provides non-overlapping (collision-free) band sharing between CR and PR users. In this case, the corresponding power regulating scheme can be viewed as a binary-type power mask over each PR band. However, the efficiency of this type of schemes depends heavily on the ability to predict/detect the presence of PR signals over various bands, i.e., the scheme requires a robust algorithm for determining white spaces. Binary-type power masks can also result in non-optimal spectrum utilization. It has been shown that allowing CR users to exploit both white (idle) and grey (partially occupied) bands give much better spectrum utilization [19]. In the second chapter of the dissertation, we derive a neighborhood-dependent multi-level adaptive power mask on CR transmissions that ensures a statistical (soft) guarantee on the performance of PRNs, and that is shown to provide better spectrum utilization. Several channel access (MAC) protocols for multi-channel networks and CRNs
16 have been proposed in literature (e.g., [59, 52, 55, 50, 13, 70, 16, 38, 68]). These schemes are often based on a greedy strategy, which selects the best available channel (or channels) for a given transmission. However, as explained in Chapter 3, when such a greedy assignment is employed in a CRN, the blocking probability for CR transmissions can increase, leading to a reduction in the CR network throughput. As shown later in the dissertation, the greedy assignment approach is not well suited to the unique features and application requirements of CRNs. Hence, new spectrum sharing algorithms are needed to improve the performance of CRNs. Accordingly, one of the main goals of this dissertation is to provide efficient spectrum access/sharing algorithms that improve the CRN performance. Specifically, in the third chapter of this dissertation, we develop novel access/sharing algorithms assuming that each CR is equipped with multiple radios and is capable of multi-channel transmissions and receptions. Using multiple transceivers greatly simplifies the task of MAC design. However, it comes at the cost of extra hardware. Therefore, in this dissertation, we also investigate the design of a dynamic channel assignment for single-transceiver CRNs2 . While channel access issues such as hidden terminals, exposed terminals, and connectivity can be easily overcome with multiple transceivers, these issues are not trivial in the case of a single half-duplex transceiver. Hence, we provide channel access solutions for single-transceiver CRNs that overcome the aforementioned issues and that improve network throughput. 1.2 Main Contributions and Dissertation Overview The main contributions of this dissertation are as follows: • In Chapter 2, we investigate a statistical approach for dynamic spectrum access and radio resource management (RRM) in opportunistic CRNs. We propose a distributed MAC protocol, called COMAC, for such networks. COMAC enables unlicensed users to dynamically utilize the available spectrum while limiting the imposed interference on PR users. COMAC is novel in three as2
A single-transceiver CR user is less complex and cheaper, which makes it more practical.
17 pects. First, it does not require CR users to coordinate their operation with PR users. Second, it does not assume any predefined CR-to-PR power mask, and thus can exploit the available spectrum more efficiently. Third, it provides PR users with a statistical guarantee on the fraction of time that their reception may be corrupted by CR users. To avoid corrupting PR receptions, COMAC requires computing the maximum power that a CR transmission can use based on current network conditions. We show how to compute this maximum power by deriving statistical models for the PR-to-CR and PR-to-PR interference. To the best of our knowledge, COMAC is the first MAC protocol for CRNs that provides a soft guarantee on the performance of PR users without assuming a predefined interference power mask. Simulation experiments illustrate that COMAC improves spectrum utilization and statistically guarantees the performance of PR users under various user deployment models and traffic loads. • One of the key challenges to enabling multi-hop CR communications is how to make CR nodes coordinate their channel access in order to maximize the CRN throughput. Although many MAC protocols have been proposed for traditional multi-channel wireless networks, these protocols are not well suited to the unique characteristics of CRNs. Specifically, typical multi-channel MAC protocols assume that the frequency channels are adjacent and that there are no constraints on the transmission power. However, a CRN may operate over a wide range of frequencies, and a power mask is often enforced on the transmission power of a CR user to avoid corrupting the transmissions of spectrumlicensed PR users. In Chapter 3, we propose a novel distance-dependent MAC protocol for CRNs, called DDMAC, that leverages the unique capabilities of CRs and the peculiar characteristics of their operating environment. DDMAC attempts to maximize the CRN throughput. It uses a novel probabilistic channel assignment mechanism that exploits the dependence between the signal’s attenuation model and the transmission distance while considering the traffic
18 profile. The protocol allows a pair of CR users to communicate over a channel that may not be optimal from one user’s perspective, but that allows more concurrent transmissions to take place, especially under moderate and high loads. To the best of our knowledge, DDMAC is the first CRN MAC protocol that utilizes the radio propagation characteristics to improve the overall network throughput. Simulation results are used to demonstrate the throughput gain under the DDMAC protocol. • Many spectrum access/sharing algorithms for CRNs have been designed assuming multiple transceivers per CR user. In practice, such an assumption may not hold due to cost consideration. In Chapter 4, we address the problem of assigning channels to CR transmissions, assuming a single half-duplex transceiver per CR. The primary goal of our design is to maximize (with respect to both spectrum assignment and transmission power) the number of simultaneous CR transmissions. Energy conservation is also addressed, but as a secondary objective. The problem is posed as a utility maximization problem subject to a target rate demand and interference constraints. In general, this optimization problem is known to be NP-hard. However, for the single-transceiver CRNs, this problem can be solved in polynomial time and is not NP-hard. We present centralized and distributed solutions that leverage the unique capabilities of CRs. We first consider a single-hop CRN, for which we introduce a CSMA-like MAC protocol that uses an access window (AW) for exchanging control information prior to data transmissions. This approach allows us to realize a centralized algorithm in a distributed manner. We then develop a distributed MAC protocol (WFC-MAC) for a multi-hop CRN. WFC-MAC improves the CRN throughput through cooperative channel assignment. It incurs low processing overhead. We compare the performance of our schemes with CSMA/CA variants. The results show that our schemes significantly decrease the blocking rate of CR transmissions, and hence improves the network throughput.
19 The conclusion are drawn in Chapter 5.
20
CHAPTER 2 Dynamic Spectrum Access Protocol Without Power Mask Constraints
2.1 Introduction In an environment where several licensed PRNs are operating, a network of CR users that co-exist with PR users needs to exploit the underutilized portion of the spectrum. In this case, the crucial challenge is how to allow CR users to share the licensed spectrum with PR users without degrading the performance of the PR users. In this chapter, we advocate a statistical approach by which CR users are allowed to communicate opportunistically while probabilistically guaranteeing the performance of PR users. Our key performance measure is the “outage probability” (pout ) of a PR user, defined as the fraction of time during which the total interference power at a PR receiver exceeds the maximum tolerable interference. 2.1.1 Goals and Contributions The contributions of this chapter are as follows. • We develop stochastic models for the PR-to-PR and the PR-to-CR interference under a Rayleigh fading channel model. In the course of constructing these models, we derive closed-form expressions for the mean and variance of the total interference at a receiving node. Closed-form expressions for the characteristic function (CF) of such interference are also obtained for integervalued path loss exponents. Numerical and simulation results indicate that the resulting distribution of the total interference is well approximated by a lognormal function.
21 • We derive an expression for the maximum allowable powers for a CR transmission based on the developed models. The computed powers provide a statistical guarantee on the PRN performance. • We design a distributed CSMA/CA-based MAC protocol for CRNs (COMAC) that does not require online interaction with PRNs. Through a local exchange of control messages, COMAC enables a pair of CR users to select the minimum number of channels to use according to the surrounding interference and the rate demand of the CR transmitter. CR users can communicate over both unused and partially used licensed channels without needing to coordinate with PRNs. Most importantly, COMAC functions without assuming a predefined CR-to-PR interference power mask. For a given PR spectrum band, the power mask is defined as the maximum permissible transmission power of a CR user over that band. This mask is needed to ensure that the CR transmission does not cause unacceptable interference to neighboring PR users operating on the same band. • We implement a channel-aware routing (CAR) mechanism for CRNs that extends the well-known minimum hop routing (min-hop) approach for improving the perceived throughput. It should be noted that interference modeling in wireless networks was previously studied under the assumption of an infinite user population, operating within an unbounded field. No multipath fading was considered. For example, [40, 62, 61] assumed that nodes are distributed according to a Poisson distribution, and characterized the distribution of the interference for an idealized infinite-size network operating within an infinite field. No multipath fading was considered. It is easy to show that their model leads to total interference whose mean and variance are infinite. Such a model cannot be applied in our work, as we consider a finite number of users.
22 2.1.2 Organization The rest of this chapter is organized as follows. The related previous work is reviewed in Section 2.2. Section 2.3 introduces our system model and assumptions. In Section 2.4, we develop stochastic models for the PR-to-PR and PR-to-CR interference, and verify these models. Section 2.5 shows how to provide a statistical guarantee on the performance of PR users. We introduce our proposed MAC protocol in Section 2.6, and evaluate its performance in Section 2.7. Finally, Section 2.8 gives concluding remarks. 2.2 Related Work One of the key challenges to enabling CR communications is how to perform opportunistic medium access control while limiting the interference imposed on PR users. Recently, several attempts were made to develop MAC protocols for CRNs (e.g., [59, 52, 55, 50, 13, 70, 16, 38, 68]). Existing work on spectrum sharing/access protocols can be classified according to their architecture (centralized or decentralized), spectrum allocation behavior (cooperative or non-cooperative), and spectrum access technique (overlay or underlay) [9]. The IEEE 802.22 working group is in the process of standardizing a centralized MAC protocol that enables spectrum reuse by CR users operating on the TV broadcast bands[1, 20]. In [13, 70, 16] centralized protocols were proposed for coordinating spectrum access. For an ad hoc CRN without centralized control, it is desirable to have a distributed MAC protocol that allows every CR user to individually access the spectrum. DC-MAC [46] is a cross-layer distributed scheme for spectrum allocation/sensing. It provides an optimization framework based on partially observable Markov decision processes, with no insights into protocol design, implementation and performance. In [45], the authors proposed a decentralized channel-sharing mechanism for CRNs based on a game-theoretic approach under both cooperative and non-cooperative scenarios. However, they did not propose an operational MAC protocol. No guarantee on the performance of PRNs was considered.
23 The FCC defined the interference temperature model [26], which provides a metric for measuring the interference experienced by PR users. Clancy [19] used this model to select an optimal bandwidth/power assignment for CR users. However, no operational protocol was proposed. It is worth mentioning that due to the lack of specific technical rules to implement the interference temperature model, the FCC has abandoned this model in 2007 [7]. In [24], the authors developed a power control approach for CR systems based on spectrum sensing side information. The objective of such an approach is to mitigate the interference to a PR user from CR transmissions. However, no operational protocol was presented. The setup was also limited in that a CR user cannot transmit over a partially occupied licensed channel. Three spectrum sharing techniques were proposed and compared in [40]: spreading-based underlay, interference avoidance overlay, and spreading-based underlay with interference avoidance. The metric of interest in the comparison was pout . The treatment did not provide guarantees on the performance of PR users. Furthermore, interference statistics were used assuming an unbounded region for outage probability analysis. In addition, The outage probability analysis ignored the interference caused by other PR users. Before closing, we note that a number of multi-channel contention-based MAC protocols were previously proposed in the context of CRNs (e.g., [59, 55, 50, 52]). The CRN MAC protocol in [59] jointly optimizes the multi-channel power/rate assignment, assuming a given power mask on CR transmissions. How to determine an appropriate power mask remains an open issue. DDMAC [52] is a spectrum-sharing protocol for CRNs that attempts to maximize the CRN throughput through a novel probabilistic channel assignment algorithm that exploits the dependence between the signal’s attenuation model and the transmission distance while considering the prevailing traffic and interference conditions. AS-MAC [55] is a spectrum-sharing protocol for CRNs that coexist with a GSM network. CR users select channels based on the CRN’s control exchanges and GSM broadcast information. Explicit coordination with the PRNs is required. In [68], the authors developed a spectrum aware MAC protocol for CRNs (CMAC). CMAC enables opportunistic access and
24 sharing of the available white spaces in the TV spectrum by adaptively allocating the spectrum among contending users. To the best of our knowledge, COMAC is the first CRN MAC protocol that provides a soft guarantee on the performance of PR users without assuming a predefined interference power mask. 2.3 System Model We consider a hybrid network, consisting of M different PRNs and one CRN. The M + 1 networks co-exist within the same geographical space. Figure 2.1 shows a conceptual view of the networks under consideration.
Figure 2.1: Opportunistic access environment containing one CRN and 3 PRNs.
The PRNs are licensed to operate on different, non-overlapping frequency bands. PR users that belong to a given PRN share the same licensed spectrum. In reality,
25 a PRN may occupy multiple, non-contiguous frequency bands. Such a PRN can be easily captured in our setup by using multiple virtual PRNs, each operating over its own band. For the ith PRN, we denote its carrier frequency, channel bandwidth, and maximum transmission power by fi , Bi , and Pt (i) , respectively. To make our analysis tractable, we model the locations of users in the ith PRN as a homogeneous Poisson random variable on a disk area of parameter (density) ρi . This model was previously used in [59, 62, 61, 19]. In our simulations, we relax this assumption and consider arbitrary deployment scenarios. Each user in the ith PRN acts as an ON/OFF source. We define the “activity factor” αi as the fraction of time that a user in the ith PRN is ON [58, 18, 64]. The source is further characterized by the distribution of its ON and OFF periods that are both taken to be exponential. Therefore, the traffic correlation is captured using a two-state Markov model. The appropriateness of the 2-state ON/OFF model has been demonstrated in several previous works, e.g., [64, 53, 61, 62, 59]. In essence, the ON/OFF behavior is attributed to the bursty nature of many types of network traffic, including data transfer and VBR video streaming. Remark: Estimating the activity behavior of PR users was investigated in [18, 64]. Specifically, in [18], αi was estimated by maintaining a run length of the idle/busy period for each channel. Whenever the idle duration is ended by a PR transmission, the run length is recorded in a circular buffer. For our purposes, we assume that a similar mechanism for estimating αi is in place. In section 2.7, we evaluate the impact of inaccurately estimating αi . CR users (unlicensed users) can opportunistically access the entire spectrum that is available to all PRNs. Each CR user is equipped with nr radio transceivers, 1 ≤ nr ≤ M , that can be used simultaneously. The CR user has a wideband sensing capability with a narrowband resolution. Such capability can be achieved using a wideband antenna, a power amplifier, and adaptive filters [9]. Thus, a CR user can sense the available spectrum in one shot (simultaneously sensing several GHz-wide bands [14]) and estimate the instantaneous interference over each band. Such advanced spectrum sensing technology is readily available through a DSP technique
26 called cyclostationary feature detection [15, 9, 14]. Alternatively, a sequential partial sensing approach can be employed at the cost of negligible switching/sensing overhead [15, 50]. It is worth mentioning that off-the-shelf wireless cards (e.g., ICS572 products [10]) can readily serve as a fully functional wideband multi-channel CR interface. Such an interface enables a CR user to perform analysis of the RF spectrum (i.e., sensing) in real time. For a given CR transmission, the aggregate rate used over the assigned channels is fixed for the duration of that transmission. Our protocol assumes the availability of a pre-specified control channel of Fourier bandwidth Bc , where Bc Bi , i = 1, . . . , M . Such a channel is not necessarily dedicated to the CRN. It may, for example, be one of the unlicensed ISM bands. Note that the existence of a dedicated common control channel is a characteristic of many MAC protocols proposed for CRNs (e.g., [59, 52, 55, 68, 46]).
PRi1
PRi 2
v rc ..
.
PRij
Figure 2.2: Illustrating the aggregate interference from PRs at receiver v.
2.4 Interference Analysis We develop stochastic models for the PR-to-PR and the PR-to-CR interference. Note that the M different PRNs are licensed to operate over non-overlapping frequency bands (orthogonal bands). This fact ensures that the interference measured at a PR receiver over a band is only due to transmitters operating on that band (no mutual interference between different PRNs). Thus, without loss of generality, we
27 consider one of the PRNs (PRN i)1 , where i = 1, 2, . . . , M , and determine the total interference at a receiver v (primary or cognitive) from only the PR users of that PRN (see Fig. 2.2). Let do (i) be the close-in distance for the ith PRN, defined as the distance from a transmitter after which the RF channel can be approximated by the free-space model [47]. Because of the highly nonlinear attenuation behavior of typical RF channels, we assume that the interference contributed to v by PR (i)
users that lie outside a disk of radius rc (rc do ) is negligible. This is inline with [67], in which rc was used to indicate the distance of the “first-tier interferers”. Our simulations (Section 2.7) relax this assumption and account for all sources of interference, including those that are very far away from the receiver. We now introduce the propagation model. Using this model, we compute the number of PR interferers at receiver v, and the probability density function (pdf) of the distance between v and an interferer. Then, we derive the characteristic function for the PR interference. 2.4.1 Wireless Channel Model We consider a Rayleigh fading model to describe the channel between any two users. Specifically, for a transmitter-receiver separation d, the received power over the ith channel2 is given by: Pr (i)
where Po
(i)
=
(i)
(i)
Pt (i) Gt Gr li2 (i) (4πdo )2
(i)
=
Po(i)
d do
(i)
−n
ξ (i) ,
d ≥ do (i)
(2.1) (i)
is the path loss of the close-in distance do , Pt (i) is the (i)
transmission power, Gt is the antenna gain of the transmitter, Gr is the antenna gain of the receiver, li is the wavelength of fi , n is the path loss exponent, and ξ (i) is a normalized random variable that represents the power gain of the fading process. For a Rayleigh fading, ξ (i) is exponentially distributed; Pr(ξ (i) ≤ y) = 1 − e−y [47]. 1
Note that we consider a generic PRN. The derived expressions then apply to any of the M
PRNs by using the PRN’s associated parameters (density, activity factor, etc.). 2 Because of our assumed 1-to-1 mapping between the PRNs and the channels, the index i is used to refer to either one.
28 2
, D, li }, where D is the antenna According to [47], do (i) is given by do (i) = max{ 2D li length. In practice, do (i) is of the same order of magnitude as the node’s dimensions. For example, for a mobile phone operating at 900 MHz with D = 5 cm, do (i) = 33 cm. For an 802.11 WLAN card operating in the 2.4 GHz band (5cm-long antenna), do (i) = 12 cm. Accordingly, it is reasonable to assume that the probability that d is less than do (i) is very small. 2.4.2 PR-to-CR Interference (i)
We now derive the statistics of the aggregate PR interference PP R−CR on a given CR receiver. Approximately, this is equal to the sum of the interference powers of all active PR transmitters within radius rc of the CR receiver, i.e., (i)
PP R−CR ≈
X
(i)
Pr,j ,
(2.2)
j
(i)
where Pr,j is the received power associated with the jth active PR transmitter of the ith channel, and the summation is carried out over all active PR transmitters in Rc . Before proceeding further, we need to determine the distribution of the distance between a PR transmitter and a CR/PR receiver. Let Ki denote the number of potential PR interferers within a disk area Rc , where Rc = πrc2 . Since the locations of the PR users are modeled as a homogeneous Poisson process, the probability of Ki = ki is given by: e−ρi Rc (ρi Rc )ki Pr{Ki = ki } = , ki = 0, 1, 2, . . . . ki !
(2.3)
The distribution of the locations of the ki interferers is that of ki independent and identically distributed (i.i.d.) uniform random variables3 [62]. Thus, the pdf of the distance r between a receiver at the center of Rc and an interferer that is randomly 3
In our simulations (Section 2.7), we study the performance of our protocol under both uniform
and skewed user distributions.
29 located inside Rc is given by [62]: fR (r) =
2r , rc2
r ≤ rc
0,
(2.4)
otherwise.
We assume that different interfering transmissions experience i.i.d. fades. This assumption is justified by noting that the distance between any two PR interferers is typically much larger than the wavelength of the carrier frequency of a PRN (e.g., (i)
for a PRN operating at 900 MHz, li = 33 cm). Let Pr,j|y denote the received power (i)
(i)
Pr,j conditioned on ξj = y. Because the probability that the distance between a PR user and a CR user is less than do (i) is approximately zero, the characteristic (i)
function (CF) of Pr,j|y can be written as: (i)
def
(i)
φP (i) (ω) = E[ejωPr,j |ξj = y] r,j|y
≈
Z
rc (i)
(i)
e
jωPo
r (i) do
−n
y
fR (r)dr.
(2.5)
do
By substituting (2.4) into (2.5) and algebraically manipulating the result, we obtain: (i)
φP (i) (ω) = 2 r,j|y
do rc
!2 Z
rc (i) do
(i) −n x
xejωPo
y
dx.
(2.6)
1
(i)
The CF of Pr,j , φP (i) (ω), can be obtained by removing the conditioning in (2.6) and r,j
algebraically manipulating the result, leading to: Z ∞ φP (i) (ω) fξ(i) (y)dy φP (i) (ω) = r,j j r,j|y 0 Z ∞ φP (i) (ω) e−y dy = r,j|y 0 ! (i) 2 Z rc (i) do xn+1 do = 2 dx. (i) rc xn − jwPo 1
(2.7)
Recall that the number of PR users in the plane is Poisson distributed with mean of ρi users per unit area. Because each PR user behaves as an ON/OFF source with activity factor αi , the number of active PR transmitters in Rc (denoted by
30 Ni ) forms a Poisson random variable with mean of αi ρi active users per unit area. (i)
Conditioned on Ni = ni , the CF of PP R−CR is given by: i n i h (i) E ejωPP R−CR |Ni = ni ≈ φP (i) (ω) r,j ni ! 2 Z rc (i) n+1 (i) x do do dx . = 2 (i) rc xn − jwPo 1
(2.8)
The CF of the total PR-to-CR interference over channel i, φP (i)
(ω), can be
P R−CR
obtained by removing the conditioning in (2.8): ∞ X e−αi ρi Rc (αi ρi Rc )ni φP (i) (ω) = P R−CR ni ! ni =0 ni ! (i) 2 Z rc n+1 (i) x do do × 2 dx . (i) n rc x − jωPo 1
(2.9)
Observe that the integral term in (2.9), denoted by ICR,i (n, ω), is a function of n, ω, and fi . Thus, by summing the series, (2.9) can be rewritten as: φP (i) (ω) = P R−CR
exp αi ρi Rc 2
(i) do
rc
!2
ICR,i (n, ω) − 1.
(2.10)
For integer values of n, ICR,i (n, ω) has a closed-form solution [57], and thus, the CF in (2.10) has a closed-form expression. As an example, the expression of ICR,i (4, ω) is given by: "
2
#
q
(i)
ωPo 3 rc 1 − 1 + j2 ICR,i (4, ω) = × (i) 2 2 do s s " !# 2 ! j r j c tan−1 − tan−1 . (i) (i) (i) ωPo do ωPo (2.11) For the case of a non-integer n, ICR,i (n, ω) can be evaluated numerically.
31 Using the fact that
(i)
do rc
1, we arrive at the following approximate expressions (i)
for the mean and variance of PP R−CR : (i)
(i)
def
def
0
P P R−CR = E[PP R−CR ] = φP (i) (0) ≈ P R−CR i h (i) (i) 2 (i) 2 2παi ρi Po do rc 2−n −παi ρi do ) − 1 , 1≤n2
,
(2.12) and
σP2 (i)
def
(i)
= var(PP R−CR ) ≈ P R−CR h 2 i2 (i) (i) −παi ρi d(i) παi ρi o 2P d e , o o (n−1) h i (i) 2 2 (i) (i) ln 2παi ρi 2Po do e−παi ρi do
rc (i) do
n>1
(2.13)
, n = 1.
Note that the above approximations for the mean and the variance show no dependence on rc for n > 2. (i)
While a closed-form expression for the pdf of PP R−CR cannot be found, numerical inversion of the CF and empirical fitting of the simulated data (Section 2.4.4) show that this pdf is well approximated by the lognormal distribution. 2.4.3 PR-to-PR Interference In addition to estimating the PR-to-CR interference, our design requires a CR user to estimate the PR-to-PR interference so that an upper bound on the CR transmission power can be computed while providing a guarantee on pout for PR users. Let bi denote the minimum distance between a PR receiver and the nearest PR interferer4 . This value is transmission-technology dependent and is fixed for a given PRN. For example, in a cellular network in which adjacent cells do not use common frequencies, 4
Mitigating interference over band i, and consequently achieving successful communications,
necessitates imposing a minimum distance between active PR users.
32 bi is the minimum reuse distance, defined as the minimum distance between a base station of a cell and a mobile terminal of another non-adjacent cell that guarantees acceptable link quality. This value is easily shown to be equal to the diameter of a cell. To characterize the PR-to-PR interference, we use a similar methodology to that used in the previous section. We replace the lower integration limit in (2.5) by b i . The CF of PR-to-PR interference is thus given by: φP (i) (ω) = P R−P R
exp αi ρi Rc 2
where
(i) do
rc
def
IP R,i (n, ω) =
Z
!2
IP R,i (n, ω) − 1
rc (i) do bi (i) do
xn+1 (i)
xn − jωPo
dx.
(2.14)
(2.15)
Consequently, the mean and variance for the PR-to-PR interference are approximately given by: (i)
P P R−P R ≈ 2 (i) (i) 2 2παi ρi Po do e−παi ρi bi rc 2−n − bi 2−n , 1 ≤ n < 2 2−n (i) (2−n)(do ) h i 2 (i) (i) 2 n=2 2παi ρi Po do e−παi ρi bi ln rbci , 2 2−n (i) (i) 2παi ρi Po do e−παi ρi b2i b(i)i , n>2 n−2
(2.17)
σP2 (i) ≈ P R−P Rh i2 2(1−n) bi −παi ρi b2i παi ρi 2Po(i) d(i) e , n>1 o (i) (n−1) d h i2 o −παi ρi b2i 2παi ρi 2Po(i) d(i) ln rbci , n = 1. o e
(2.18)
do
and
Similar to the case of PR-to-CR interference, for integer values of n, the CF in (2.14) has a closed-form expression. Furthermore, we found that the lognormal function (i)
well approximates the distribution of PP R−P R .
33 2.4.4 Model Verification We now use MATLAB simulations to empirically verify the validity of the derived PR-to-CR interference model (results for the validation of the PR-to-PR interference model are similar). We consider a circular field of radius 100 meters in which four PRNs are uniformly distributed. The transmission power for a PR user is 1 Watt. The antenna length (D) is 5 cm. Time is divided into slots. At any given slot, each user in PRN i transmits with probability αi . Other PRN parameters are shown in Table 2.1. We set rc = 100 m.
First, we assess the goodness of the
Table 2.1: Simulation parameters used to verify interference analysis. PRN 1 2 3 4
fi 900 MHz 1.5 GHz 2.4 GHz 4.0 GHz
ki 300 400 400 200
αi 0.6 0.5 0.4 0.2
(i)
approximations in (2.12) and (2.13) for the mean and variance of PP R−CR . Table 2.2 shows the analytical approximations and the measured values for the means and variances, and the associated relative error (r ) when n = 4 (similar behavior was observed for n = 2). Each empirical value in the table is the average of 100 runs, each lasting for 100000 time slots. The results show that the derived expressions well approximate the measured statistics. Figures 2.3(a) and (b) report r as a function of the number of PR users (ki ) for the PR-to-CR interference mean and variance, respectively. The results show that the derived expressions well approximate the measured statistics, with r < 1%, irrespective of the number of PR users. To validate the conjecture that the distribution of the interference model is well approximated by a lognormal distribution, we compute the pdf of the interference for PRN 1 with n = 2 in two ways: by constructing the histogram of the simulated data and by numerically inverting the CF in (2.10). Figure 2.4(a) plots the empirical and numerically (1)
computed pdfs of PP R−CR against the theoretical lognormal distribution with mean and variance given in (2.12) and (2.13), respectively. Visual inspection of the figure indicates
34
Table 2.2: Mean and Variance of the PR-to-CR Interference for n = 4. PRN
Approx. 5.35 × 10−8 2.14 × 10−8 6.68 × 10−9 6.02 × 10−10
2
2
1.8
1.8 PRN 1 PRN 2 PRN 3 PRN 4
1.6 1.4 εr (Mean)
r (%) 0.269% 0.914% 0.829% 0.426%
1.2
1.4
1 0.8 0.6
1 0.8 0.6 0.4
0.2
0.2
200
r (%) 0.63% 0.92% 0.64% 0.58%
1.2
0.4
0
Variance Simulation 5.38 × 10−8 2.12 × 10−8 6.64 × 10−9 5.98 × 10−10
PRN 1 PRN 2 PRN 3 PRN 4
1.6
εr (variance)
1 2 3 4
Mean Simulation 1.26 × 10−5 5.02 × 10−6 1.60 × 10−6 1.43 × 10−7
Approx. 1.27 × 10−5 5.07 × 10−6 1.58 × 10−6 1.43 × 10−7
0 250
300 No. of PR users (k i)
350
400
(a) r vs. Ni (mean)
200
250
300 No. of PR users (k i)
350
400
(b) r vs. Ni (variance)
Figure 2.3: Mean and variance verification when n = 4. the adequacy of the lognormal distribution.
Figure 2.4(b) shows the probability plot of
the empirical data against three different distributions: Gamma, Weibull, and Lognormal5 . In this figure, only the plotted points that correspond to the lognormal form reasonably straight lines and follow the empirical distribution fairly closely. Similar behaviors were observed for other PRNs and different values of n. 5
The Data-probability plot is a graphical technique for assessing whether or not a data set
follows a given distribution. If the probability plot approximately forms a straight line, then the distribution of the empirical data is well approximated by the suggested distribution.
35
Probability plot for different distributions (n=2)
10000
f(PPR−CR)
6000
Probability
0.9999 0.9995 0.999 0.995 0.99
8000
0.95 0.9 0.75
4000
Empirical pdf Lognormal pdf Numerical pdf
2000
0
0
5
10
15 −4 x 10
PPR−CR
(a) Histogram
0.5
0.25 0.1 0.05 0.01 0.005 0.001 0.0005 0.0001
Gamma Weibull Lognormal
10−5
−4
10
PPR−CR
10−3
(b) Data-probability plot
Figure 2.4: Interference model verification when n = 2. 2.5 Guaranteeing Outage Probability for PR users Because the outage probability (pout ) is our primary performance metric, we statistically bound pout for each PRN. Our objective can be stated as follows: With probability 1 − β, where β 1, the transmissions of CR users should not disturb the reception of any PR user. Therefore, we require that pout ≤ β. To provide such a guarantee, we need to (i)
compute the maximum allowable transmission power PC,β that a CR transmitter can use over channel i such that all communicating PR users within the communication range of the transmitting CR are not impacted by this transmission with probability 1 − β. We enforce an exclusive channel occupancy policy on CR transmissions, whereby a channel occupied by a CR user cannot be simultaneously allocated to another CR user in the same vicinity (inline with the CSMA/CA mechanism). This policy ensures that the interference measured at a PR receiver is mainly due to at most one CR transmitter and to other PR transmitters. Accordingly, we compute an upper bound on the amount of interference that can be introduced by a CR transmitter over each channel. Consider the jth PR user of the ith PRN. With probability 1 − β, the following condition should be satisfied by every CR user: (i)
(i)
(i)
(i)
PP R−P R,j + gC,j PC,β ≤ PL , ∀j = 1, 2, . . . , and ∀i = 1, 2, . . . , M
(2.19)
36 (i)
where PP R−P R,j is the total PR-to-PR interference power measured at the jth PR receiver (i)
of the ith PRN, gC,j is the gain between a CR transmitter and the jth PR receiver, and (i)
(i)
PL is the interference power limit of a PR receiver in the ith PRN. The value of P L , which is sometimes referred to as the load or interference margin, is typically known (i)
for a given PRN (e.g., set by the FCC) [26, 19]. PL provides an upper bound on the potential interference that could be tolerated by a PR receiver (i.e., defined at the PR receiver not at the CR transmitter). Consequently, using this interference limit, we derive the maximum permissible transmission power of a CR user such that a given outage probability is guaranteed. (i)
Because we assume no active coordination between CR and PR users, g C,j is difficult (i)
to measure at a CR user. To proceed with our analysis, an estimate of g C,j that preserves (i)
(i)
the required bound on pout is needed to compute PC,β . The problem of selecting gC,j was (i)
studied in [24, 19]. The gC,j was selected based on the shortest distance between a PR receiver and a CR transmitter. In [19], a probabilistic argument was used for computing the shortest distance. In [24], the shortest distance was derived from the spectrum sensing side information measured at a CR user. We now derive the CDF of the distance between a CR transmitter and the closest active PR receiver. Based on such a CDF and following the same methodology in [19], we (i)
propose a mathematical formulation for selecting a value of gC,j that preserves the target pout . For a given PRN, let the distance between a CR transmitter located at the center of (i)
a disk of radius rc (rc do ) and the closest active PR receiver be denoted by Rmin . Then, Rmin = min{Rj : j ∈ Γ}, where Rj is a random variable representing the distance between a CR transmitter and the jth PR receiver, and Γ is the set of active PR receivers. According to (2.4), the CDF of Rj is given by FRj (r) =
r2 . rc2
Conditioning on Ni = ni , the
CDF of Rmin is given by: FRmin |Ni =ni (r) = 1 −
ni Y
1 − FRj (r)
j=1
"
=1− 1−
r rc
2 #n i
.
(2.20)
Given that Ni is a Poisson random variable (Section 2.4.2), the CDF of Rmin can be
37 obtained by removing the conditioning in (2.20) and algebraically manipulating the result: 2
FRmin (r) = 1 − e−αi ρi πr .
(2.21)
(i)
Let r∗ denote the distance used in setting gC,j in (2.19). This r ∗ can be selected based on a target percentage of FRmin (i.e., FRmin (r∗ ) = 1 − p∗ ). Formally, with probability p∗ , where p∗ is very close to one, the distance between a CR transmitter and its closest PR receiver is at least r ∗ . By substituting FRmin (r∗ ) = 1 − p∗ in (2.21) and solving for r ∗ , we obtain: r∗ =
s
− ln(p∗ ) . α i ρi π
(2.22)
Depending on the relative location of a PR receiver with respect to a CR transmitter, there are two possible scenarios where outage can occur at a PR receiver: 1. The PR receiver falls within a distance less than r ∗ from a CR transmitter. The likelihood of this scenario is 1 − p∗ . In this case, we conservatively set Pr[outage|r < r∗ ] ≈ 1. 2. The PR receiver is at a distance greater than r ∗ from a CR transmitter. The def
likelihood of this scenario is p∗ . In this case, let γ = Pr[outage|r > r ∗ ]. Accounting for the above two scenarios, the overall outage probability can be computed via a straightforward application of Bayes’s rule, i.e., pout = Pr[outage|r < r ∗ ]Pr[r < r ∗ ] + Pr[outage|r > r ∗ ]Pr[r > r ∗ ] = 1 × (1 − p∗ ) + γ × p∗ = 1 − (1 − γ)p∗ . Recall that we require pout ≤ β, which implies: 1−β γ ≤1− . p∗
(2.23)
(2.24)
Note that γ cannot be negative. Thus, for a valid bound on γ, the following constraint must be satisfied:
1−β p∗
< 1.
(2.25)
Equations (2.23) and (2.24) reveal that in order to preserve the required bound on p out (i.e., β), the condition in (2.19) should be satisfied by every CR user that is located at a
38 distance greater than r ∗ , with probability 1 − γ. To satisfy this condition with probability (i)
(i)
(i)
1 − γ, we compute the (1 − γ)-quantile of PP R−P R,j , denoted by Pγ . Because PP R−P R is approximately lognormally distributed, its (1 − γ)-quantile is given by: Pγ(i)
(σ
=e
P
Φ (i) P R−P R
−1 (1−γ))
(2.26)
where Φ−1 is the (1 − γ)-quantile of the standard normal distribution. (i)
By substituting Pγ in (2.19) and rearranging the equation, we obtain an upper bound on the interference that a CR transmitter is allowed to contribute to the ith PRN while ensuring pout ≤ β: (i) PC,β
(i)
(i)
≤
PL − P γ (i)
gC,j
,
i = 1, 2, . . . , M.
(2.27)
Accordingly, the maximum allowable transmission powers for a CR user over various → − (1) (2) (M ) channels are given by the vector P C,β = [PC,β , PC,β , . . . , PC,β ].
2.6 The COMAC Protocol COMAC is a distributed and asynchronous MAC protocol for ad hoc CRNs that uses the previous analysis to enable opportunistic CR communications while providing soft guarantees on the performance of PR users. The proposed protocol uses a contentionbased handshaking for exchange of control information. Before we describe the protocol’s operation in detail, we first define and compute the different transmission regions around a CR user A. These regions describe A’s “view” of its neighborhood.
2.6.1 Transmission Regions for a CR User Each CR user A is associated with a data region and a control region. Within these regions, other CR and PR users may exist. The data region of A is defined as the area in which A’s data transmission can be correctly decoded by another CR user. Let r data (A) be the radius of this region. With probability 1 − β, COMAC protects all PR receptions that are within distance rdata (A) and that share channels with A. The control region of A is defined as the region in which A’s control packets can be correctly decoded. Let the radius of this region be rctrl (A). As described below, COMAC requires rctrl (A) ≥ 2rdata (A) to enforce exclusive channel occupancy among different CR transmissions.
39 In computing A’s maximum transmission range over channel i (ai ), we assume channel stationarity for the duration of one data packet. The received SINR at channel i can be computed as: (i)
µi =
C(fi )a−n PA i (i)
(i)
(Pth + PP R−CR )
,
i = 1, . . . , M
(2.28)
(i)
where Pth is the measured thermal noise over channel i, PA is A’s transmission power, (i)
and C(fi ) is a frequency-dependent constant, given by C(fi ) =
(i)
Gt Gr li2 . (4π)2 (do (i) )2−n
According
to COMAC, CR users transmit data packets using the maximum allowable power vector → − → − ( P C,β ), derived in Section 2.5. Using P C,β and (2.28), the maximum distance at which a CR receiver can correctly decode A’s data transmission over channel i (a i ) is given by: v u (i) u C(fi ) PC,β n t , i = 1, . . . , M (2.29) ai = (i) (i) µ∗i (Pth + PP R−CR ) where µ∗i is the SINR threshold required at the CR receiver to achieve a target bit error
rate over channel i. In (2.29), the CR-to-CR interference is ignored because of the aforementioned exclu(i)
sive channel occupancy policy among CR users6 . The PR-to-CR interference PP R−CR is estimated by its average expected interference at a CR transmitter, given in (2.12). Note that neighboring CR users typically experience similar average PR-to-CR interference [9], i.e., they share a similar view of the spectrum conditions. Thus, it makes little difference whether the parameters are computed at the transmitter or at the receiver. Remark: In computing ai , we do not consider the expected worst-case PR-to-CR interference since it typically leads to small control and data transmission ranges, which might jeopardize the network connectivity. On the other hand, ignoring the PR-to-CR interference results in longer control ranges, leading to over-conservative channel assignment and consequently lower CRN throughput.
2
Let M be the set of all M channels. In COMAC, CR user A maintains a list of currently available channels, denoted by LAC(A), which consists of the channels in M that are not currently used by any of A’s CR neighbors. LAC(A) is determined through 6
Our simulations take into account the effect of the hidden-terminal problem due to imperfect
control by considering the interference from active neighboring CR transmissions that use common channels (if any).
40 the overheard control packets. We set A’s data transmission range to: rdata (A) =
min
j∈LAC(A)
aj .
(2.30)
We impose the following constraint on rctrl (A) to control the CR-to-PR interference: Proposition 1 If rctrl (A) ≥ 2 maxj∈LAC(A) aj , then there is no overlap between the data region of A and the data region of any other CR transmitter that overlaps with A in one or more data channels. Proof. By definition, for any channel j ∈ LAC(A), rdata (A) ≤ aj and rctrl (A) ≥ 2aj . Because of the exclusive channel occupancy, within rctrl (A) range no CR transmission other than A’s can take place over channel j. Thus, the distance between A and any other CR transmitter, say C, is at least 2aj . If C is outside the control region of A and wants to reuse channel j, it will choose its rdata (C) to be at most aj . According to the proposition, C will choose its rctrl (C) to be at least 2aj . Consequently, the data regions of A and C will not overlap, and only A’s transmission will cause interference to PR users located in A’s data region.
2
Remark: In general, the transmission range is a decreasing function of the transmission rate. Noting that the control channel requires a relatively low data rate, and consequently a low SINR threshold. Hence, the control range in Proposition 1 can be easily enforced through power control. Let Pctrl (A) be the minimum power level that is needed by a CR user A to support the range rctrl (A) over the control channel. In computing Pctrl (A), we account for the channel-specific RF attenuation and interference behavior. Formally, we set Pctrl (A) = µ∗c I (c) /(C(fc )rctrl (A)), where fc , µ∗c , C(fc ), and I (c) respectively denote the carrier frequency of the control channel, the SINR threshold required at the CR receiver to achieve a target bit error rate over the control channel, (c)
(c)
a frequency-dependent constant (C(fc ) = Gt Gr /(4π)2 fc2 (do (c) )2−n ), and the average noise-plus-interference power over the control channel. To reduce the required value of Pctrl (A), the control channel should be selected in the lower portions of the spectrum, where higher transmission ranges can be achieved [68].
2
In Section 2.7, we study the impact of different settings of rctrl (A) (as a function of rdata (A)) on the protocol’s performance.
41 2.6.2 Spectrum Access We propose a spectrum access mechanism that enables the CR transmitter and receiver to agree on the set of channels to use. This mechanism also ensures that with probability 1 − β the ensuing data transmission will not disturb any of the PR users in the vicinities of the CR transmitter and receiver. The spectrum access mechanism is described as follows. Suppose that CR user A has data to transmit to CR user B at a total data rate RA . This RA is supported through P the aggregate rate of all selected channels, i.e., RA ≤ m i=1 ri , where m is the number
of channels that will be assigned to A’s transmission, m ≤ nr , and ri is the data rate of
the ith selected channel. Let Ω(A, B) be the set of channels assigned to the transmission A → B, where m = kΩ(A, B)k. As described in Section 2.6.3, the selection of Ω(A, B) depends on spectrum state information (SSI). For now, it suffices to say that SSI includes T def the following information: (1) LAC(A, B) = LAC(A) LAC(B); (2) the instantaneous
interference level at B over each channel in LAC(A, B); and (3) the channel gain between A and B, computed using the received signal strength of A’s control packets. If A does not sense a carrier over the control channel for a randomly selected backoff
period, it computes its rdata (A) according to (2.29) and (2.30). It then sends a Request-toSend (RTS) message at power Pctrl (A) (computed according to Proposition 1). The RTS (i)
packet includes LAC(A), Pctrl (A), RA , and PC,β (A), ∀i ∈ LAC(A). The neighbors of A, other than B, that can correctly decode the RTS will stay silent until either they receive another control packet, denoted by DCTS (explained below), or the expected time-out for that packet expires. Upon receiving the RTS packet, B determines the SSI and proceeds with the channel assignment process, whose purpose is to determine whether or not there exists a feasible set of channels Ω(A, B) ⊆ LAC(A, B) that can support the total traffic demand RA . Depending on the outcome of the channel assignment process, B decides whether or not A can transmit. If not, then B does not respond to A. Otherwise, B sends a Clear-to-Send (CTS) message to A, which contains Ω(A, B) and the duration (Tpkt (A)) needed to reserve the assigned channels for the ensuing data transmission and associated ACK packet. The CTS implicitly instructs the CR neighbors of B to refrain from transmitting over the set of assigned channels for the duration Tpkt (A). Once A receives the CTS, it replies back with a “Decided-Channels-to-Send” (DCTS) message,
42 informing its neighbors of Ω(A, B) and Tpkt (A). It also announces the success of the RTS/CTS exchange between A and B to A’s neighbors, which may not have heard B’s CTS. After completing the RTS/CTS/DCTS exchange, the transmission A → B proceeds. Once completed, B sends back an ACK packet to A over the channel in Ω(A, B) that has the highest rate. Because there is no interference between data and control packets, a CR that hears the RTS (CTS) packet defers its transmission only until the end of the control packet handshaking. This allows for more parallel transmissions to take place in the same vicinity. Before concluding this section, we give the formats of the various control packets. For a CR transmitter A and a CR receiver B, the formats of the RTS, CTS, and DCTS are: (i)
RTS(A → B) = {A, B, LAC(A), Pctrl (A), RA , PC,β (A)}. CTS(B → A) = {B, A, Ω(A, B), Tpkt (A)}. DCTS(A → B) = {A, B, Ω(A, B), Tpkt (A)}.
2.6.3 Channel Assignment It is known that using the maximum possible channels for a transmission reduces the CRto-PR interference [26, 59]. However, this may lead to channel over-assignment, which reduces the opportunity for assigning available channels to other CR transmitters [9, 59]. In our work, we statistically bound the CR-to-PR interference while using the minimum possible number of channels. Three parameters impact channel assignment: (1) the SSI, → − (2) P C,β (A), and (3) RA . Based on the above parameters, receiver B acts as follows: • When B receives A’s RTS, it first checks LAC(A, B) and removes any channel i whose received SINR µi is less than µ∗i (note that the transmission power and instantaneous interference are known at B). • B sorts the rest of the available channels in a descending order of their data rates, calculated according to the receiver SINR and any predefined rate-vs-SINR relationships (e.g., Shannon’s equation, staircase function, etc.). It then iteratively picks channels from the top of the sorted list until either the aggregate rate is satisfied, the sorted list is exhausted (i.e., no feasible channel assignment can be found), or
43 the number of selected channels exceeds nr . In the latter two cases, B will not respond to A’s RTS, prompting A to back off and retransmit later. It is easy to show that this channel assignment is optimal in terms of minimizing the number of selected channels. Algorithm 1 summarizes the channel assignment process.
* *
CR user
*
*
C
*
*
*
*
* A
*
*
*
C
*
B
*
Users in PRN 1 Users in PRN 2
D
*
*
*
*
*
+
*
D A
B
*
*
(a) Possible reuse
channel
(b) Impossible channel reuse
Figure 2.5: Scenarios in which a CR transmitter C can/cannot reuse the channels assigned to A. Solid circles indicate data-transmission ranges, while dashed circles indicate control-transmission ranges. Figure 3.11 depicts two scenarios for the operation of COMAC. In the first scenario (Figure 3.11(a)), the two transmitters A and C cannot hear each other’s control packets. So, according to Proposition 1, the transmissions A → B and C → D can overlap in their data channels. In Figure 3.11(b), node C falls in the control region of node A (and vice versa). The exclusive channel occupancy policy prevents A and C from using common channels. However, the two transmissions can proceed simultaneously if A and C can find non-intersecting channels to support their rates.
2.7 Performance Evaluation We now evaluate the performance of COMAC and study its effect on the performance of PR users. Our simulation programs are written in CSIM (a C-based process-oriented discrete-event simulation package) [3].
44
Algorithm 1: Channel Assignment (i) Input: LAC(A, B), PC,β (A), I(B), GAB , RA A feasible channel assignment
Output: Ω(A, B) or
indicate channel assignment is infeasible for all i ∈ ACL(A, B) Compute µi using (2.28) if µi < µ∗i ACL(A, B) = ACL(A, B) − {i} else Compute data rate of ith channel (ri ) for all i ∈ ACL(A, B) Sort the channels in a descending order of ri Ω(A, B) = φ RAT E = 0 Let U be the set of the sorted channels while U 6= φ Pick a channel, say j, from the top of U U ← U − {j} Ω(A, B) ← Ω(A, B)
S
{j}
RAT E ← RAT E + rj if RAT E ≥ RA return Ω(A, B) break else if U == φ return “no feasible assignment found”
45 In our analysis, we assumed that the interference at a PR receiver is equal to the sum of the interference powers of all other interferers within an interference radius r c . We also assumed that COMAC enforces an exclusive channel occupancy policy on CR transmissions (i.e., no CR-CR interference). Note, however, that hidden-terminal problem can occur in this scenario due to imperfect control. Our simulations relax these assumptions and account for all sources of interference, including those that are far away from a receiver (primary or cognitive) and use common channels. We focus on one-hop CR communications and investigate the effect of coexistence between the CRN and the PRNs on network performance. Our performance metrics include the outage probability for PR users, p out , and the CRN goodput, defined as the average number of successfully received packets per time slot. For simplicity, we consider a fixed-packet size (2 Kbytes) and a fixed rate demand (RA = 10 Mbps) for all CR users. We set the control-packet size to 120 bits. A time slot corresponds to the transmission of one packet at RA . We also measure the end-to-end goodput in multi-hop routing scenarios.
2.7.1 Simulation Setup We simulate a system consisting of 8 PRNs and 1 CRN. Users in these networks are distributed over a 500×500 meters2 area. We study both uniform and non-uniform node deployments. The first 4 PRNs operate in the 900 MHz band, occupying 4 non-overlapping (i)
2.5 MHz channels with PL = 2 × 10−9 W. The other 4 PRNs operate in the 2.4 GHz (i)
frequency band, occupying 4 non-overlapping 2.5-MHz channels with PL = 1 × 10−10 W. The activity factors for the 8 PRNs are 0.1, 0.2, 0.3, 0.4, 0.1, 0.2, 0.3, and 0.4, respectively. The number of PR users in each PRN is 200. The transmission power for each PR transmitter is 1 W and the antenna length (D) is 5 cm. We set the minimum distance between a PR receiver and the nearest PR interferer (bi ) to 25 meters for all i. The CRN consists of 200 users. Each CR user generates packets according to a Poisson process with rate λ (in packet/time slot), which is the same for all users. We set n r = 4; i.e., a CR user can use up to four data channels simultaneously. The signal propagation model in our simulations follows (3.4) with n = 4. We set µ∗i to 5 dB for all i. For all experiments, we select the value of r ∗ (defined in Section 2.5) such that FRmin (r∗ ) = 1 − p∗ = 10−3 . The reported results are averaged over 100 runs.
46 2.7.2 Single-hop Scenarios We first investigate the effect of CR transmissions on the performance of PR users assuming uniform node deployment. Figure 2.6(a) illustrates pout versus time7 for two PRNs (1 and 6) with β = 0.05. The reported results are cumulative over time, i.e., 0-100, 0-200, 0-300, etc. It can be observed that pout is always less than β = 0.05 for both PRNs. As time progresses, pout converges to a value less than 0.05. These results are in line with the analysis in Section 2.5. For the next experiments, we focus on the performance for PRN 1 (other PRNs depicted similar behaviors). Figure 2.6(b) demonstrates pout as a function of λ at β = 0.01, 0.05, and 0.1. The results show that the bound on pout is always satisfied. The impact of different values of rctrl (.) on pout is shown in Figure 2.6(c) with β = 0.05. The figure illustrates that for rctrl (.) ≥ 2 rdata (.), the statistical guarantee is satisfied. However, for rctrl (.) < 2 rdata (.), the statistical guarantee is not always satisfied. Figure 2.6(d) shows the CDF of the observed pout (Fpout ) with β = 0.05. The figure reveals that < 5% of the time the total interference power at a PR receiver exceeds the maximum tolerable interference (i.e., Fpout (β = 0.05) = Pr[pout < β] < 0.95.). Thus, the statistical guarantee is satisfied. For a given β, Figures 2.7(a) and (b) depict the channel usage, defined as the fraction of time in which a specific channel is used for CR transmissions. These figures reveal that the carrier frequency and PRN activity factor affect channel usage (recall that the 8 PRNs differ in their αi values). The smaller the value of αi , the higher is the utilization of channel by the CRN. The CRN utilization of the lower four channels is higher than that for the higher four channels (because of the lower attenuation). Even though channels with lower carrier frequencies and smaller activity factors are favored for CR transmissions (lower attenuation), under moderate and high traffic load, there are no significant differences in channel usage among all channels. Furthermore, channel usage remains fairly fixed in that traffic regime. Figure 2.8(a) shows the effect of β on the CRN connection blocking rate, defined as the fraction of CR packet attempts that need to back off due to channel unavailability. 7
All figures reporting pout show only the PR user that experienced the highest interference
among all PR users in the given PRN.
47 As demonstrated, the blocking probability is smaller at larger β. This is because a larger β increases the maximum allowable powers for CR users, and consequently decreases the required number of channels to support the aggregate rate demand. Figure 2.8(b) indicates that a larger value of β results in improved CRN goodput. This can be deduced from (2.26) → − and (2.27), as larger β results in increasing P C,β . Consequently, each CR user can use fewer number of channels and more CR transmissions can take place concurrently. We study the impact of different control transmission ranges on the CRN goodput. Similar to the experiment in Figure 2.6(c), four control transmission radii are simulated: r ctrl = rdata , 1.5 rdata , 2 rdata , and 3 rdata . Figure 2.8(c) shows the CRN goodput as a function of packet generation rate under different values of rctrl and a fixed β = 0.05. The figure shows that increasing rctrl (relative to rdata ) has two conflicting consequences: (1) the transmission floor reserved by a CR user increases, and (2) the CR-to-CR interference is reduced. The latter consequence reduces the mutual interference (i.e., improves goodput), while the former reduces the number of simultaneous transmissions (i.e., reduces goodput). Based on our simulations, we can draw the following observations: (1) for large values of r ctrl , the first consequence dominates (i.e., there are fewer simultaneous CR transmissions). Consequently, the CRN goodput decreases with an increase in rctrl , (2) for relatively small values of rctrl (e.g., between rdata and 1.5 rdata ), we observe that the effect of the second consequence dominates, resulting in better goodput performance when r ctrl = 1.5 rdata than when rctrl = rdata . The system performance under Raleigh and Rician channel models is investigated in Figure 2.8(d). Two Rician factors (Kf ) are simulated: Kf = 1 and ∞8 . For a fair comparison, we considered normalized random variables to present the fading processes. The results show that both channel models give almost the same throughput performance. In both cases, the outage probability guarantee is still achieved. We also investigate the effect of selecting p∗ (equivalently r ∗ ) (defined in Section 2.5) on the performance under different CR traffic loads. Figure 2.9(a) plots γ over the feasible range of p∗ for different values of β. For β = 0.1, Figures 2.9(a) and (b) indicate that using the largest possible value of p∗ (consequently the largest possible γ) results in improved CRN goodput. This can be deduced from (2.26) and (2.27), as larger γ results in increasing → − → − P C,β for CR users. The increase in P C,β reduces the number of channels assigned to a CR transmitter, which allows for more concurrent transmissions. Figures 2.9(c) indicates 8
When Kf = ∞, the Rician channel reduces to the AWGN channel.
48 that for all values of p∗ in Figure 2.9(b), the required guarantee is always achieved. Similar observations can be made for other values of β. Next, we study the impact of αi on performance. Three different activity profiles are simulated: high (αi = 0.8), moderate (αi = 0.4), and low (αi = 0.1). Figure 2.10(a) shows that the CRN goodput decreases for higher PRN activity profiles. This is expected since the larger αi , the higher will be the PRNs spectrum utilization, which decreases the maximum allowable powers for CR users. Figure 2.10(b) indicates that the bound on pout is always achieved under different activity profiles. In Figure 2.10(c), we study the effect of inaccurately estimating αi . We operate COMAC assuming an estimated activity factor αei of 0.4, and vary its actual value in the simulations. It is noted that the required
bound on pout is not satisfied when αi exceeds αei by more than 20%. Thus, αi has to be
conservatively estimated.
Finally, we investigate the robustness of COMAC under a skewed user deployment.
We assume that the network field is divided into four quadrants with respective PR user densities 60%, 25%, 10%, and 5%. Figure 2.11(a) illustrates that the required bound on pout is still satisfied. However, Figure 2.11(b) shows that a minor reduction in the CRN goodput may occur compared with a uniform node deployment.
2.7.3 Multi-hop Scenarios We implement a channel-aware routing (CAR) mechanism for CRNs that extends the wellknown minimum hop routing (min-hop) approach. We assume shortest path routing using Dijkstra’s algorithm is used in the CRN network. CAR exploits the channel availability information to produce end-to-end routes that improve the end-to-end CRN goodput. It uses the hop count and the residual capacity of each link, defined as the maximum data rate a link can support over all the available channels. The selected route is one with the minimum number of hops that can support a given demand rate. This can be achieved by first pruning any link that has a residual capacity less than the source demand, then applying min-hop routing on the pruned graph. The potential benefit of using CAR in contrast to min-hop routing is demonstrated in Figure 2.12, where a source S attempts to transmit data to a destination D. Figure 2.12(a) shows node connectivity in the original graph. Min-hop routing always selects route S-A-D for data transmissions, as shown in Figure 2.12(b). CAR will also use route S-A-D if it satisfies the required rate. However,
49 if some of the links cannot support the source demand, as demonstrated in Figure 2.12(c), CAR will select route S-B-E-D. We investigate the end-to-end goodput performance of COMAC as a function of the packet generation rate under min-hop routing and CAR. In our configuration, described above, we randomly select any pair of nodes as source and destination. For min-hop routing, the paths are computed only once at the beginning of the experiment. For CAR, we compute the route for every flow of packets between the source and destination. All the packets in that flow follow the same path. The measured average path length between any source and destination in our network was about seven hops. Figure 2.13 shows that CAR achieves up to 25% increase in the end-to-end goodput. This improvement is magnified under high loads.
2.8 Conclusion In this chapter, we proposed a MAC protocol for opportunistic CRNs. Our protocol, COMAC, improves spectrum utilization while limiting the interference imposed on licensed users. COMAC provides a statistical performance guarantee for PR users in terms of the outage probability pout . We first developed stochastic models for the PR-to-PR and the PR-to-CR interference under a Rayleigh fading channel model, and derived closed-form expressions for the mean and variance of each interference component. Furthermore, closedform expressions were obtained for the characteristic function of the total interference under typical path loss exponents. Empirical results validated the developed interference models. Using numerical fitting, we found that the actual distributions of the PR-to-CR and PR-to-PR interference are well approximated by a log-normal distribution. From the interference analysis, we derived a closed-form expression for the maximum allowable powers for CR transmissions that ensure a statistical bound β on pout for PR users. We integrated our theoretical analysis in the design of the COMAC protocol. Our simulation results show that COMAC statistically guarantees the performance of PR users under different CR traffic loads and for different values of β. Results also show that channel usage is reasonably balanced across various channels, even when the PR activity factors over such channels and the associated carrier frequencies are significantly different. Although uniform node deployment was used in our analysis, our simulations verified that
50 the performance is not significantly impacted by the distributions of users in PRN/CRN. Finally, our simulation results showed that exploiting the available channel information for the routing decisions can improve the end-to-end throughput of the CRN by up to 25%.
51
0.08
0.15
α = 0.1, f = 0.9 GHz α = 0.2, f = 2.4 GHz Bound, β= .05
0.07
β = 0.1 β = 0.05 β = 0.01
0.13
0.06
0.11 0.09
0.04
pout
pout
0.05
0.03
0.07
0.02
0.05
0.01
0.03
0 −0.01
500
1000
1500
Time slots
2000
2500
0.01 0 0.02
0.06
(a) pout vs. time (λ = 0.06)
rctrl rctrl rctrl rctrl
0.14 0.12
λ
0.14
0.18
0.22
(b) pout vs. λ
= rdata = 1.5 rdata = 2 rdata = 3 rdata
1 0.9 0.8 0.7
0.08
CDF
pout
0.1
0.1
0.06
0.6 0.5 0.4
0.04
0.3
0.02
0.2
β = .05
0 0.02
0.06
0.1
λ
0.1
0.14
0.18
(c) pout (for different rctrl (.)) vs. λ
0 0
0.025
0.05
0.075
pout
0.1
0.125
0.15
(d) CDF of observed pout (β = 0.05)
Figure 2.6: Performance of a PRN.
52
CH 1 CH 2 CH 3 CH 4 CH 5 CH 6 CH 7 CH 8
Channel Usage (%)
20 18 16 14 12 10 8 6 4
18
PRN 1
15
PRN 2
PRN 3 PRN 4
12
PRN 5 PRN 6
PRN 7
PRN 8
9 6 3
2 0 0
β = 0.05
21
Channel usage (%)
22
0.05
0.1
λ
0.15
0.2
(a) CR channel usage vs. λ
0
1
2
3
4
5
Channel No.
6
7
8
(b) CR channel usage (high load)
Figure 2.7: Channel usage for the CRN.
53
12
17
Blocking probability (%)
16
Goodput (packet/time slot)
β = 0.01 β = 0.05 β = 0.1
15 14 13 12 11 10 0.02
0.05
0.08
λ
0.11
0.14
0.17
0.2
(a) CR blocking rate
16 14 12
rctrl = rctrl = rctrl = rctrl =
6 4 2 0.1
λ
β = 0.1 β = 0.05 β = 0.01
4 2 0 0
0.05
0.1
λ
0.15
0.2
0.25
14
rdata 1.5 rdata 2 rdata 3 rdata
8
0.05
6
(b) CRN goodput for different values of β
10
0 0
8
Goodput (Packet/time slot)
Goodput (Packet/ Time slot)
18
10
0.15
0.2
(c) CRN goodput for different values of rctrl
β =0.1
12 10 8 6
Rayleigh Rician, K = 1
4
Rician, K = ∞
f f
2 0 0
0.04
0.08
0.12 λ
0.16
0.2
0.24
(d) Goodput vs. λ (Rician and Raleigh channel models)
Figure 2.8: Performance of the CRN.
54
0.12
0.1
0.08
β=0.1
β=0.05
γ
0.06
0.04
0.02
β = 0.01
0 0.9
0.92
0.94
p*
0.96
0.98
1
(a) γ vs. p∗ 11
Goodput(Packet/time slot)
β =0.1
10 9 8 λ =0.05 λ =0.11 λ =0.25
7 6 5 4
3 0.9
0.92
0.94
p*
0.96
0.98
.9999
(b) Goodput vs. p∗
0.15
β = 0.1
0.13
pout
0.1 0.07 λ=0.05 λ=011 λ =0.25
0.04 0.01
0.9
0.92
0.94
p*
0.96
0.98
0.9999
(c) pout vs. p∗
Figure 2.9: Impact of selecting p∗ .
55
β = 0.05
Goodput (Packet/time slot)
18
Low PR activity factors Moderate PR activity factors High PR activity factors
16 14 12 10 8 6 4 2 0 0
0.05
0.1
λ
0.15
0.2
0.25
(a) CRN goodput for different αi values 0.08
Low profile Modrate profile High profile
0.07
pout
0.06 0.05 0.04 0.03 0.02 0.01
0.05
0.1
λ
0.15
0.2
(b) pout vs. λ (different αi ) 0.14
Estimated α i
0.12
Estimation error 20%
pout
0.1 0.08 0.06 0.04 0.02 0.2
0.3
0.4
Actual
0.5
αi
0.6
0.7
(c) pout under inaccurate αi
Figure 2.10: Impact of activity profile on performance.
56
Non−uniform node distribution
15
β = 0.1 β = 0.05 β = 0.01
0.14 0.12
pout
0.1 0.08 0.06 0.04 0.02 0
0.05
0.1
λ
0.15
0.2
0.25
(a) pout vs. λ
Goodput(Packet/time slot)
0.16
12
β = 0.1, Skewed β = 0.01, Skewed β = 0.1, Uniform β = 0.01, Uniform
β = 0.1
9
6 β = 0.01 3
0 0
0.05
0.1
0.15
λ
0.2
0.25
(b) CRN goodput for different β
Figure 2.11: Performance under skewed and uniform user deployments.
D
A S
B
S
E
C
D
A
G
B C
F
S
C
(b) A
D
X
B
X E
X
S
F
S
B C
C
G F
A
D
X E
X F
(e)
E
(d)
X X
D
B
G
(c) A
G F
(a) A
E
S G
D
B C
E
G F
(f)
Figure 2.12: Illustration of channel-aware routing.
57
End−to−end goodput (packet/ time slot)
2.5
CAR Min−hop routing
2
1.5
1
0.5
0 0
0.05 0.1 0.15 0.2 Packet generation rate (packet/ time slot)
0.25
Figure 2.13: End-to-end throughput of min-hop routing and CAR as a function of the packet generation rate.
58
CHAPTER 3 Distance- and Traffic-Aware Channel Assignment in Cognitive Radio Networks
3.1 Introduction A CRN has unique characteristics that distinguish it from traditional multi-channel wireless networks. Unlike traditional wireless networks, which typically occupy contiguous bands [43, 54, 30], a CRN is expected to operate over a set of widely-separated noncontiguous frequency bands. Communication on such bands exhibits different RF attenuation and interference behaviors. It is well known that signal attenuation increases with the distance between the two communicating users and also with the carrier frequency used for communication [48]. Therefore, when assigning transmission channels in a CRN, it is necessary to consider the signal attenuation model and the interference conditions to improve spectrum utilization. Another characteristic of a CRN is that users must operate using a relatively low transmission power (i.e., abide by a power mask) to avoid degrading the performance of the PR users [9, 59]. These peculiar characteristics call for new MAC protocols that efficiently utilize the available spectrum while improving the overall network throughput.
3.1.1 Motivation Channel assignment mechanisms in traditional multi-channel wireless networks typically select the “best” channel, or set of channels, for a given transmission (e.g., [43, 30, 59, 44]). In these mechanisms, the best channel is often defined as the one that supports the highest rate. We refer to this approach as the best multi-channel (BMC) approach. When the BMC approach is employed in a CRN, the blocking probability for CR transmissions, defined as the percentage of CR packet requests that are blocked due to the unavailability of a feasible channel assignment, can increase, leading to a reduction in the network throughput. To illustrate, consider an environment in which two PRNs and one CRN
59 coexist. PRN 1 operates over a low-frequency band (CH1), while PRN 2 operates over a high-frequency band (CH2). Suppose that PRN 2 introduces a higher average PRto-CR interference. Consequently, a CR receiver experiences a higher average signal-tointerference-plus-noise ratio (SINR) over CH1 than over CH2. Assume that two CR users A and C need to send data to CR users B and D, respectively (see Figure 3.1). Also assume that the distance between A and B (dAB ) is less than that between C and D (dCD ). Figure 3.1(a) shows that when the CR users employ the BMC approach, the transmission A → B uses CH1, whereas the transmission C → D uses CH2. A → B is allowed to proceed because it operates over a low carrier-frequency channel with low PRto-CR interference for a short transmission distance. On the other hand, C → D requires relatively higher transmission power to overcome the high attenuation associated with the high-frequency/high-interference channel and the long transmission distance. If the required transmission power exceeds the specified power mask, C → D cannot proceed. However, both A → B and C → D have much better chances of proceeding simultaneously if each CR transmitter selects channels while keeping in mind the constraining power mask of the other transmitter (Figure 3.1(b)). As a numerical example, assume that PRN 1 and PRN 2 operate in the 900 MHz and 2.4 GHz bands, respectively. Assume that dAB = 10 meters and dCD = 50 meters. Also assume that a CR transmission is successful if the received SINR over the selected channel is greater than the SINR threshold. For both channels, we set the SINR threshold and the interference mask to 5 dB and 60 mW, respectively. Assume that CR receivers B and D experience the same level of total interference over both channels (0.05 µW). Given the above parameters and using the propagation model in [47] with path loss exponent of 2, the required transmit powers over CH1 and CH2 for A → B are 2.2 mW and 16 mW, respectively. For C → D, these powers are 56.18 mW and 399.5 mW. According to the BMC scheme (Figure 3.1(a)), A → B can proceed over CH1 (the power mask is not violated), whereas C → D cannot proceed over CH2 (the required transmit power exceeds the power mask). On the other hand, if A → B uses CH2 and C → D uses CH1, both transmissions can proceed simultaneously (Figure 3.1(b)). It is worth mentioning that in a given (one-hop) neighborhood, the optimal channel assignment that maximizes the number of simultaneous CR transmissions can be formulated as an integer linear programming (ILP) problem [11, 12]. Since computing the optimal
60
C
CH 2
D
X
A
B
CH 1
(a) BMC channel assignment
C
CH 1 A
D B
CH 2
(b) Distance-dependent channel assignment
Figure 3.1: Scenarios in which two CR transmissions can/cannot proceed simultaneously. solution for the ILP problem grows exponentially with the size of the network [11], heuristic algorithms with suboptimal performance are needed. Such algorithms should attempt to compute channel assignment with reasonable computational/communication overhead.
3.1.2 Goals and Contributions In this chapter, we develop a novel CSMA-based MAC protocol that aims at enhancing the throughput of the CRN subject to a power mask constraint. The proposed protocol (DDMAC) employs an intelligent stochastic channel assignment scheme that exploits the dependence between the RF signal attenuation model and the transmission distance while taking into consideration the local traffic conditions. The channel assignment scheme accounts for the interference conditions and the power constraints at different bands. In particular, the scheme assigns channels with lower average SINR to shorter transmission distances, and vice versa. In addition, our scheme associates more preferable channels to the most frequent transmission distances and less preferable channels to the less frequent distances. In other words, the assignment process identifies a “preferable” channel list for each CR user. Such a list indicates which channels are preferable to use depending on the estimated distance between the transmitter and the receiver. We propose two variants for the channel assignment scheme. The first variant is suitable for offline planning of spectrum sharing in networks with known deployment and traffic patterns, while the
61 other is suitable for online dynamic network operation with unknown traffic patterns. The second one employs a stochastic learning technique that adapts to network dynamics (i.e., mobility, interference conditions, and traffic conditions). The primary advantage of our assignment scheme is that it is based on passive learning. This is because in DDMAC, CR users always listen to the control channel in order to overhear control-packet exchanges, including those not destined to them. CR users use the control information to identify the preferable channels. DDMAC has the following attractive features: • It does not make any assumptions about the activity patterns of the underlying networks or about user distribution. • It is easy to implement in practical settings and its processing overhead is small. • It is transparent to PR users, i.e., does not require coordination with them. • It inherently improves the fairness among CR users, compared to typical multichannel CSMA-based protocols. • Under low load and several available channels, DDMAC gracefully degrades to the BMC approach. To evaluate the performance of DDMAC, we conduct simulations over a dynamic CRN with mobile users. Our simulation results show that by being distance- and traffic-aware, DDMAC significantly improves network throughput while preserving fairness. The results also indicate that compared with typical multi-channel CSMA-based protocols, DDMAC decreases the connection blocking rate in a CRN by up to 30%. By injecting artificial errors into the estimated distances, our evaluation reveals that DDMAC is robust against estimation errors. It should be noted that selecting a preferable channel list was also proposed in the MMAC protocol [60]. However, MMAC does not support multiple-channel assignment (i.e., MMAC is limited to one channel per user). Specifically, the channel selection criterion in MMAC is to use a channel with the lowest count of source-destination pairs that have selected the channel. In DDMAC, the preferable channel list per node is constructed by accounting for the challenges associated with CRs (i.e., low transmit power, presence of PR
62 users, widely-separated non-contiguous available bands). Unlike DDMAC, the objective in MMAC was not to address spectrum sharing while improving the overall throughput, but rather to handle multi-channel hidden terminals using a single transceiver and to balance the channel usage over all available channels. In addition, MMAC requires global network synchronization, which is not a requirement in DDMAC.
3.1.3 Organization The rest of this chapter is organized as follows. Section 3.2 gives an overview of related work. In Section 3.3.1, we introduce our system model and state the main assumptions. The SINR analysis is presented in Section 3.3.2. Section 3.3.3 illustrates the effect of the carrier frequency and transmission distance on the path loss. In Section 4.2.4, we formulate the optimal channel assignment problem. Section 3.5 introduces our proposed distance- and traffic-aware channel assignment algorithm. Section 3.6 describes the proposed DDMAC protocol and outlines its benefits and associated overhead. We evaluate DDMAC in Section 3.7. Finally, Section 3.8 gives concluding remarks.
3.2 Related Work Recently, several attempts were made to develop MAC protocols for CRNs (e.g., [59, 46, 64, 65, 40, 55, 45]). Existing literature on spectrum sharing/access protocols can be classified according to their architecture (centralized or decentralized), spectrum allocation behavior (cooperative or non-cooperative), and spectrum access technique (overlay or underlay) [9]. The IEEE 802.22 working group [1] is in the process of standardizing a centralized MAC protocol that enables spectrum reuse by CR users operating on the TV broadcast bands. References [13, 70, 16] are examples of centralized protocols that were proposed for coordinating spectrum access. DSAP [13] is a centralized scheme for spectrum sharing in CRNs. It enables a central entity to lease spectrum to users in a limited geographical region. In [70], a new user-central system architecture for dynamic channel allocation was proposed. According to this architecture, CR users can dynamically and adaptively change their service provider based on their desired criteria (e.g., quality/cost metrics). For an ad hoc CRN without centralized control, a distributed MAC protocol in needed
63 to enable every CR user to individually access the spectrum without corrupting the communication of PR users. In [59], the authors developed a CRN MAC protocol with a common control channel. This protocol jointly optimizes the channel/power/rate assignment, assuming a given power mask on CR transmissions. DC-MAC [46] is a cross-layer distributed scheme for spectrum allocation/sensing. It provides an optimization framework based on partially observable Markov decision processes, assuming that PR and CR users share the same slotted transmission structure. In [64], the authors investigated continuous-time Markov models for dynamic spectrum access in open spectrum wireless networks. Using such models, a distributed random access protocol is proposed to achieve airtime fairness between dissimilar unlicensed users. The FCC defined the interference temperature model [26], which provides a metric for measuring the interference experienced by licensed receivers. In [65], the authors studied the issue of spectrum sharing among a group of spread-spectrum users subject to constrains on the SINR and on the interference temperature. In [19], the interference temperature model was used for optimal selection of spectrum and transmission powers for CR users. Three spectrum sharing techniques were proposed and compared in [40]: spreading-based underlay, interference avoidance overlay, and spreading-based underlay with interference avoidance. The metric of interest in the comparison was the outage rate of PR transmissions. Interference statistics were used assuming an infinite number of users and an unbounded region for outage probability analysis. AS-MAC [55] is a spectrum-sharing protocol for a CRN that coexists with a GSM network. In AS-MAC, the GSM network is assumed to provide input to the CRN over a broadcast channel. Each CR user selects channels based on the CRN’s control packets and information from the GSM network. Thus, explicit coordination with the PRNs is required. In [45], the authors proposed a decentralized channel-sharing mechanism for CRNs based on a game-theoretic approach for both cooperative and non-cooperative scenarios. In [69], the concept of a time-spectrum block is introduced to model spectrum reservation in a CRN. Based on this concept, the authors presented centralized and distributed CRN protocols with a common control channel for spectrum allocation. The above protocols were designed without exploiting the dependence of the number of allowable CR transmissions on the carrier frequency and the transmission distance. They are limited to the analytical aspects of MAC design, with no complete operational
64 details. To the best of our knowledge, DDMAC is the first CRN MAC protocol that aims at improving the CRN throughput by exploiting the dependence on the RF signal’s attenuation model and the transmission distance while considering the prevailing traffic and interference conditions.
3.3 Preliminaries Users in PRN1 CR
CR
Users in PRN2
CR
Figure 3.2: Example of an opportunistic CRN that coexists with two PRNs.
Figure 3.3: Operating spectrum in the hybrid network.
3.3.1 Network Model We consider a CRN with decentralized control (i.e., an ad hoc network). This CRN coexists geographically with M different PRNs. PR users are legacy radios that cannot be controlled by the CRN. Figure 3.2 shows a conceptual view of the scenario under consideration with M = 2. The PRNs are licensed to operate over non-overlapping frequency
65 bands. We assume that all the PRN bands have the same bandwidth (BW ). In reality, a PRN may occupy multiple, non-contiguous, frequency bands. Such a PRN can be easily represented in our setup by using multiple equal-bandwidth virtual PRNs, each operating over its own carrier frequency. For the ith PRN, we denote its carrier frequency by f i . As shown in Figure 3.3, the available bandwidth (BW ) of a PRN is divided into L adjacent but non-overlapping frequency channels each of Fourier bandwidth W (in Hz). Such L channels are collectively referred to as a band. Let N denote the total number of channels in all bands; N = LM . Without loss of generality, we assume that BW is sufficient to support at least one CR transmission. This is an acceptable assumption in many wireless systems that are built to operate in the unlicensed bands, including IEEE 802.11/a/b/g-compliant devices. Each CR user is equipped with nt radio transceivers, 1 ≤ nt ≤ L, that can be used simultaneously. In theory, a CR user can transmit over an arbitrary segment of the available bandwidth by using tunable filters. In practice, however, a CR typically implements a bank of fixed filters, each tuned to a given carrier frequency with fixed bandwidth, allowing the CR user to choose from a fixed number of channels. In our setup, we assume the latter (more practical) capability, which can be used to approximate the tunable filter scenario. To avoid corrupting the transmissions of licensed users, a mask is enforced on the (i)
transmission power of a CR user over each band, i.e, Pt
(i)
≤ Pmask , i = 1, 2, . . . , M . The
determination of an appropriate power mask is an important topic, which has been investigated under certain simplifying assumptions (e.g., [19, 53]). The spectrum sharing protocols in [19] and [53] were designed such that the maximum transmission powers of CR users over various bands are dynamically computed based on the PR’s interference margins (set by the FCC) and local traffic conditions. In [53], the authors provided a neighborhood-dependent adaptive power mask on CR transmissions that ensures a statistical (soft) guarantee of the outage probability of PRNs (the probability that the total interference power at a PR receiver exceeds the maximum tolerable interference). The authors provided closed-form expressions for the resulting power mask. For our purposes, we assume that a similar mechanism for determining the power mask is in place. A CR user transmits data to other CR users using the maximum allowable power vector P mask . When not transmitting, a CR user is capable of measuring the total noise-plus-interference
66 M L N, N BW , W nt Pt (i) (i) Pmask I (i) (i) SINRj PL (fi ) Pr (fi ) d J (i) cj Cj µ∗i mj Mj R Rc , r c Ri ri , ri−1 Di di , di−1 m Twin pi (t) pei (t) α Ωi (A), K CCL(A, B) Φ(A, B)
Number of PRNs Number of channels in a PRN N is the total number of channels, N = {1, . . . , N } Bandwidth of a band and a channel, respectively Number of transceivers per CR user CR transmit power Interference power mask on channel i Noise-plus-interference over channel i Measured SINR over band i at receiver j Path loss associated with band i Received power at a CR receiver over band i Transmitter-receiver distance Set of all CR transmission requests in a locality ith selected channel’s data rate for transmission j Rate demand of the jth CR transmission SINR threshold over channel i Number of selected channels for the jth transmission Mj is the set of mj selected channels for the jth transmission Random variable represents the distance to the intended receiver RC is the transmission region, Rc = πrc2 ith ring around a CR in static channel assignment Radii that define Ri , i = 1, . . . M ith ring around a CR in dynamic channel assignment Radii that define Di Number of non-overlapping Di rings Observation window time Probability of Di at time t Weighted average of pi (t) Forgetting factor CR A’s preferable channel list for region i, i = 1, ..., K Common channel list available for A → B transmission Preferable available channels for A → B transmission
Table 3.1: Summary Of Notations Used In This Chapter. I (i) over all bands i = 1, 2, . . . , M 1 . This requires a wideband sensing capability with a narrowband resolution. The technology to support such capability is readily available through a wideband antenna, a power amplifier, adaptive filters, and a DSP technique called cyclostationary feature detection [15, 9, 14]. Thus, a CR user can simultaneously sense several GHz-wide bands and estimate the instantaneous interference over each band [14]. Alternatively, a sequential partial sensing approach can be employed at the cost of negligible switching/sensing overhead [15, 51]. It is worth mentioning that off-the-shelf wireless cards can readily serve as a fully functional wideband multi-channel CR interface. Such an interface enables a CR user to perform analysis of the RF spectrum (i.e., sensing) in real time. 1
The quantity I (i) includes the PR-to-CR interference as well as the thermal noise.
67 3.3.2 Analysis of the Average SINR Based on the aforementioned characteristics of the CRN, the average measured SINR (SINR) at a CR receiver over band i is mainly determined by: (1) the path loss associated with that band (PL (fi )); (2) the average value of the measured I (i) over that band (I
(i)
),
which can be estimated based on the sensing history and the spectrum occupancy statistics (i)
(i)
[33, 46]; and (3) the enforced power mask Pmask . Formally, SINR (dB) is given by: (i)
(i)
(i)
SINR (dB) = Pmask (dB) − PL (fi )(dB) − I (dB).
(3.1)
Table 3.1 summarizes the main notation used in this chapter.
3.3.3 Carrier Frequency and Distance Effects on Path Loss In this section, we discuss the effect of the carrier frequency and transmission distance on the path loss. For a given carrier frequency f , let do (f ) be the close-in distance, i.e., the distance from the transmitter after which the RF channel can be approximated by the free-space model; do (f ) can be determined from measurements or can be estimated by [47]: do (f ) = max
2Da2 f c , Da , c f
(3.2)
where Da is the antenna length of the transmitter and c is the speed of light. Let Po (f ) denote the received power at the close-in distance. Then, Po (f ) can be estimated as follows [47]: Po (f ) =
c2 Gt (f )Gr (f ) Pt (f ) (4πdo (f ))2 f 2
(3.3)
where Gt (f ) and Gr (f ) are the transmit and receive antenna gains, respectively. Let Pr (f ) denote the received power at distance d from the transmitter, d ≥ do (f ). Then, do (f ) n Pr (f ) = Po (f ) d
(3.4)
where n is the path loss exponent (typically, 2 ≤ n ≤ 6). Note that, in practice, d o (f ) is of the same order of magnitude as the node’s dimensions. For example, for a mobile phone operating in the 900 MHz band with D = 5 cm, do (f ) = 33 cm. For an 802.11 WLAN card operating in the 2.4 GHz band and the same antenna size, d o (f ) = 12 cm. Accordingly, it is reasonable to assume that the probability that d is less than d o (f ) is very small (i.e., Pr(d < do (f )) ≈ 0).
68 Using (3.2), (3.3), and (3.4), the path loss PL (f ) can be expressed as: PL (f ) = 10 log
where
Pt (f ) = −10 × Pr (f )
o n c2 γDan−2 c 2Da2 f log , , ∀ f s.t. D ≥ max a 2 n c f d n f o 2Da2 f cn γ c ∀ f s.t. f ≥ max Da , c log f n dn , n o n−2 log c4−n γ(2Da2 ) , ∀ f s.t. 2Da2 f ≥ max D , c a c f f 4−n dn def
γ=
Gt (f )Gr (f ) . (4π)2
Note that the dependence of PL (f ) on d (i.e.,
1 dn )
(3.5)
(3.6)
is the same for any given carrier fre-
quency. Figure 3.4 depicts the path loss for a wide range of carrier frequencies and two values of n at a distance d = 1 meter. This figure and equation (3.5) reveal that the signal attenuation increases as the distance between the two communicating users increases, and as the frequency used for communication increases. These observations provide the motivation for our distance-dependant channel assignment, discussed in Section 3.5. 70
Path Loss (dB)
60 50 40 30 20 0
n=4 n=2 1
2
3 f (Hz)
4
5
6
9
x 10
Figure 3.4: Path loss vs. carrier frequency for two path loss exponents (Da = 5 cm, Gt (f ) = Gr (f ) = 1).
69 3.4 Optimal Channel Assignment Problem Our objective is to maximize the number of simultaneous CR transmissions, and consequently the overall network throughput. Toward this end, we define the term local spectrum utilization as the total number of simultaneous CR transmissions that can be supported in a given (one-hop) locality while meeting a predefined power mask. Before formulating the problem, we discuss the requirements for a successful CR transmission.
3.4.1 CRN Transmission Requirements Within a given neighborhood, multiple CR users may contend for access to one or more of the available channels. Let N and J denote the set of all N channels and the set of all CR transmission requests in the local neighborhood at a given time, respectively. We assume that the jth CR transmission (j ∈ J ) is successful if both of the following two conditions are met: • It is possible to find mj available channels from the set N such that (i)
P mj
(i) i=1 cj
≥ Cj ,
where cj is the data rate of the ith selected channel and Cj is the total rate demand for the jth CR transmission. • Let Mj be the set of mj selected channels. Then, the received SINR of every i ∈ Mj (i)
(SINRj ) must be greater than the SINR threshold (µ∗i ) that is required at the CR receiver to achieve a target bit error rate over channel i.
3.4.2 Maximizing the Utilization of Local Spectrum (i)
Let δj be a binary variable denoting whether or not channel i is assigned for transmission j. Formally, (i)
δj
1, if channel i is assigned for transmission j = 0, otherwise.
(3.7)
Similar to [28, 12], the problem of maximizing the total number of simultaneous CR
70 transmissions in a given neighborhood can be formally stated as follows: maxδ(i) ∈{0,1} j P P
(i)
P
j∈J
1
(i)
hP
(i) (i)
i∈N
δ j cj ≥ C j
j∈J
δj ≤ 1, ∀i ∈ N
i∈N
δj ≤ nt , ∀j ∈ J
(i)
i
(3.8) (3.9) (3.10)
(i)
SINRj ≥ µ∗i , ∀j ∈ J , s.t. δj = 1
(3.11)
where 1[.] is the indicator function. The constraint in (3.9) ensures that a channel cannot be assigned to more than one CR transmission in the same vicinity. The constraint in (3.10) ensures that at most nt channels can be assigned to a CR transmission. For an ad hoc CRN, the above optimization problem must run in a distributed manner at each CR user in the network. This implies that each CR user must exchange instantaneous SINR and rate demand information with neighboring CR users before selecting channels, which incurs high control overhead and delay (i.e., information may not be up-to-date). Even if perfect knowledge of the SINR of each link and the rate demands are available, the above ILP problem belongs to the class of NP-hard problems [11]. In this chapter, we develop a heuristic channel assignment scheme that provides a suboptimal solution with low complexity and good spectrum utilization. Our heuristic exploits distance and traffic awareness. The key idea behind it is to assign channels with low SINR to short-distance transmissions. Also, local traffic information is used to assign more channels to more likely transmission distances.
3.5 Distance-Dependent Channel Assignment Algorithm In this section, we describe our proposed channel assignment mechanism. The assignment process identifies a “preferable” channel list for each CR user. Such a list indicates which channels are preferable to use depending on the estimated distance between the transmitter and the receiver. It is worth mentioning that many techniques for estimating the transmitter-receiver distance in wireless networks have been proposed based on the Received Signal Strength Indicator (RSSI), the Time of Arrival (ToA), and the Time Difference of Arrival (TDoA) [66]. For our purposes, we assume that any of these schemes is in place. In Section 3.7, we investigate the robustness of our scheme to inaccurate distance estimation.
71 Two variants of the channel assignment mechanism are proposed. The first variant is suitable for offline planning of spectrum sharing in networks with known traffic patterns, whereas the second variant is for online spectrum allocation in dynamic (mobile) networks with unknown traffic patterns.
3.5.1 Spectrum Assignment for Known Traffic Profiles Given a CR user with a packet to transmit, let r be the estimated distance to the intended receiver; r ≤ rc , where rc is the maximum transmission range. rc represents the largest distance from a CR transmitter over which the transmission at maximum power can be correctly decoded over all selected channels in the absence of interference from other def
terminals (CR or PR users). Let FR (r) = Pr{R ≤ r}. The functional form of FR depends on both node distribution as well as the distance traffic profile, which for now we assume to be given. Given FR , the channel assignment process is conducted as follows: • The available bands are divided according to their measured SINR (given in (3.1))2 into M sets S1 , S2 , . . . , SM , where each band consists of multiple channels. The set S1 contains the frequency channels of the band that has the highest SINR, S2 contains the next highest SINR, and so on. def
• A CR user, say A, divides its maximum transmission region Rc = πrc2 into M nonoverlapping “rings” R1 , . . . , RM . The ith ring contains the CR users whose distances to A fall in (ri−1 , ri ], where i = 1, . . . , M and 0 = r0 ≤ r1 ≤ r2 ≤ . . . ≤ rM = rc . The rings are divided such that the probability of communicating with a CR receiver that falls within any of the M rings is the same, i.e., FR (ri ) − FR (ri−1 ) =
1 , M
i = 1, . . . , M.
(3.12)
User A computes the radii ri , i = 1, . . . M , by substituting for FR (ri ) in (3.12) and solving for ri . • Finally, A constructs a preferable channel list for each ring by assigning channels with lower SINR to shorter transmission distances and channels with higher SINR 2
Note that PL ’s dependence on d is the same for all bands. Thus, for the purpose of SINR
comparison, we set d = 1 meter.
72 to longer transmission distances, i.e., assign SM to R1 , SM −1 to R2 , . . ., and S1 to RM . To illustrate the idea, we consider a uniformly distributed CRN and assume that a CR transmitter randomly chooses a destination for its data from within R c . Therefore, FR (r) is given by: FR (r) =
r2 , rc2
1,
r ≤ rc r ≥ rc
.
(3.13)
Using (3.12) and (3.13), and noting that r0 = 0, we arrive at the following expression for ri : ri =
s
r2 1 + i−1 M rc2
rc =
r
i rc . M
(3.14)
Figure 3.5 illustrates the non-overlapping rings around a CR transmitter when M = 4. Within these rings, other CR and PR users may exist. Assume rc = 100 meters. Then, r1 , . . . , r4 are given by 50, 70.71, 86.6, 100 meters, respectively.
Figure 3.5: Four regions around a CR transmitter for assigning channels.
3.5.2 Spectrum Assignment for Unknown Traffic Profiles For offline spectrum planning, we assumed in the previous section a fixed network and prior knowledge of the distance traffic pattern (i.e., the form of FR ). During network
73 operation, however, the distance traffic pattern may change with time, depending on network dynamics and user mobility. Because users only possess local knowledge of their neighborhoods, it is difficult to maintain the optimal network performance. Nevertheless, we can develop a stochastic learning algorithm that performs well and uses only localized information. Stochastic learning techniques have been widely used in wireless networks for online traffic prediction, tracking, and power control [35, 17]. Our proposed learning approach is a distributed algorithm that runs at each CR user in the network. A CR user, say A, evenly divides its maximum transmission region Rc into m non-overlapping regions, where m M . The ith region, Di , forms a ring, defined by the area {(x, y) : d2i−1 < x2 + y 2 ≤ d2i }, where di = i rmc , and di−1 < di i = 1, . . . , m. CR user A maintains an mentry transmission distance table. The ith entry in that table corresponds to the region D i , and contains the number of overheard CR packet requests during the recent observation window Twin for which the transmitter-receiver distances fall in the range (di−1 , di ] (how to convey transmitter-receiver distance information will be discussed later). Note that the proper setting of Twin depends on the dynamics of the network. The effect of Twin is studied in Section 3.7.
Figure 3.6: Time diagram of pmf’s updating process. To initialize the assignment algorithm, all CR users employ the BMC scheme discussed in Section 3.1. At any time t, CR user A constructs its transmission distance table based on control packets it overheard during the observation window [t − Twin , t]. Using the transmission distance table, A estimates the current probability mass function p i (t) of the distance r at time t (see Figure 3.6). It then computes an exponentially weighted average of pi (t) : pei (t) = αpi (t) + (1 − α)pei (t − Twin ),
(3.15)
where α is a forgetting factor, 0 < α ≤ 1. Once pei (t) is computed, A computes the
74 preferable channel list for each ring. Let Ωi (A) denote the preferable channel list for ring Di at CR user A (how to construct Ωi (A) will be given later). The new preferable channel lists will be used during the next observation window time. The proposed channel assignment process merges the Di ’s into K regions according to pi (t), where K ≤ M . It then assigns preferable channels for each region. The process is now described in detail: P P 1. User A determines the integer k such that | k−1 ei (t) − m ei (t)| is minimized, i=0 p i=k p i.e., it divides the regions into two groups; short-distance and long-distance groups. The probabilities of the short-distance and long-distance groups are given by: Pshort =
k−1 X
pei (t)
(3.16)
pei (t).
(3.17)
i=0
and Plong =
m X i=k
2. User A divides the M bands into two frequency sets: low SINR frequency set and high SINR frequency set. It assigns the low SINR frequency set to the short-distance group and the high SINR frequency set to the long-distance group. The numbers of bands in the high (nH ) and low (nL ) frequency sets depend on Pshort and Plong , as follows: nH nL
Pshort = M Pshort + Plong = M − nH
(3.18)
where d.e is the ceiling function. 3. Step 1 and 2 are repeated for every group until either only one band is assigned to that group or the group contains only one region. Note that when repeating the above process for a group, m in (3.17) and M in (3.18) are replaced by the number of regions in that group and the number of channels assigned to that group, respectively. Using this recursive procedure, the preferable channel list Ωi (A), for all i, is computed for one observation window.
75 3.5.3 Complexity Claim 1: The worst-case complexity for selecting the preferable channel list Ω i (A), for all i, may be obtained using the above recursive procedure in O(mK) time, where K ≈ min[N, m]. Proof: In the worst case, our proposed algorithm requires O(m) comparisons to perform one iteration (steps 1 and 2). In addition, it requires at most K = min[N − 1, m − 1] iterations to obtain Ωi (A), for all i. Hence, Ωi (A), for all i, may be obtained using the proposed algorithm with a complexity of O(mK), where K ≈ (m min[N, m]). For N ≥ m, K ∼ O(m). On the other hand, for N < m, K ∼ O(N ).
3.5.4 Examples We illustrate the previously discussed channel assignment process using the following examples.
Example 1 Consider four PRNs and one CRN. Each PRN occupies two adjacent non-overlapping channels. The PRNs are labeled such that f1 < f2 < f3 < f4 . Consider a CR user A with SINR
(1)
> SINR
(2)
> SINR
(3)
> SINR
(4)
. Suppose that A divides its transmission region
Rc into 8 rings, D1 , D2 , . . . , D8 . At a given time t, assume that the weighted average pmf {pei (t) : i = 1, . . . , 8} is given by {0.25, 0.1, 0.15, 0.05, 0.05, 0.15, 0.05, 0.2}. Figure 3.7 shows
how the proposed channel assignment process is conducted. The outcome of this process is as follows:
• Band 4, which includes two channels, is assigned to all CR transmissions whose distances are in D1 (i.e., Ω1 (A) = {4}). • Band 3, which includes two channels, is assigned to all CR transmissions whose distances are in D2 and D3 (i.e., Ω2 (A) = Ω3 (A) = {3}). • Band 2, which includes two channels, is assigned to all CR transmissions whose distances are in D4 , D5 , and D6 (i.e., Ω4 (A) = Ω5 (A) = Ω6 (A) = {2}). • Band 1, which includes two channels, is assigned to all CR transmissions whose distances are in D7 and D8 (i.e., Ω7 (A) = Ω8 (A) = {1}).
76
Di 1 2 3 4 5 6 7 8
p~ i
0.25 0.1 0.15 0.05 0.05 0.15 0.05 0.2
{f4, f3} 0.5
{f4} {f3} {f2}
{f2, f1} 0.5
{f1}
0.25 0.25 0.25 0.25
Figure 3.7: Example that illustrates the channel assignment process in a dynamic CRN.
Example 2 Consider 8 PRNs and one CRN. The PRNs are labeled such that f1 < f2 < . . . < f8 . Suppose that A divides its transmission region into 2 rings. At a given time t, assume that the weighted average pmf {pei (t) : i = 1, 2} is given by {0.25, 0.75}. Then, the outcome of our preferable channel assignment is as follows:
• Channels 1 and 2 (total of 2 channels are assigned to all CR transmissions whose distances are in D1 ).
• Channels 3, . . . , 8 (total of 6 channels are assigned to all CR transmissions whose distances are in D2 ). The above example reveals that our algorithm assigns more preferable channels (total of 6 channels) to the more frequently used transmission distances (D2 , pe2 (t) = 0.75).
3.6 DDMAC Protocol
Based on the channel assignment process presented in Section 3.5, we now propose a distributed, asynchronous MAC protocol for CRNs. The proposed DDMAC is a CSMA/CAbased scheme that uses contention-based handshaking for exchanging control information. It is worth mentioning that the most common configuration for upcoming CRNs is to use CSMA/CA-like MAC access [69, 34, 9, 32, 59, 55, 71, 51, 53]. Thus, in designing the channel access in DDMAC, we focused on extending the CSMA/CA scheme due to its
77 maturity and wide deployment in many wireless packet networks. It should be noted that the handshaking procedure is essential in multi-channel systems. Besides mitigating the hidden-terminal problems, there are two other main objectives for the use of RTS/CTS: (1) conducting and announcing the channel assignment, and (2) prompting both the transmitter and the receiver to tune to the agreed on channels before transmission commences. Before describing our protocol in detail, we first state our main assumptions.
3.6.1 Assumptions In designing DDMAC, we make the following assumptions: • For each frequency channel, the channel gain is stationary for the duration of three control packets and one data and ACK packet transmission periods. As explained in [42], this assumption holds for typical mobility patterns and transmission rates. • Channel gains between two CR users are symmetric. This is a typical assumption in any RTS/CTS-based protocol, including the IEEE 802.11 scheme. • CR transmissions use the maximum allowable power vector (Pmask ). The key idea behind this choice is as follows. It is well-known that using as many channels as possible for a transmission reduces the CR-to-PR interference [59] due to the reduction in transmission power. However, because DDMAC enforces an exclusive channel occupancy, which prevents two neighboring CR users from using common channels3 , such a channel assignment policy may lead to channel over-assignment, which reduces the opportunity for finding available channels by other neighboring CR transmitters (thus reducing the CRN’s throughput). Therefore, in DDMAC, we tackled the CR-to-PR interference problem by assuming a given power mask to protect PR users while trying to use the least possible number of selected channels per transmission. This can be done by transmitting at the highest possible transmission power over each selected channel, which results in less number of assigned channels per CR transmission. This increases the opportunity for finding available channels by other neighboring CR transmitters. 3
The exclusive channel occupancy excludes CR-to-CR interference although it still allows for
the typical co-channel PR-to-CR interference, thus largely simplifying the CR-to-PR interference management process.
78 • The total rate demand of a CR user A (denoted by CA ) is met by aggregating the transmission rates of several selected channels. Note that CA can vary from one packet to another. • A prespecified control channel with Fourier bandwidth Bc is available, where Bc B. This channel does not need to be reserved for the CRN. It can, for example, be one of the subchannels in an ISM band. • Contending CR users follow similar interframe spacings and collision avoidance strategies of the 802.11 protocol (over the control channel) by using physical-carrier sensing and backoff before initiating control packet exchanges. We also assume that data packet sizes are significantly larger than control packets, and therefore, the use of the RTS/CTS handshake is justified.
3.6.2 Channel Access in DDMAC The channel access mechanism allows the CR transmitter and receiver to agree on the set of channels to use for communication and to allocate their rates. Rate is allocated in a manner that ensures that the power mask and the rate demands are met. A CR user A views its transmission region as K non-overlapping regions, where each region is associated with a preferable channel list Ωi (A), i = 1, . . . , K, determined according to Section 3.5. This user maintains an N -entry channel list and an m-entry transmission distance table (as described in Section 3.5). The jth entry of the channel list indicates the status of the jth channel; 1 if the channel is available and 0 if the channel is occupied or reserved by any of A’s CR neighbors. Recall that each CR user is equipped with nt transceivers. One of these transceivers is tuned to the control channel, while the other nt −1 transceivers can be tuned to any data channels. As a result, CR users can always hear control messages over the common control channel even when they are transmitting/receiving data over other data channels. Thus, every CR user listens to the control channel, and accordingly updates its channel list and transmission distance table. Suppose that CR user A has data to transmit to another CR user B at an aggregate rate demand CA . Then, A reacts as follows: • If user A does not sense a carrier over the control channel for a random duration of time, it sends an RTS message at the maximum (known) power Pmax . This Pmax
79 is constrained by the power mask imposed on the prespecified control channel. The RTS includes CA , the packet size (in bytes), and the list of all available channels at A (see Figure 3.8). RTS
Transmitter Receiver ID ID
Packet size
Rate demand (CA)
Available channel list
PCA/ EPCA
Transmitter Receiver ID ID
TX duration
Distance (d)
Assigned channel list
NCA/ ENCA
Transmitter Receiver ID ID
Distance (d)
Figure 3.8: Formats of DDMAC control packets.
• The neighbors of A (other than B) that can correctly decode the RTS refrain from accessing the control channel until they receive one of two possible control packets, denoted by EPCA and ENCA (explained below). • Upon receiving the RTS packet, B estimates the distance between A and B (d AB ) (using one of the techniques described in Section 3.5). It identifies the preferable channel list Ωi (B) that corresponds to dAB . Based on the available channels at A and B, and the instantaneous interference level over these channels as measured at B, user B removes any channel that has a received SINR less than its threshold SINR and determines the common channel list that is potentially available for A → B transmission, denoted by CCL(A, B). User B then computes the intersection between Ωi (B) and CCL(A, B) to identify a preferable set of channels for A → B (Φ(A, B)). To achieve good throughput, B sorts the channels in Φ(A, B) in a descending order of their maximum possible data rate (calculated according to Shannon’s formula4 ). Then, user B appends the rest of the common available T channels that are not in Φ(A, B) i.e., CCL(A, B) Φ(A, B) , also listed in a descending order of their maximum possible data rate, to the bottom of the sorted
preferable channels. User B cumulatively adds channels from the top of the new 4
Other rate-vs-SINR relationships, such as a staircase function, can be used for calculating the
achievable data rates.
80 sorted list until either the aggregate rate CA is satisfied or the list is exhausted, i.e., no feasible channel assignment is found. • If there is no feasible channel assignment, then B responds by sending a NegativeChannel-Assignment (NCA) message that includes the distance dAB (see Figure 3.8). The purpose of this packet is to help B’s neighbors estimate the network distance traffic pattern and prompt A to back off and retransmit later. If B can find a set of available channels that can support a total demand CA , it sends a PositiveChannel-Assignment (PCA) message to A, which contains the assigned channels for the transmission A → B, the distance dAB , and the duration needed to hold the assigned channels for the ensuing data transmission and corresponding ACK packet. The PCA packet implicitly instructs B’s CR neighbors to mark the set of assigned channels as unavailable for the indicated transmission duration. It also helps these neighbors estimate the network distance traffic pattern. • Depending on which control message is received, user A reacts as follows: – If A receives an NCA message, it responds by sending an Echo-NCA (ENCA) message, which includes the distance dAB . The purpose of this packet is to help A’s neighbors estimate the network distance traffic pattern. – If A receives a PCA message, it replies back with an Echo-PCA (EPCA) message, informing its neighbors of the selected channel list, the distance dAB , and the transmission duration. This EPCA also announces the success of the control packet exchange between A and B to A’s neighbors, which may not have heard B’s PCA. • Once the RTS-PCA-EPCA exchange is completed, the data transmission A → B proceeds. Once completed, B sends back an ACK packet to A over the best assigned channel, i.e., the channel that has the highest rate. A time diagram of the RTSPCA-EPCA-DATA-ACK exchange is depicted in Figure 3.9. It is worth mentioning that there is no interference between data and control packet transmissions because the two are separated in frequency. Therefore, a CR user that hears the RTS packet from A defers its attempt to access the control channel until it receives
81 ABA CDC R P T C S A
E P C A
R P T C S A
E P C A
……….. ACK
Data TX 1
DATA A->B ACK
Data TX 2
. . .
DATA C->D
Figure 3.9: RTS-PCA-EPCA-DATA-ACK packet exchange. an EPCA or an ENCA packet from A. In addition, a CR user that receives only a PCA or an NCA should defer its attempt to access the control channel for the expected time of the EPCA/ENCA packet (to avoid a collision between control packets). This allows for more parallel transmissions to take place in the same neighborhood (see Figure 3.9). Remark: DDMAC’s channel assignment is performed on a per-packet basis, with the assigned channels on the interfaces dynamically changing (i.e., for each packet or a handful of packets) based on channel availability. This type of channel assignment requires channel switching to occur at a very small time scale, which is in the range of micro-seconds5 . This is a typical requirement for opportunistic spectrum sharing due to the high fluctuations in channel availability due to the dynamics of PR users.
3.6.3 Spatial Reuse and DDMAC Note that, we consider a CSMA/CA-based multi-hop CRN environment, which consists of multiple contention regions (neighborhoods). For these contention regions, spatial reuse is possible. Specifically, non-neighboring CR users may access the same channels on a different coverage area. This allows multiple CR transmissions to proceed simultaneously (spatial frequency reuse). To illustrate the idea of spatial reuse, Figure 3.10 depicts two scenarios for the operation of DDMAC. In the first scenario (Figure 3.10(a)), the two transmitters A and C cannot hear each other’s control packets. So, according to CSMA/CA, the transmissions A → B and C → D can overlap in their data channels, i.e., 5
Current radio technology allows channel switching to be done in a few microseconds (i.e.,
< 10 µs [23, 51])
82 the assigned channels for A → B transmission are reserved only within the area of A’s and B’s control range (spatial reuse case). In Figure 3.10(b), node C falls in the control region of node A (and vice versa). The exclusive channel occupancy policy prevents A and C from using common channels. However, the two transmissions can proceed simultaneously if A and C can find two non-intersecting sets of channels to support their rates. * *
CR user
*
*
C
*
*
*
*
* A
*
B
*
Users in PRN 1 Users in PRN 2
D
*
*
*
*
* *
*
C
*
+
*
D A
B
*
*
(a) Allowed channel reuse
(b) Unallowed channel reuse
Figure 3.10: Scenarios in which a CR transmitter C can/cannot reuse the channels assigned to A. Solid circles indicate data-transmission ranges, whereas dashed circles indicate control-transmission ranges.
3.6.4 Worst-Case Scenarios for DDMAC We illustrate two extreme scenarios under which the DDMAC protocol gracefully degrades into the BMC scheme. Recall that a CR receiver A divides its transmission range into m regions. • Scenario I: At a given time t, assume that the weighted average pmf {pei (t) : i = 1, . . . , m} has a value of 1 at i = m and 0 otherwise (i.e., most likely, transmission
distances are within Dm ). This scenario represents the case where all A’s neighbors are located near the border of its transmission range (Figure 3.11(a)). According to the proposed channel assignment algorithm, the preferable channel list is identified
83 as follows: Ωi (A) = φ : i = 1, . . . , m − 1. Ωi (A) = N : i = m. Recall that N denotes the set of all available channels.
rc d m-1
A d1
rc
A d1
d m-1
(a) Scenario I
(b) Scenario II
Figure 3.11: Illustration of two worst-case scenarios in DDMAC.
• Scenario II: At a given time t, assume that the weighted average pmf {pei (t) : i = 1, . . . , m} has a value of 1 at i = 1 and 0 otherwise (i.e., most likely, transmission
distances are within D1 ). This scenario represents the case where all A’s neigh-
bors are located close to A (Figure 3.11(b)). According to the proposed channel assignment algorithm, the preferable channel list is identified as follows: Ωi (A) = N : i = 1. Ωi (A) = φ : i = 2, . . . , m. According to DDMAC, the sorted channel list from which a CR user assigns channels to its transmission is constructed by appending the common sorted available channels that are not in the sorted preferable channels to the bottom of the sorted preferable channels list. Thus, for the above two scenarios and depending on the transmitter-receiver distance, the sorted channel list of DDMAC is as follows: • If the distance falls in the range (dm−1 , dm = rc ] or (0, d1 ], the preferable channel list is the set of all available channels. Therefore, the sorted channel list of DDMAC
84 is the same as that of the BMC scheme. Consequently, DDMAC gracefully degrades into the BMC scheme. • If the distance falls within the transmission range Rc but not in the range (dm−1 , dm ] or (0, d1 ], the preferable channel list is empty whereas the available channel list contains all the available common channels. Therefore, the sorted channel list of DDMAC is the same as that of BMC. Consequently, DDMAC gracefully degrades into the BMC scheme.
Protocol Overhead Claim 2: DDMAC and BMC have comparable overheads. Proof: Both DDMAC and BMC use a three-way handshake to send one data packet. Thus, DDMAC does not introduce any additional control message overhead.
3.7 Protocol Evaluation We now evaluate the performance of the DDMAC via simulations and compare it with CSMA/CA variants. Our results are based on simulation experiments conducted using CSIM (a C-based, process-oriented, discrete-event simulation package [3]). Each CR user generates packets according to a Poisson process with rate λ (in packet/time slot), which is the same for all users. For simplicity, data packets are assumed to be of a fixed size (2 Kbytes). Each CR user requires an aggregate transmission rate of 5 Mbps. We divide time into slots, each of length 3.3 ms. A time slot corresponds to the transmission of one CR packet at a rate of 5 Mbps. We set the CRN SINR threshold to 5 dB and (i)
the thermal noise to Pth = 10−21 Watt/Hz for all channels. Because DDMAC and the compared with CSMA/CA-based protocols have the same maximum transmission ranges and use the same channel access mechanism, it is reasonable to assume that all protocols achieve the same forward progress per hop. Consequently, our performance metrics are: (1) one-hop throughput, i.e., the destination of a packet is restricted to one hop from the source, (2) connection blocking rate, and (3) the fairness index [31]. The connection blocking rate is defined as the percentage of CR packet requests that are blocked due to the unavailability of a feasible channel assignment. We use Jain’s fairness index [31] to
85 quantify the throughput fairness of a scheme6 . Fairness index values closer to 1 indicate better fairness. The signal propagation model in (3.4) is used with n = 4, the antenna length (D) is 5 cm, and Gt (f ) = Gr (f ) = 1 for every carrier frequency f .
3.7.1 Single-hop Scenarios Simulation Setup We first simulate a small-scale network for the purpose of highlighting the advantages and operational details of DDMAC. DDMAC is compared to three multi-channel CSMAbased protocols that use different channel selection schemes: an optimal scheme (which uses exhaustive search), the BMC scheme [43, 30] (which is based on a greedy strategy that selects the best available channels for a given transmission), and a naive scheme (which always tries to select high-frequency channels if available for a given transmission, while leaving low-frequency channels for other users). Specifically, we consider a single-hop CRN, where all users can hear each other. This CRN coexists with two PRNs in a 100 meter × 100 meter field. The PRNs operate in the 600 MHz and 2.4 GHz bands. Each PRN band consists of one channel of bandwidth 1.5-MHz. The number of PR users in each PRN is 50. Each user in the ith PRN acts as an ON/OFF source, where it is ON while transmitting and OFF otherwise. We define the “activity factor” α i as the fraction of time in which the ith type PR user is ON (i.e., the probability that the source is in the ON state). The source is further characterized by the distribution of its ON and OFF periods that are both taken to be exponential. We set the average ON period to be the duration of one time slot. Therefore, the traffic correlation is captured using a two-state Markov model. The appropriateness of the 2-state ON/OFF model has been demonstrated in several previous works, e.g., [64, 53, 61, 62, 59]. In essence, the ON/OFF behavior is attributed to the bursty nature of many types of network traffic, including data transfer and VBR video streaming. The αi probabilities for the two PRNs are 0.5 and 0.3, respectively. The transmission power for each PR user is 0.5 Watt. 6
For our simulation setup, CR user demands are uniform. The destination CR user is uniformly
selected from the one-hop neighbors and the packet generation rate are the same for all CR users. Thus Jain’s fairness index provides a meaningful metric for comparing the fairness of DDMAC and BMC.
86 For the CRN, we consider 40 mobile users. We assume that a CR user can use up to (1)
(2)
two data channels simultaneously. We set the interference mask to Pmask = Pmask = 50 mW. The random waypoint model is used for mobility, with the speed of a CR user uniformly distributed between 0 and 2 meters/sec. We also set the forgetting factor α to 0.6, the observation window Twin = 0.5 second, and the number of rings around a CR user m = 12. For a fair comparison, we let all schemes use the maximum allowable power vector Pmask .
Results Under the above setup, Figure 3.12 shows that DDMAC improves the overall one-hop throughput by up to 25% compared to BMC and 34% compared to the naive approach. In addition, its throughput is within 7% of the optimal one, obtained via exhaustive search. Note that the exhaustive search implies that the instantaneous SINR values, location information, and rate demands are known to the decision-making entity that assigns channels to CR users (i.e., such search requires global information). Even if perfect knowledge of the SINR of each link and the rate demands are available, for large-scale networks, finding the optimal solution requires exhaustive search over a large state space, which grows exponentially with the number of CR users and the available channels.
Throughput (Packet/time slot)
1.5 1.25 1 0.75
Optimal DDMAC BMC Naive approach
0.5 0.25 0 0
10
20 30 λ (Packet/sec)
40
Figure 3.12: Throughput vs. λ for a small-scale network (comparison with the optimal scheme).
87 3.7.2 Multi-hop Scenarios Simulation Setup We now evaluate the performance of the DDMAC protocol under more realistic (largescale) network scenarios and contrast it with a typical multi-channel CSMA-based protocol that uses BMC as a channel selection scheme [43, 30]. We consider four PRNs and one CRN. Users in each PRN are uniformly distributed over a 500 meter × 500 meter area. The PRNs operate in the 600 MHz, 900 MHz, 2.4 GHz, and 5.7 GHz bands, respectively. Each PRN band consists of three non-overlapping 1-MHz channels. The number of PR users in each PRN is 300. The αi probabilities for the four PRNs are 0.5, 0.3, 0.3, 0.1, respectively. The transmission power for each PR user is 0.5 Watt. For the CRN, we consider a random-grid topology7 , where 225 mobile CR users are placed within the 500 meter × 500 meter field. The field is split into 225 smaller squares, one for each CR user. The location of a mobile user within the small square is selected randomly. For each generated packet, the destination is selected randomly from the onehop neighbors. Within each small square, the random waypoint model is used for CR mobility, with the speed of a CR user uniformly distributed between 0 and 2 meters/sec. We assume that a CR user can use up to three data channels simultaneously. We set (1)
(2)
(12)
the interference mask to Pmask = Pmask = . . . = Pmask = 50 mW. Note that, in our design, we assume that DDMAC enforces an exclusive channel occupancy policy on CR transmissions (i.e., no CR-CR interference). However, hidden-terminal problem can occur in this scenario due to imperfect control. Our simulations relax these assumptions and account for all sources of interference, including those that are far away from a receiver and use common channels. The reported results are the average of 100 experiments. Remark: Our simulations only address the MAC layer aspects and assume that route computations have already been carried out. Using a destination from a node’s one-hop neighbor is used only to convey the need for channel access and does not have any implications on the application. A “destination” in this context could be the next hop where a packet should be forwarded to or the final packet destination. Randomly 7
Note that random-grid topologies are realistic topologies. The random-grid topology models
constrained scenarios. For example, a building could have various offices, where each office may contain several wireless devices.
88 selecting neighbor as a destination is realistic in terms of packet forwarding, especially when multiple flow streams (like file transfers, messaging, or VoIP) pass through a node.
Blocking rate (%)
0.45 BMC DDMAC
0.4
0.35
0.3 0
0.1
0.2 0.3 0.4 Packet genaration rate
0.5
Single hop througput (Packet/time slot)
(a) Blocking rate vs. λ.
40 35 30 25 20 15 10
BMC DDMAC BMC−Raleigh DDMAC−Raleigh
5 0 0
0.1 0.2 0.3 0.4 Packet generation rate (Packet/sec)
0.5
(b) Throughput vs. λ.
Figure 3.13: Performance of a CRN.
Results We first compare the performance of DDMAC to that of the BMC scheme. For a fair comparison, we let both schemes use the maximum allowable power vector P mask . We set α to 0.6, Twin to 0.5 second, and m to 12. Figures 3.13(a) and (b) show that under moderate and high traffic loads, DDMAC significantly reduces the connection blocking
89 rate and improves the overall one-hop throughput by up to 30%. This improvement is attributed to the increase in the number of simultaneous transmissions in DDMAC. Note that under low traffic load, the throughput of DDMAC gracefully degrades to that of BMC due to the availability of a sufficient number of channels. The system performance under Raleigh channel model (i.e., varying channel conditions) is also investigated in Figure 3.13(b). We have considered normalized random variables to present the fading processes. The results show that, under varying channel conditions, the same trends that were noticed for AWGN channel have been observed. Figure 3.13(b) also shows that the impact on throughput performance is almost the same under AWGN and Raleigh channel models. Similar observations have also been reported in an experimental study in [63] for IEEE 802.11b wireless LAN. In [63], the authors showed that the impact on throughput and packet error rate were virtually identical under AWGN and Raleigh channel models. In Figure 3.14(a), we focus on the per-user throughput performance under DDMAC8 . As shown in this figure, although DDMAC requires a pair of CR users to communicate over a set of channels that may not be optimal from one user’s perspective, the per-user throughput of DDMAC under moderate and high traffic loads is still greater than that of the BMC scheme. This is because DDMAC attempts to serve a given CR transmission first using the preferable channel list and preserves the “better” channels for other transmissions. However, if the aggregate rate of this transmission cannot be satisfied using the preferable list, DDMAC attempts to serve this transmission using the remaining available channels. Next, we compare the fairness index of DDMAC to that of BMC. Compared to BMC, Figure 3.14(b) shows that DDMAC slightly improves the network fairness and preserves long-term fairness properties. This improvement occurs because DDMAC motivates cooperation among neighbors to maximize their network-wide benefit. The effect of dividing the CR user’s transmission range is depicted in Figure 3.15(a) for different values of λ. As m increases, the throughput also increases up to a certain point. For m ≥ 12, no significant improvement is observed in the CRN throughput. This is because the proposed channel assignment mechanism merges the m regions into K ≤ m regions, i.e., over-splitting Rc is not useful. In Figure 3.15(b), we study the impact of α and Twin on the performance of DDMAC. 8
This figure shows the average worst-case throughput performance among all CR users.
90 Scheme BMC DDMAC(Twin DDMAC(Twin DDMAC(Twin DDMAC(Twin DDMAC(Twin
= 0.03 s) = 0.3 s) = 0.4 s) = 1 s) = 4 s)
α∗ 0.1 0.6 0.6 0.8 1.0
Best throughput (packet/slot) 25 26 33.6 33.85 33.89 28
Table 3.2: Performance of DDMAC at the optimal α as a function of Twin . We set λ = 0.3 packet/slot. The network throughput versus α for different values of Twin is shown in the figure. It is clear that the throughput depends on the choice of α and Twin . As Twin increases, α should increase to give much more importance to recent observations without entirely discarding older observations. Table 3.2 shows the best throughput performance and the associated optimal value of α (α ∗ ), obtained from simulation, for different values of Twin . It is clear that if Twin is too small or too large, the throughput reduces significantly. We also investigate the robustness of DDMAC to inaccurate distance estimation, which mainly results from multi-path propagation, reflection, and fading effects. We introduce uniform estimation errors (ξ ∼ Uniform[−, ]) into the distance d. Thus, the estimated distance de is given by (1 + ξ) d. Figure 3.16(a) shows the effect of inaccurate distance estimation on the perceived throughput as a function of under different traffic loads.
As the figure indicates, there are no significant changes in the throughput for different values of . Furthermore, Figure 3.16(b) gives the percentage of reduction in throughput due to inaccurate d as a function of λ for different values of . This figure shows that the maximum percentage of reduction in throughput due to inaccurate d is less than 6%. The results in Figure 3.16 indicate that channel assignment in DDMAC is quite robust to distance estimation errors. This is because DDMAC requires only rough estimates of user distribution, distances among users, and local traffic conditions in order to dynamically adapt channel assignments to current network traffic. Finally, we study the end-to-end throughput for both BMC and DDMAC. Specifically, for each generated packet, the destination node is randomly selected to be any node in the network. We use a min-hop routing policy, but we ignore the routing overhead. For both schemes, the next-hop candidates are nodes that are within the transmission range of
91 the transmitter. Figure 3.17 shows that under moderate and high traffic loads, DDMAC significantly improves the overall network throughput (inline with the results in Figure 3.13(b)).
3.8 Conclusions In this chapter, we proposed an opportunistic distance-dependent MAC protocol for CRNs (DDMAC). DDMAC improves the CRN throughput through cooperative channel assignment, taking into consideration the non-adjacency of frequency channels and the imposed power masks. We presented a heuristic stochastic channel assignment scheme that dynamically exploits the dependence between the signal attenuation model and the transmission distance. Our scheme accounts for traffic dynamics. It assigns channels with lower average SINR to shorter transmission distances to increase the number of simultaneous transmissions. We integrated the channel assignment process in the design of DDMAC. We compared the performance of DDMAC with that of a reference multi-channel MAC protocol that is designed for typical multi-channel systems (BMC). We showed that, under moderate and high traffic loads, DDMAC achieves about 30% increase in throughput over the BMC scheme, with manageable processing overhead. Although DDMAC requires a pair of CR users to communicate on a channel that may not be optimal from a user’s perspective, we showed that the average per-user throughput of DDMAC under moderate and high traffic loads is greater than that of the BMC scheme. Furthermore, DDMAC preserves (even slightly improves) throughput fairness relative to BMC. In summary, DDMAC provides better spectrum utilization by reducing the connection blocking probability and increasing the system throughput. To the best of our knowledge, DDMAC is the first CRN MAC protocol that utilizes the radio propagation characteristics to improve the overall network throughput.
Single−hop throughput (packet/ time slot)
92
0.12 0.1 0.08 0.06 0.04 BMC DDMAC
0.02 0 0
0.1
0.2 0.3 Packet generation rate
0.4
0.5
(a) Per-user single-hop throughput.
1
Fairness Index
0.9 0.8
BMC DDMAC
0.7 0.6 0.5 0
0.1
0.2 0.3 0.4 Packet generation rate
0.5
(b) Fairness index vs. λ.
Figure 3.14: Per-user throughput and fairness Performance.
Single−hop throuhput (Packet/time slot)
93
40
λ = 0.2 λ = 0.25 λ = 0.3
35
30
25
20
5
10 m
15
Single−hop throuput (Packet/time slot)
(a) Throughput vs. number of rings (m) around a CR user.
38
Twin = 0.03 s
36
Twin = 0.3 s
34
Twin = 1 s
32
λ = 0.3
Twin = 4 s
30 28 26 24 22 0.1
0.2
0.3
0.4 0.5 0.6 0.7 Forgitting factor α
0.8
0.9
1
(b) Throughput vs. α for different Twin values.
Figure 3.15: Performance of DDMAC.
Single−hop throughput (Packet/time slot)
94
40 35 30 25 20 15 Low load
10
Moderate load
5 0 0.05
High load 0.1
0.15 ε
0.2
0.25
Percentage of reduction in throughput
(a) Throughput vs. . 6 5
ε = 5%
4
ε = 25%
ε = 15%
3 2 1 0 0
0.1
0.2 0.3 0.4 Packet generation rate
0.5
(b) Percentage of reduction in throughput vs. λ.
Figure 3.16: Impact of inaccurate distance estimation on DDMAC.
95
End−2−end Throughput (Packet/time slot)
10 8 6
BMC DDMAC
4 2 0 0
0.1
0.2 0.3 0.4 Packet generation rate
0.5
Figure 3.17: End-to-end throughput vs. λ
96
CHAPTER 4 Cooperative Adaptive Spectrum Sharing for Throughput Enhancement in Cognitive Radio Networks
4.1 Introduction Thus far, we have discussed MAC protocols that assume multiple transceivers per CR user, which may not often be the case for low-cost CR systems. In contrast, in this chapter, we consider a single half-duplex transceiver per CR node. Our goal is to improve network throughput, and conserve energy as a secondary objective, subject to PR interference constraint and user rate demands.
4.1.1 Goals and Contributions The contribution of of this chapter is three-fold. We first formulate the channel assignment and power allocation problem that maximizes the number of simultaneous CR transmissions. Then, for a single-transceiver CRNs, we present centralized and distributed cooperative channel assignment algorithms that leverage the unique capabilities of CRs. For a single-hop CRN, we develop a CSMA-based MAC protocol with access window, called AW-MAC, which realizes the optimal centralized channel assignment algorithm in a distributed manner. We then develop a novel CSMA-based MAC protocol called WFC-MAC for a multi-hop CRN. WFC-MAC improves the CRN throughput performance through cooperative channel assignment among neighboring CR users. According to WFC-MAC, a CR user that intends to transmit has to account for potential future transmissions in its neighborhood. Our protocols do not require interaction with PRNs, and can be incorporated into existing multi-channel systems (using currently available hardware) with little extra processing overhead. To evaluate the performance of our protocols, we conduct simulations for a singlehop and a multi-hop CRN with mobile users. Simulation results show that our protocols significantly improve the network throughput over two previously proposed schemes (i.e.,
97 BMC-MAC [30] and DDMAC [52]). The results also indicate that our protocols preserve fairness. For single-hop scenarios, we show that AW-MAC achieves the best throughput (up to 39% improvement over BMC-MAC scheme) at no additional cost in energy consumption. In multi-hop scenarios, WFC-MAC achieves the best throughput at the cost of energy consumption.
4.1.2 Organization The rest of this chapter is organized as follows. In Section 4.2, we introduce our system model, state our assumptions, and formulate the optimal channel assignment/power control problem. Section 4.3 introduces the centralized channel assignment algorithm. Section 4.3.2 describes the proposed AW-MAC protocol. In Section 4.4, we introduce the distributed channel assignment algorithm and the proposed WFC-MAC protocol. In Section 4.5, we analysis the throughput of our proposed protocols. Section 4.6 presents our simulation results. The related previous work is reviewed in Section 4.7. Our concluding remarks are presented in Section 4.8.
4.2 Problem Formulation and Design Constraints 4.2.1 Network Model We consider a decentralized opportunistic CRN that geographically coexists with K different PRNs. The PRNs are licensed to operate on non-overlapping frequency bands. The kth PRN, k = 1, . . . , K, occupies mk adjacent but non-overlapping frequency channels, each of Fourier bandwidth W . Let M denote the set of all non-overlapping channels in P all PR bands, and let M = |M| = K k=1 mk . For k = 1, . . . , K, let fk and Bk = mk W
respectively be the carrier frequency and bandwidth associated with the kth PRN (in Hz).
CR users continuously scan the spectrum, identifying potential spectrum holes and exploiting them for their transmissions.
These users also control their transmission
powers to avoid no harmful interfering with PR receptions. The problem of identifying spectrum holes and selecting appropriate powers is overcomplicated by the presumingly non-cooperative nature of the PRNs, which usually do not provide online feedback to CR users. To address this issue, a multi-level frequency-dependent power mask (1)
(2)
(M )
(Pmask = {Pmask , Pmask , . . . , Pmask }) on the CR transmissions is often adopted [52]. En-
98 forcing such a power mask allows for simultaneous spectrum sharing between neighboring CR and PR users. According to this approach, CR users can exploit both idle and partially-utilized bands, potentially leading to better spectrum utilization. Note that the multi-level power mask is a generalization of the widely-used binary-level power mask, whereby for band i, i = 1, . . . M , the CR transmission power is 0 if any PR user operates (i)
on band i, or Pmax 1 if none of them is active. The power mask can also be made timedependent to reflect the fluctuations in channel availability due to the dynamics of PR users. Determination of an appropriate multi-level power mask is a challenging problem, which has been investigated under certain simplifying assumptions (e.g., [19, 53]). Specifically, in [53], the authors provided a multi-level frequency- and neighborhood-dependent adaptive power mask on CR transmissions that ensures a statistical guarantee on the performance of PRNs. This adaptive power mask is computed dynamically based on the traffic profiles and interference margins of PR users. For our purposes, we assume that a similar mechanism for determining Pmask is in place. Without loss of generality, we consider a scenario where an idle PRN channel is sufficient to support one CR packet transmission. This is an acceptable assumption in many wireless systems that are built to operate in the unlicensed bands, including IEEE 802.11/a/b/g-compliant devices. Let N denote the set of all N CR transmission requests at a given time and in a given neighborhood. While not transmitting, a CR user periodically measures the levels of interference over the M data channels and estimates the total noise-plus-interference power vector. This can be done, for example, via a sequential partial sensing approach (1)
(2)
(M )
with negligible switching/sensing overhead [15, 50]. Let Ij = {Ij , Ij , . . . , Ij
} denote
the total noise-plus-interference power vector as measured by the CR receiver of the jth CR transmission.
4.2.2 Feasibility Constraints For the CR jth transmission, both the transmitter and receiver need to cooperatively select appropriate frequency channels and transmission powers over that channels while meeting the following constraints: 1
(i)
Pmax is the smaller of the FCC regulatory maximum transmission power over band i and the
maximum power supported by the CR’s battery.
99 1. Exclusive channel occupancy policy: A selected channel cannot be assigned to more than one CR transmission in the same neighborhood (inline with the CSMA/CA mechanism). 2. nt transceivers per CR user : Each CR user is equipped with nt transceivers, which can be used simultaneously. The operation is half-duplex, i.e., while transmitting over given channels, the CR user cannot receive/listen, even over other channels. 3. CR-to-PR interference constraints: For a CR transmission j and a selected channel (i)
i, the transmission power Pj i.e.,
(i) Pj
≤
must satisfy the power constraint over that channel,
(i) Pmask .
4. Rate demands: Each CR transmission j requires a given data rate Rj . 5. Total power constraint: The total transmission power over the selected channels is limited to Pmax . This constraint is imposed, for example, by the CR’s battery.
4.2.3 Problem Statement Maximum Number of CR Transmissions Multi-transceiver Problem Given the set of CR transmission requests N , the set of available channels M, the power (i)
mask vector Pmask , ∀i ∈ M, the maximum total transmission power Pmax , the number of transceiver per CR user nt , and the rate demand Rj , ∀j ∈ N , our goal is to maximize the number of simultaneous CR transmissions subject to the previously mentioned constraints and such that: a. Every CR request (link) is either assigned the required rate demand or that link is blocked. b. A CR user cannot transmit and receive at the same time. c. A CR transmitter cannot transmit to more than one destination at a time. d. A CR receiver cannot receive from more than one transmitter at a time. e. A CR transmitter (receiver) can transmit (receive) over up to n t channels.
100 Maximum Number of CR Transmissions Single-transceiver Problem Given the set of CR transmission requests N , the set of available channels M, the power (i)
mask vector Pmask , ∀i ∈ M, the maximum total transmission power Pmax , the number of transceiver per CR user nt = 1, and the rate demand Rj , ∀j ∈ N , our goal is to maximize the number of simultaneous CR transmissions subject to the previously mentioned constraints and such that: a. Every CR request (link) is either assigned the required rate demand or that link is blocked. b. A CR user cannot transmit and receive at the same time. c. A CR transmitter cannot transmit to more than one destination at a time. d. A CR receiver cannot receive from more than one transmitter at a time. e. A CR transmitter (receiver) can transmit (receive) over over only one channel.
4.2.4 Problem Formulation Formally, our goal is to compute a feasible channel assignment that assigns channels and transmission powers to CR requests such that the number of simultaneous CR transmissions is maximized subject to the previously mentioned constraints. If multiple solutions exist for this optimization problem, then we seek the one that requires the least amount of energy. (i)
Let αj be a binary variable that is defined as follows: (i)
αj
1, if channel i is assigned to transmission j = 0, otherwise.
(4.1)
101 The maximum rate resource assignment problem is stated as follows: i X X h (i) (i) maximizeα(i) ∈{0,1},P (i) 1 α j rj ≥ R j j
j
i∈M j∈N
s.t.
X
(i)
αj ≤ 1, ∀i ∈ M
j∈N
0≤
(i) Pj
X
≤
i∈M (i) Pmask , ∀i
X
(i)
(i)
αj ≤ nt , ∀j ∈ N ∈ M and ∀j ∈ N
(i)
αj P j
≤ Pmax , ∀j ∈ N
(4.2)
i
where 1[.] is the indicator function, (i)
(i) rj
def
= W log2 1 +
(i)
(i)
G j Pj (i)
Ij
is the Shannon capacity
for link j on channel i, and Gj is the frequency-dependent attenuation factor associated with the jth transmission over channel i. The first two constraints ensure that at most nt channel can be assigned per transmission and a channel cannot be assigned to more than one transmission. The last two constraints reflect the maximum allowable transmission power over a given channel and the maximum total transmission power for each transmission, respectively. The optimization problem in (4.2) is a mixed integer non-linear program (MINLP) and is NP-hard2 for the multi-transceiver case (i.e., maximum rate multi-transceiver problem) [11, 12]. However, in the special case of single-transceiver per CR user (i.e., maximum rate single-transceiver problem), this MINLP is not NP-hard and may be solved optimally in polynomial time. Specifically, the maximum rate single-transceiver problem is the same as assigning channels to independent (distinct) links such that the number of CR transmissions is maximized. Hence, this problem corresponds to the well-known maximum weighted bipartite matching problem, which has a polynomial-time solution [56] (page 418). 2
The solution complexity grows exponentially with the number of CR users and the available
channels.
102 4.3 Optimal Channel Assignment In this section, we first present a centralized algorithm for the channel assignment problem based on bipartite matching. The objective of this algorithm is to maximize the total number of simultaneous CR transmissions by means of power management. Note that centralized algorithms are easy to implement in single-hop networks where all users are within radio range of each other.
4.3.1 Proposed Algorithm In our context, a centralized algorithm implies that the instantaneous SINR values, location information, and rate demand are known to the decision-making entity that assigns channels and transmission powers. For a finite number of available channels and given rate demands, a CR user computes the minimum required power over each channel. Using this fact and noting that the graph connecting the set of CR transmission requests and the set of available channels is a bipartite graph3 , our optimization problem can be transformed into a bipartite perfect matching problem. The maximum matching of this bipartite graph problem is the set containing the maximum number of CR transmissions that can proceed simultaneously. If there are multiple feasible channel assignments with maximum matching, the one requiring the smallest total transmission power will be selected. In the following, we develop an algorithm that transforms our optimization problem into a bipartite perfect matching problem. Formally, the algorithm proceeds in three steps: Step 1. Compute the minimum required powers: For every CR transmission request j ∈ N and every channel i ∈ M, the algorithm computes the minimum required (i)
transmission power Pj,req that can support the rate demand Rj , i.e.,
(i)
Pj,req =
2
Rj W
(i)
− 1 Ij (i)
Gj
.
(4.3)
Then the algorithm identifies every possible channel/transmission combination (i, j) whose (i)
(i)
(i)
Pj,req satisfies Pj,req ≤ Pmask . Let χ be the set of all such combinations. We say the channel/transmission combination (i, j) is prohibited if (i, j) * χ. 3
A bipartite graph is a graph whose vertex set can be decomposed into two disjoint sets such
that no two vertices in the same set are connected.
103 Links
Channels
Figure 4.1: Bipartite graph with M = N = 3. Step 2. Formulate the perfect bipartite matching problem: The algorithm creates M nodes, each corresponding to one of the available channels. Let these nodes constitute the channel set C. The algorithm also creates N nodes to represent the CR transmission requests. Let these nodes constitute the request set R. If N > M , the algorithm creates N − M additional nodes CD = {M + 1, . . . M + N } to represent dummy S channels and updates C as C = C CD . On the other hand, if N < M , the algorithm
creates M − N additional nodes RD = {N + 1, . . . M + N } to represent dummy requests S and updates R as R = R RD . Then, the algorithm connects the nodes in C to the nodes
in R. Any (i, j) assignment that contains a dummy node is also a prohibited assignment. (i)
Let wj
denote the arc weight of link (i,j) on the bipartite graph. For all prohibited (i)
assignments, the algorithm sets wj to a very large number Γ. Formally, w(i) = P (i) , if P (i) ≤ P (i) , j ∈ R and i ∈ C j,req j,req j mask (i) (i) w(i) = Γ, if Pj,req > Pmask , j ∈ R and i ∈ C. j
(4.4)
Figure 4.1 shows an example of the bipartite graph with M = N = 3.
The above bipartite graph transforms the assignment problem into a perfect bipartite matching (because the number of CR transmissions is equal to the number of channels, and every node in the request set is connected to every node in the channel set). Now, the algorithm seeks a one-to-one assignment for the max{M, N } × max{M, N } bipartite graph constructed in Step 2, with the weights defined in (4.4), so as the channel utilization is maximized while selecting the minimum transmission powers. The optimal solution of this optimization problem can be found using the Hungarian algorithm, which has a
104 polynomial-time complexity and codes are readily available for its implementation [21, 56]. Remark: The above problem may also be solved as a flow problem, i.e., computing a maximum number of disjoint paths with minimum sum of link weights. Step 3. Find the optimal resource allocation: The algorithm removes all prohibited assignments in the optimal solution of the perfect bipartite matching by setting (i)
them to 0, and modifies αj as follows: α∗(i) = 1, if α(i) = 1 and w (i) < Γ, j ∈ R and i ∈ C j j j α∗(i) = 0, if α(i) = 1 and w (i) = Γ, j ∈ R and i ∈ C. j j j ∗(i)
(4.5)
Using the revised optimal assignment {αj }, the algorithm computes Z P P ∗(i) i∈C j∈R αj . Depending on Z, there are two possibilities:
=
• If Z = 0, there is no feasible channel assignment.
• If Z > 0, there is a feasible channel assignment. In this case, Z represents the maximum number of possible concurrent transmissions.
4.3.2 Channel Access Protocol for Single-hop CRNs Assumptions For each frequency channel, we assume that its gain is stationary for the duration of a few control packets and one data packet. This assumption holds for typical mobility patterns and transmission rates [42]. We also assume symmetric gains between two users, which is a common assumption in RTS/CTS-based protocols, including the IEEE 802.11 scheme. Our protocols assume the availability of a prespecified control channel of Fourier bandwidth Bc , where Bc W , i = 1, . . . , M . Such a channel is not necessarily dedicated to the CRN. It may, for example, be one of the unlicensed ISM bands. Note that the existence of a common control channel is a characteristic of many MAC protocols proposed for CRNs (e.g., [52, 55, 68, 46, 53]).
Operational Details To execute the centralized algorithm presented in the previous section in a distributed manner, we require the instantaneous SINR information of all contending CR users in a
105 given locality to be known to all CR users in that locality before assigning channels and transmission powers. In a single-hop network, this issue can be handled during the “admission phase” by introducing a contention period known as the access window (AW). The AW consists of M fixed-duration access slots (AS). A series of control packet exchanges take place during these slots, after which several data transmissions can commence concurrently. We note here that the use of an AW for contention was originally proposed in the MACA-P protocol [8] and was later integrated in the design of POWMAC [42]. However, in both protocols the objective was not to address spectrum sharing (channel assignment), but rather to prevent collisions between control and data packets (in MACAP) and to address single-channel transmission power control (in POWMAC). During the AW, communicating CR users announce their instantaneous SINR information. A CR user that has packets to transmit and that is not aware of any already established AW in its neighborhood can asynchronously initiate an AW. Each AS consists of the sum of an RTS duration, a CTS duration, and a maximum backoff interval (explained below), and two fixed short interframe spacing (SIFS) periods4 . Control packets are sent at the maximum (known) power Pmax . This Pmax is constrained by the maximum permissible transmission power imposed on the control channel. Upon receiving an RTS packet from a CR user, say A, that is initiating an AW, other CR users in the network synchronize their time reference with A’s AW. AW M T ctrl Tctrl Ctrl
Data+Ack Tdata
AW
M T ctrl
Tctrl
……..
……..
Data+Ack Tdata …….. ……..
CH1
CH2
……..
. . . . . .
. . . . . .
. . . . . .
. . . . . .
……..
CH M
t AS
AS
Figure 4.2: Basic operation of AW-MAC 4
As defined in the IEEE 802.11b standard [2], a SIFS period consists of the processing delay
for a received packet plus the turnaround time.
106 Suppose that a CR user C overhears A’s RTS, and has a data packet to send. C contends for the control channel in the next access slot of A’s AW as follows. It first backs off for a random duration of time (T ) that is uniformly distributed in the interval [0, Tmax ]; Tmax is a system-wide backoff counter. After this waiting time and if no carrier is sensed, user C sends its RTS packet in the current AS. Note that Tmax is in the order of a few microseconds whereas a time slot is in milliseconds, so the backoff mainly serves to prevent synchronized RTS attempts. For illustration purposes, Figure 4.2 shows a time diagram of the channel access process, assuming fixed data-packet sizes and equal rate demands. Tctrl and Tdata in the figure denote the durations (in seconds) of one RTS/CTS packet exchange and one data plus ACK packets transmissions, respectively. After all the control packets have been exchanged, the channel assignment and power management algorithm of Section 4.3 is executed at every communicating node. Remark: Another design possibility that can achieve improvement in the CRN throughput is to use two half-duplex transceivers per CR user. One transceiver would be tuned to the control channel, while the other could be tuned to any data channel in M. In this case, there is no interference between data and control transmissions because the two are separated in frequency. This way, the reservations of the subsequence AW can be conducted while current data transmissions are taking place (i.e., mimicing a fullduplex operation). This reduces the control overhead and improves the overall throughput at the cost of an additional transceiver. We refer to the channel access mechanism that uses AW assignment with one transceiver as AW-MAC, and the one that uses AW assignment with two transceivers as 2-radio AW-MAC. Figure 4.3 shows the basic operation of 2-radio AW-MAC. In Section 4.5, we study the potential throughput improvement of 2-radio AW-MAC over AW-MAC.
4.4 Distributed Channel Assignment for Multi-hop CRNs In this section, we present a distributed channel assignment scheme for a multi-hop CRN. This scheme aims at approximating the centralized algorithm presented in Section 4.3. It attempts to improve spectrum utilization in a purely distributed manner while relying only on information provided by the two communicating nodes. We first identify the key challenges involved in realizing the centralized algorithm in a distributed manner. Then,
107 AW MT ctrl
Ctrl
Data+Ack Tdata
T ctrl T ctrl
T ctrl Tctrl
…….
…….
Data+Ack T data T ctrl
Data+Ack T data Tctrl
…….
……..
……..
CH 1 ……..
CH 2
CH M
…….. . . . . . .
. . . . . .
AS
. . . . . .
……..
……..
MT ctrl
MT ctrl
AW
AW
t
Figure 4.3: Basic operation of 2-radio AW-MAC.
we describe our distributed scheme in detail.
4.4.1 Challenges To execute our centralized algorithm in a multi-hop environment, the algorithm must run in a distributed manner at each CR user in a given locality (i.e., contention region). This implies that each CR user that belongs to a contention region must exchange instantaneous SINR information with other neighboring CR users in that region before selecting channels and powers. This incurs high control overhead and delay. Moreover, in a multihop environment, CR users may belong to multiple contention regions that differ in their channel availability. To overcome such challenges, we develop a heuristic channel assignment scheme that provides a suboptimal solution with low complexity and that achieves good spectrum utilization.
4.4.2 Channel Assignment The main consideration in our distributed scheme is to enable cooperation among neighboring CR users. A CR user that intends to transmit has to account for potential future transmissions in its neighborhood. It does that by assigning to its transmission the worst feasible channel, i.e., the least-capacity available channel that can support the required rate demand. We refer to this approach as the worst feasible channel (WFC) scheme. Note that a user determines the worst feasible channel for its transmission using only local in-
108 formation. WFC scheme preserves better channels for potential future CR transmissions. Even though WFC requires a pair of CR users to communicate on a channel that may not be optimal from one user’s perspective, it allows more CR transmissions to take place simultaneously, especially under moderate to high traffic loads. Compared to previously proposed channel assignment schemes (evaluated in Section 4.6), our approach avoids unnecessary blocking of CR transmissions, and has a great potential to improve network throughput by means of cooperative channel assignment.
4.4.3 Channel Access Protocol Protocol Overview Based on the WFC algorithm, we propose a distributed multi-channel MAC protocol for multi-hop ad hoc networks with a single half-duplex radio per node. The proposed protocol is an extension of the single channel RTS-CTS-DATA-ACK handshaking scheme used in the 802.11 standard. It differs from previous designs in that it exploits the “dualreceive single-transmit” capability of radios (i.e., each radio is capable of receiving over two channels simultaneously, but can transmit over one channel at a time). The operation is half-duplex, i.e., while transmitting, the radio cannot receive/listen, even over other channels. It can be implemented using one transceiver with slight upgrade in the receive chains of the circuitry. This capability is readily available in some recent radios. For example, QUALCOMM’s RFR6500 radio [4] supports “simultaneous hybrid dual-receive operation, which allows for 1X paging signal monitoring during a 1xEV-DO connection, while monitoring other frequency bands for hand-off”. Another example is Kenwood’s TH-D7A Dual-Band Handheld Transceiver [2], which supports simultaneous reception over both data and voice channels using a single antenna. Though a simple enhancement of the transceiver circuitry, the dual-receive capability makes the MAC design much easier. In particular, if we assume a common (or coordinated) control channel, a CR user that is not transmitting any data can tune one of its two receive branches to the control channel while receiving data over the other receive branch. This way, the multi-channel hiddenterminal problem can be alleviated.
109 Operational Details To facilitate multi-channel contention and reduce the likelihood of CR collisions, each CR user, say A, maintains a free-channel list (FCL) and a busy-node list (BNL). The FCL(A) represents channels not occupied by other CR users within the A’s one-hop communication range. BNL(A) consists of the IDs of CR users that are currently busy transmitting/receiving data packets in A’s neighborhood. The FCL(A) and BNL(A) are continuously updated according to the channel access dynamics and overheard control packets. The proposed protocol follows similar interframe spacings and collision avoidance strategies of the 802.11 scheme (implemented here over the control channel) by using physical carrier sensing and backoff before initiating control-packet exchanges. Upon accessing the control channel, communicating CR users perform a three-way handshake, during which they exchange control information, conduct the channel assignment, and announce the outcome of this channel assignment to their neighbors. The details of the channel access mechanism are now described. Suppose that CR user A has data to transmit to CR user B at a rate demand RA . If A does not sense a carrier over the control channel for a randomly selected backoff period, it proceeds as follows: • If FCL(A) is empty or B is busy (based on BNL(A)), A backs off and attempts to access the control channel later. Otherwise, A sends an RTS message at power Pmax . The RTS packet includes FCL(A) and RA . • A’s neighbors other than B, that can correctly decode the RTS will stay silent until either they receive another control packet from A, denoted by FCTS (explained below), or until the expected time for the FCTS packet expires. • Upon receiving the RTS packet, B proceeds with the channel assignment process, whose purpose is to determine whether or not there exists a feasible channel assignment that can support RA . • Depending on the outcome of the channel assignment process, B decides whether or not A can transmit. If not, then B does not respond to A, prompting A to back off, with an increased backoff range that is similar to 802.11, and retransmit later. Otherwise, B sends a CTS message to A that contains the assigned channel, the transmit power, and the duration (Tpkt (A)) needed to reserve the assigned channel.
110 The CTS implicitly instructs B’s CR neighbors to refrain from transmitting over the assigned channel for the duration Tpkt (A). • Once A receives the CTS, it replies back with a “Feasible-Channel-to-Send” (FCTS) message, informing its neighbors of the assigned channel and Tpkt (A). Such a threeway handshake is typically needed in multi-channel CSMA/CA protocols designed for multi-hop networks (e.g., [30, 52, 59]). For single-hop networks, where all users can hear each other, there is no need for the FCTS packet. Likewise, in singlechannel multi-hop networks, the FCTS packet is also not needed. • After completing the RTS/CTS/FCTS exchange, the transmission A → B proceeds. Once completed, B sends back an ACK packet to A over the assigned data channel. Figure 4.4 illustrates a time diagram of the channel access process. TX 1 CTRL Data CH
R C T T S S
F C T S
TX 2 R C T T S S
F C T S
………..
Data TX 1
ACK ACK
Data CH
Data TX 2
. . .
t
Figure 4.4: Time diagram of control packet exchange in WFC-MAC.
When used with the WFC assignment, the above protocol is referred to as WFC-MAC. Note that, while receiving a data packet over a given data channel, a CR user still listens to other control packet exchanges taking place over the control channel, and can update its FCL and BNL accordingly. However, a CR user that is transmitting a data packet will not be able to listen to the control channel, so its FCL and BNL may become outdated. We refer to this problem as transmitter deafness, which is primarily caused by the halfduplex nature of the radios. To remedy this problem, when the receiver sends its ACK, it includes in this ACK any changes in the ACL and BNL that may have occurred during the transmission of the data packet. The transmitter uses this information to update its own tables. Remark: Because there is no interference between data and control packets, a CR user that hears the RTS (CTS) packet defers its transmission only until the end of the
111 control packet handshaking. This allows for more parallel transmissions to take place in the same vicinity.
4.5 Throughput Analysis In this section, we use simplified analysis to evaluate the maximum achievable throughput of various channel access schemes in single-hop topologies. We assume that a CR user transmits data in the form of fixed-size packets at a fixed transmission rate. Recall that Tctrl denotes the transmission duration of one RTS plus one CTS packets, and Tdata denotes the duration for data plus ACK transmissions. Assume that Tctrl can be expressed in terms of Tdata as Tctrl = δTdata . It is worth mentioning that according to the IEEE 802.11 specifications, Tdata is at least an order of magnitude larger than Tctrl (i.e., 0 < δ 1). As an example, consider data- and control- packet sizes of 4-KB and 120 bits, respectively [5]. Also consider a transmission rate of 5Mbps. Then, δ ≈ 0.0073. We now provide expressions for the maximum achievable throughput under the various schemes, defined as the maximum number of simultaneous CR transmissions that can be supported in a Tdata + M Tctrl = (1 + M δ)Tdata duration. For the single-transceiver AW-MAC, according to Figure 4.2, the maximum number of data packets that can be potentially transmitted in a Tdata + M Tctrl duration is M . Under 2-radio AW-MAC, at steady state, the maximum number of data packets that P Tctrl 2 can be potentially transmitted in the same duration is M + M i=1 M Tdata = M + M δ =
M (1 + M δ) (see Figure 4.3). Under both WFC-MAC and BMC-MAC (which is similar to WFC-MAC but uses the BMC channel assignment), for a given channel i, Figure 4.5 shows that an RTS/CTS exchange can immediately follow the transmission of the previous data
packet over that channel. Thus, the maximum achievable throughput in the Tdata +M Tctrl PM −1 ctrl duration is M + i=1 (M − i − 1) TTdata . With some manipulations, this quantity can be 2
written as M + δ( M2 − 23 M + 1).
Computing the maximum achievable throughput in this way is rather optimistic since we are assuming that for 2-radio AW-MAC/AW-MAC, all AW slots result in successful RTS/CTS exchanges, and that for the BMC-MAC/WFC-MAC and a given data channel, an RTS/CTS exchange follows immediately the transmission of the previous data packet over that channel.
112 M Tctrl Tctrl Tctrl Ctrl
Tdata ……..
……
(M-1)Tctrl ………...
CH1 (M-2)Tctrl
CH2
………...
………... . . . . .
CH M-1 CHM
Tctrl
. . . . . ………...
………...
t
Figure 4.5: Basic operation of the distributed spectrum access scheme.
Figure 4.6 shows the maximum achievable throughput as a function of M for two data-packet sizes and various channel access schemes. For practical data- and controlpacket sizes [5], where δ 1, the figures reveal that various channel access schemes achieve comparable throughput performance. More importantly, the use of two halfduplex transceivers per CR user provides a minor improvement in the system throughput over a single-transceiver design. The figures also demonstrate that the throughput gain due to two transceivers is larger at smaller data-packet sizes (i.e., larger δ) and larger M . This is because a larger δ (or M ) means larger a AW duration, which results in more overhead for the single-transceiver solution.
4.6 Performance Evaluation We now evaluate the performance of the proposed protocols via simulations. Our proposed protocols (AW-MAC and WFC-MAC) are compared with two multi-channel MAC protocols: BMC-MAC [30] and DDMAC [52]. As mentioned before, BMC-MAC selects the best available channel for data transmission. Although BMC-MAC was not originally designed for a CRN environment, we adapt its operation to such an environment by modifying the channel selection criteria: if the transmission power associated with the best available channel satisfies the CR-to-PR interference mask, then it will be selected; otherwise no channel will be assigned, prompting the transmitter to back off. DDMAC is a CSMA-based spectrum-sharing protocol for CRNs. It attempts to maximize the CRN throughput through a probabilistic channel assignment algorithm that exploits the
25
Maximum Achievable Throughput
Maximum Achievable Throughput
113
2−radio AW−MAC 20
AW−MAC WFC/BMC−MAC
15 10 5 0
5
10 M
15
(a) Data-packet size = 4 KB
20
25 2−radio AW−MAC 20
AW−MAC WFC/BMC−MAC
15 10 5 0
5
10 M
15
20
(b) Data-packet size = 8 KB
Figure 4.6: Maximum achievable throughput (in packet/ (Tdata + M Tctrl )) vs. total number of frequency channels (control-packet size = 120 bits). dependence between the signal’s attenuation model and the transmission distance while considering current traffic and interference conditions. For a fair comparison, in BMCMAC, WFC-MAC, and DDMAC, CR users employ the same channel access mechanism described in Section 4.4.3. They differ in the channel assignment approach. Thus, the maximum achievable throughput under DDMAC is the same as the one obtained in Section 4.5 for WFC-MAC/BMC-MAC. It is worth mentioning that DDMAC involves more processing overhead, as it requires distance and traffic estimation. In our evaluation, we first study the network performance in a single-hop CRN. Then, we study it in a multi-hop mobile CRN5 . Our results are based on simulation experiments conducted using CSIM, a C-based, process-oriented, discrete-event simulation package [3].
4.6.1 Simulation Setup We consider four PRNs and one CRN that coexist in a 100 meter × 100 meter field. Users in each PRN are uniformly distributed. The PRNs operate in the 600 MHz, 900 MHz, 2.4 GHz, and 5.7 GHz bands, respectively. Each PRN consists of three 2.5-MHz-wide channels, resulting in a maximum of 12 channels for opportunistic transmissions. The 5
Our simulations account for the effect of the hidden-terminal problem by considering the
interference from active CR transmissions that use common channels (if any).
114 number of PR users in each PRN is 100. We divide the time into slots, each of length 6.6 ms. A time slot corresponds to the transmission of one data packet at a transmission rate of 5 Mbps. In any given slot, each user in the kth PRN attempts to transmit over its own band with probability pk . The probabilities for the four PRNs are 0.5, 0.3, 0.3, 0.1, respectively. The transmission power for each PR user is 0.5 Watt. For the CRN, we consider 200 mobile users. The random waypoint model is used for mobility, with the speed of a CR user uniformly distributed between 0 and 2 meters/sec. For each generated packet, the destination node is selected randomly. Each CR user generates fixed-size (4KB) data packets according to a Poisson process of rate λ (in packet/time slot). Each user requires a transmission rate of 5 Mbps. We set the CRN SINR threshold to 5 dB (i)
and the thermal noise power density to Pth = 10−21 Watt/Hz for all channels. We set (1)
(2)
(12)
the interference mask to Pmask = Pmask = . . . = Pmask = 50 mW and the control-packet size to 120 bits. The reported results are averaged over 100 runs. Our performance metrics include: (1) the network throughput, (2) the CR blocking rate, (3) the average energy consumption for successfully transmitting one data packet (Ep ), and (4) the fairness index. The CR blocking rate is defined as the percentage of CR requests that are blocked due to the unavailability of a feasible channel. We use the fairness index in [31] to quantify the fairness of a scheme according to the throughput of all the CR users in the network.
4.6.2 Single-hop Network We first study the throughput performance. Figures 4.7(a) and (b) show that 2-radio AW-MAC provides only minor improvement in the network throughput over the single transceiver AW-MAC (this result is inline with the analysis in Section 4.5). Because both 2-radio AW-MAC and AW-MAC use the same channel assignment algorithm and provide comparable throughput performance, in the following, we focus on the performance of AWMAC and compare it with the performance of the other protocols. Specifically, Figures 4.7(a) and (b) show that under moderate and high traffic loads, AW-MAC significantly outperforms the other protocols. At steady state, AW-MAC reduces the CR blocking rate and improves the overall one-hop throughput by up to 39% compared to BMC-MAC, 15.7% compared to DDMAC, and 8.8% compared to WFC-MAC. This improvement is mostly attributed to the increase in the number of simultaneous CR transmissions. WFC-
115 MAC outperforms both BMC-MAC and DDMAC. This is because WFC-MAC attempts to serve a given CR transmission first using the worst feasible channel and preserves better channels for potential future transmissions. Under light loads, all protocols give comparable throughput performance. In Figure 4.7(c), we study the impact of the channel assignment strategy on E p . It is clear that WFC-MAC and DDMAC perform the worst in terms of energy consumption. At the same time, the figure reveals that 2-radio AW-MAC, AW-MAC, and BMC-MAC have comparable performance with respect to Ep . Thus, the throughput advantage of AW-MAC does not come at the expense of additional energy consumption. Figure 4.8 shows that all schemes achieve comparable fairness. This can be attributed to the fact that in all of these schemes CR users contend over the control channel using a variant of the CSMA/CA mechanism. Finally, Figure 4.9 depicts the channel usage, defined as the fraction of time in which a specific channel is used for CR transmissions. For WFC-MAC and DDMAC, channel usage is roughly evenly distributed among all channels, irrespective of the traffic load. However, for AW-MAC and BMC-MAC, the channels with lower carrier frequencies are favored for CR transmissions (because of the lower attenuation characteristics of these channels).
4.6.3 Multi-hop Network In order to study the performance in a multi-hop environment, we use the same simulation setup described in Section 4.6.1, but with the following changes: • A 500 meter × 500 meter field is considered for the 200 mobile CR users. (1)
(2)
(12)
• The interference mask is set to Pmask = Pmask = . . . = Pmask = 100 mW. • The number of PR users in each PRN is 300. PR users are uniformly distributed over the field. The purpose behind these changes in the setup is to give rise to hidden terminals. Because BMC-MAC, WFC-MAC, and DDMAC use the same channel access mechanism and the same maximum transmission range, it is safe to assume that they also achieve the same forward progress per hop. Consequently, we can focus on the one-hop
116 throughput, i.e., the packet destination is restricted to one hop from the source. Note that hidden-terminal problems still exist and they will clearly impact the network performance. As shown in Figures 4.10(a) and 4.10(b), WFC-MAC achieves lower CR blocking rate and higher network throughput than the other two protocols under moderate and high traffic loads. On the other hand, under low traffic load, all protocols achieve comparable throughput performance.
Figure 4.10(c) shows that BMC-MAC outperforms WFC-
MAC and DDMAC in terms of Ep under different traffic loads. Similar fairness and channel usage properties to the single-hop scenarios are observed here. Note that no single strategy is always best in all traffic regimes. Under light traffic, BMC-MAC provides the same throughput performance as WFC-MAC and DDMAC, but outperforms them in terms of Ep . However, under moderate and high traffic loads, WFCMAC performs better in terms of throughput at the cost of Ep .
4.7 Related Work One of the key challenges to enabling CR communications is how to perform opportunistic medium access control while limiting the interference imposed on PR users. Recently, several attempts were made to develop MAC protocols for CRNs (e.g., [36, 37, 59, 52, 55, 51, 13, 70, 16, 38, 68]). Existing work on spectrum sharing/access protocols can be classified according to their architecture (centralized or decentralized), spectrum allocation behavior (cooperative or non-cooperative), and spectrum access technique (overlay or underlay) [9]. The IEEE 802.22 working group is in the process of standardizing a centralized MAC protocol that enables spectrum reuse by CR users operating on the TV broadcast bands[1]. In [13, 70, 16] centralized protocols were proposed for coordinating spectrum access. For an ad hoc CRN without centralized control, it is desirable to have a distributed MAC protocol that allows every CR user to individually access the spectrum. DC-MAC [46] is a cross-layer distributed scheme for spectrum allocation/sensing. It provides an optimization framework based on partially observable Markov decision processes, with no insights into protocol design, implementation and performance. In [45], the authors proposed a decentralized channel-sharing mechanism for CRNs based on a gametheoretic approach under both cooperative and non-cooperative scenarios. However, they did not propose an operational MAC protocol. No guarantee on the performance of PRNs
117 was considered. The FCC defined the interference temperature model [26], which provides a metric for measuring the interference experienced by PR users. Clancy [19] used this model to select an optimal bandwidth/power assignment for CR users. However, no operational protocol was proposed. It is worth mentioning that due to the lack of specific technical rules to implement the interference temperature model, the FCC has abandoned this model in 2007 [7]. In [24], the authors developed a power control approach for CR systems based on spectrum sensing side information. The objective of such an approach is to mitigate the interference to a PR user from CR transmissions. However, no operational protocol was presented. The setup was also limited in that a CR user cannot transmit over a partially occupied licensed channel. Three spectrum sharing techniques were proposed and compared in [40]: spreading-based underlay, interference avoidance overlay, and spreadingbased underlay with interference avoidance. The metric of interest in the comparison was pout . The treatment did not provide guarantees on the performance of PR users. Furthermore, interference statistics were used assuming an unbounded region for outage probability analysis. In addition, The outage probability analysis ignored the interference caused by other PR users. Before closing, we note that a number of multi-channel contention-based MAC protocols were previously proposed in the context of CRNs (e.g., [59, 55, 51, 52]). The CRN MAC protocol in [59] jointly optimizes the multi-channel power/rate assignment, assuming a given power mask on CR transmissions. How to determine an appropriate power mask remains an open issue. DDMAC [52] is a spectrum-sharing protocol for CRNs that attempts to maximize the CRN throughput through a novel probabilistic channel assignment algorithm that exploits the dependence between the signal’s attenuation model and the transmission distance while considering the prevailing traffic and interference conditions. AS-MAC [55] is a spectrum-sharing protocol for CRNs that coexist with a GSM network. CR users select channels based on the CRN’s control exchanges and GSM broadcast information. Explicit coordination with the PRNs is required. In [68], the authors developed a spectrum aware MAC protocol for CRNs (CMAC). CMAC enables opportunistic access and sharing of the available white spaces in the TV spectrum by adaptively allocating the spectrum among contending users.
118 4.8 Conclusions In this chapter, we investigated the design of cooperative dynamic channel assignment in single-transceiver CRNs with adaptive power management. Our solution maximizes the network throughput as a primary objective, followed by minimizing energy consumption as a secondary objective. We first presented centralized (optimal) and distributed (heuristic) channel assignment algorithms that leverage the unique capabilities of CRs. For singlehop CRNs, we developed a CSMA-based MAC protocol with access window (AW) for exchanging control messages. Our AW-MAC realizes the centralized channel assignment algorithm in a distributed manner. Based on our distributed assignment, we also developed a distributed, asynchronous MAC protocol (WFC-MAC) for multi-hop CRNs. We studied the performance of our protocols and contrasted them with two previously proposed multichannel MAC protocols (i.e., BMC-MAC and DDMAC). We showed that for single-hop CRNs, AW-MAC performs the best in terms of throughput and energy consumption under various traffic conditions. Under moderate-to-high traffic loads, AW-MAC achieves about 39% increase in throughput over BMC-MAC at no additional cost in energy. It achieves about 15.7% throughput improvement over DDMAC, with even less energy consumption and processing overhead. For multi-hop scenarios, our simulation results show that WFCMAC is the best strategy in terms of throughput at the cost of energy consumption. This was observed for different traffic loads. In addition, under low traffic load, we found that BMC-MAC is a good scheme in terms of energy consumption, as both BMC-MAC and WFC-MAC have the same throughput performance in such traffic regime.
119
(a) Blocking ratio vs. λ
Thropughput (Packet/sec)
3600 3200
1%
2800
15.7%
2400 39%
2000
WFC−MAC BMC−MAC DDMAC AW−MAC 2−radio AW−MAC
1600 1200 800 400 0 0
10
20 λ (Packet/sec)
30
(b) Throughput vs. λ 0.128
WFC−MAC BMC−MAC DDMAC AW−MAC 2−radio AW−MAC
Ep (mJ)
0.096
0.064
0.032
0 4
8
12
16 20 24 λ (Packet/sec)
28
32
36
(c) Ep vs. λ
Figure 4.7: CRN performance in single-hop scenarios.
120
Fairness Index
1
AW−MAC WFC−MAC BMC−MAC DDMAC
0.9
0.8
0.7 0
10
20 λ (Packet/sec)
30
Figure 4.8: Fairness index in single-hop scenarios (2-radio AW-MAC depicted similar behavior as AW-MAC).
0.15
Low Load
0.1 Channel Usage (%)
0.05 0 0.15
1
2
3
4
5
6
7
8
9 10 11 12
Moderate Load
0.1 0.05 0 0.15 0.1
1
2
3
4
5
6 7 8 9 10 11 12 High Load BMC−MAC WFC−MAC AW−MAC DDMAC
1
2
3
4
5
0.05 0
6 7 Ch. No.
8
9 10 11 12
Figure 4.9: CR channel usage in single-hop scenarios (2-radio AW-MAC depicted similar behavior as AW-MAC).
121
CR Blocking Rate
0.55
WMC−MAC BMC−MAC DDMAC
0.5
0.45
0.4
0.35
0.05
0.1 0.15 0.2 λ (Packet/time slot)
0.25
(a) Blocking ratio vs. λ
Througput (Packet/ time slot)
2.8 8%
2.6
15.5%
2.4 2.2 2
WMC−MAC BMC−MAC DDMAC
1.8 1.6 1.4
0.05
0.1 0.15 0.2 λ (Packet/time slot)
0.25
(b) Throughput vs. λ
2
−4
x 10
WFC−MAC BMC−MAC DDMAC
E
p
(mJ)
1.8 1.6 1.4 1.2 1 0.05
0.1 0.15 0.2 λ (Packet/time slot)
0.25
(c) Ep vs. λ
Figure 4.10: CRN performance in multi-hop scenarios.
122
CHAPTER 5 Conclusions
In Chapter 2, we proposed a MAC protocol for opportunistic CRNs. Our protocol, COMAC, improves spectrum utilization while limiting the interference imposed on licensed users. We first developed stochastic models for the PR-to-PR and the PR-to-CR interference under a Rayleigh fading channel model, and derived closed-form expressions for the mean and variance of each interference component. Furthermore, closed-form expressions were obtained for the characteristic function of the total interference under typical path loss exponents. From the interference analysis, we derived a closed-form expression for the maximum allowable powers for CR transmissions that ensure a statistical bound β on p out for PR users. We integrated our theoretical analysis in the design of the COMAC protocol. Our simulation results show that COMAC statistically guarantees the performance of PR users under different CR traffic loads and for different values of β. Results also show that channel usage is reasonably balanced across various channels, even when the PR activity factors over such channels and the associated carrier frequencies are significantly different. Although uniform node deployment was used in our analysis, our simulations verified that the performance is not significantly impacted by the distributions of users in PRN/CRN. Finally, our simulation results showed that exploiting the available channel information for the routing decisions can improve the end-to-end throughput of the CRN by up to 25% In Chapter 3, we presented a novel opportunistic distance-dependent MAC protocol for CRNs (referred to as DDMAC). DDMAC improves the CRN throughput through cooperative channel assignment, taking into consideration the non-adjacency of frequency channels and the imposed power masks. We presented a heuristic stochastic channel assignment scheme that dynamically exploits the dependence between the signal attenuation model and the transmission distance. Our scheme accounts for traffic dynamics. It assigns channels with lower average SINR to shorter transmission distances to increase the number of simultaneous transmissions. We integrated the channel assignment process in
123 the design of DDMAC. We compared the performance of DDMAC with that of a reference multi-channel MAC protocol that is designed for typical multi-channel systems (BMC). We showed that, under moderate and high traffic loads, DDMAC achieves about 30% increase in throughput over the BMC scheme, with manageable processing overhead. Although DDMAC requires a pair of CR users to communicate on a channel that may not be optimal from a user’s perspective, we showed that the average per-user throughput of DDMAC under moderate and high traffic loads is greater than that of the BMC scheme. Furthermore, DDMAC preserves (even slightly improves) throughput fairness relative to BMC. In summary, DDMAC provides better spectrum utilization by reducing the connection blocking probability and increasing the system throughput. To the best of our knowledge, DDMAC is the first CRN MAC protocol that utilizes the radio propagation characteristics to improve the overall network throughput. In Chapter 4, we investigated the design of cooperative dynamic channel assignment in single-transceiver CRNs with adaptive power management. Our solution maximizes the network throughput as a primary objective, followed by minimizing energy consumption as a secondary objective. We first presented centralized (optimal) and distributed (heuristic) channel assignment algorithms that leverage the unique capabilities of CRs. For singlehop CRNs, we developed a CSMA-based MAC protocol with access window (AW) for exchanging control messages. Our AW-MAC realizes the centralized channel assignment algorithm in a distributed manner. Based on our distributed assignment, we also developed a distributed, asynchronous MAC protocol (WFC-MAC) for multi-hop CRNs. We studied the performance of our protocols and contrasted them with two previously proposed multichannel MAC protocols (i.e., BMC-MAC and DDMAC). We showed that for single-hop CRNs, AW-MAC performs the best in terms of throughput and energy consumption under various traffic conditions. Under moderate-to-high traffic loads, AW-MAC achieves about 39% increase in throughput over BMC-MAC at no additional cost in energy. It achieves about 15.7% throughput improvement over DDMAC, with even less energy consumption and processing overhead. For multi-hop scenarios, our simulation results show that WFCMAC is the best strategy in terms of throughput at the cost of energy consumption. This was observed for different traffic loads. In addition, under low traffic load, we found that BMC-MAC is a good scheme in terms of energy consumption, as both BMC-MAC and WFC-MAC have the same throughput performance in such traffic regime.
124 REFERENCES
[1] IEEE 802.22 Working Group on Wireless Regional Area Networks. http://www.ieee802.org/22/. [2] Kenwood TH-D7A dual-band handheld http://www.kenwoodusa.com/Communications/AmateurRadio/Portables/TH-D7A(G).
transceiver.
[3] Mesquite Software Incorporation. www.mesquite.com. [4] Qualcomm Announces Sampling of the Industry’s First Single-Chip Receive Diversity Device for Increased CDMA2000 Network Capacity. http://www.qualcomm.com/press/releases/2005/050504-rfr6500.html. [5] The Cisco Aironet 350 Series of wireless http://www.cisco.com/warp/public/cc/pd/witc/ao350ap.
LAN.
[6] FCC, spectrum policy task force report, ET docket no. 02-155, Nov. 2002. [7] Interference Temperature Operation FCC, ET Docket No. 03-237; FCC 07-78, 2007. [8] A. Acharya, A. Misra, and S. Bansal. MACA-P : a MAC for concurrent transmissions in multi-hop wireless networks. In Proceedings of the First IEEE PerCom 2003 Conference, pages 505–508, Mar. 2003. [9] I. Akyildiz, W.-Y. Lee, M. Vuran, and S. Mohanty. Next generation dynamic spectrum access cognitive radio wireless networks: A survey. Computer Networks, 50(13):2127–2159, 2006. [10] A. (AN-SR-7). Building a versatile low latency cognitive radio for multi-mission applications with the ICS-572. Technical report, Sep 2003. http://www.gefanucembedded.com/news-events/whitepapers/3067. [11] E. Arikan. Some complexity results about packet radio networks. IEEE Transactions on Information Theory, 30(4):681–685, 1984. [12] A. Behzad and I. Rubin. Multiple access protocol for power-controlled wireless access nets. IEEE Transactions on Mobile Computing, 3(4):307–316, 2004. [13] V. Brik, E. Rozner, S. Banarjee, and P. Bahl. DSAP: A protocol for coordinated spectrum access. In Proceedings of the IEEE DySPAN Conference, pages 611– 614, Nov. 2005.
125 [14] D. Cabric and R. Brodersen. Physical layer design issues unique to cognitive radio systems. In Proceedings of the IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Sept. 2005. [15] D. Cabric, S. Mishra, and R. Brodersen. Implementation issues in spectrum sensing for cognitive radios. In Proceedings of the 38th Asilomar Conference on Signals, Systems and Computers, pages 772–776, Nov. 2004. [16] D. Cabric, S. Mishra, D. Willkomm, R. Brodersen, and A. Wolisz. A cognitive radio approach for usage of virtual unlicensed spectrum. In Proceedings of the 14th IST Mobile and Wireless Communications Summit, June 2005. [17] Y. Chang, T. Ho, and L. Kaelbling. Mobilized ad-hoc networks: A reinforcement learning approach. In Proceedings of the International Conference on Autonomic Computing, pages 240–247, May 2004. [18] D. K. C.-C. Chia-Chun Hsu, A. Weit. A cognitive MAC protocol using statistical channel allocation for wireless ad-hoc networks. In proceeding of Wireless Communications and Networking Conference (WCNC07), pages 105–110, March 2007. [19] T. C. Clancy. Achievable capacity under the interference temperature model. In Proceedings of the IEEE INFOCOM Conference, pages 794–802, May 2007. [20] C. Cordeiro, K. Challapali, D. Birru, and S. Shankar. IEEE 802.22: The first worldwide wireless standard based on cognitive radios. In Proceedings of the IEEE DySPAN Conference, November 2005. [21] V. Dendeit and H. Emmons. Max-Min matching problems with multiple assignments. Journal of Optimization Theory and Application, 91(2):491–511, 1996. [22] T. Fujii and Y. Suzuki. Ad-hoc cognitive radio: Development of frequency sharing system by using multi-hop netowrk. In Proceedings of the IEEE DySPAN Conference, November 2005. [23] R. Garces and J. Garcia-Luna-Aceves. Collision avoidance and resolution multiple access for multichannel wireless networks. In Proceedings of the IEEE INFOCOM Conference, pages 595–602, March 2000. [24] K. Hamdi, W. Zhang, and K. B. Letaief. Power control in cognitive radio systems based on spectrum sensing side information. In Proceedings of the IEEE ICC Conference, pages 5161–5165, June 2007.
126 [25] H. Harada. Software defined radio prototype toward cognitive radio communication systems. In Proceedings of the IEEE DySPAN Conference, November 2005. [26] S. Haykin. Cognitive radio: Brain-empowered wireless communications. IEEE Journal on Selected Areas in Communications, 23(2):201–220, Feb. 2005. [27] S. Hijazi, B. Natarajan, M. Michelini, Z. Wu, and C. Nassar. Flexible spectrum use and better coexistence at the physical layer of future wireless systems via a multicarrier platform. IEEE Wireless Communications, 11(2):8–14, April 2004. [28] A. Hoang and Y. Liang. Maximizing spectrum utilization of cognitive radio networks using channel allocation and power control. In Proceedings of Vehicular Technology Conference (VTC), pages 1–5, Sept. 2006. [29] S. Huang, X. Liu, and Z. Ding. Opportunistic spectrum access in cognitive radio networks. In Proceedings of the IEEE INFOCOM Conference, pages 1427–1435, April 2008. [30] N. Jain, S. Das, and A. Nasipuri. A multichannel CSMA MAC protocol with receiver-based channel selection for multihop wireless networks. In Proceedings of the 9th Int. Conf. on Computer Communications and Networks (IC3N), pages 432–439, Oct. 2001. [31] R. Jain. The Art of Computer System Performance Analysis. New York: John Wiley & Sons, 1991. [32] J. Jia, Q. Zhang, and X. Shen. HC-MAC: A Hardware-Constrained Cognitive MAC for Efficient Spectrum Management. IEEE Journal on Selected Areas in Communications, 26(1):106–117, January 2008. [33] S. D. Jones, N. Merheb, and I.-J. Wang. A cognitive MAC protocol using statistical channel allocation for wireless ad-hoc networks. In Wireless Communications and Networking Conference (WCNC), March 2007. [34] H. Khalife, S. Ahuja, N. Malouch, and M. Krunz. Probabilistic path selection in opportunistic cognitive radio networks. In Proceedings of the IEEE GLOBECOM Conference, Nov. 2008. [35] S. Kiran and R. Chandramouli. An adaptive energy-efficient link layer protocol using stochastic learning control. In Proceedings of the IEEE ICC Conference, pages 1114–1118, May 2003. [36] S. Krishnamurthy, M. Thoppian, S. Venkatesan, and R. Prakash. Control channel based MAC-layer configuration, routing and situation awareness for
127 cognitive radio networks. In Proceedings of the IEEE Military Communications Conference (MILCOM), October 2005. [37] P. Kyasanur and N. H. Vaidya. Protocol design challenges for multi-hop dynamic spectrum access networks. In Proceedings of the IEEE DySPAN Conference, November 2005. [38] L. Ma, X. Han, and C.-C. Shen. Dynamic open spectrum sharing MAC protocol for wireless ad hoc networks. In Proceedings of the IEEE DySPAN Conference, pages 203–213, Nov. 2005. [39] D. Maldonado, B. Le, A. Hugine, T. W. Rondeau, and C. W. Bostian. Cognitive radio applications to dynamic spectrum allocation. In Proceedings of the IEEE DySPAN Conference, November 2005. [40] R. Menon, R. Buehrer, and J. Reed. Outage probability based comparison of underlay and overlay spectrum sharing techniques. In Proceedings of the IEEE DySPAN Conference, pages 101–109, Nov. 2005. [41] R. Mukhopadhyay, Y. Park, P. Sen, N. Srirattana, J. Lee, C. Lee, S. Nuttinck, A. Joseph, J. D. Cressler, and J. Laskar. Reconfigurable RFICs in silicon-based technologies for a compact intelligent RF front-end. IEEE Transactions on Microwave Theory and Techniques, 53(1):81–93, January 2005. [42] A. Muqattash and M. Krunz. POWMAC: A single-channel power control protocol for throughput enhancement in wireless ad hoc networks. IEEE Journal on Selected Areas in Communications, 23(5):1067–1084, May 2005. [43] A. Nasipuri and S. Das. Multichannel CSMA with signal power-based channel selection for multihop wireless networks. In Proceedings of the IEEE Vehicular Technology Conference (VTC), pages 24–28, Sept. 2000. [44] A. Nasipuri and S. Das. Performance of multi-channel ad hoc networks. International Journal of Wireless and Mobile Computing, 1(3/4):191–203, 2006. [45] N. Nie and C. Comaniciu. Adaptive channel allocation spectrum etiquette for cognitive radio networks. In Proceedings of the IEEE DySPAN Conference, pages 269–278, Nov. 2005. [46] L. T. Q. Zhao and A. Swami. Decentralized cognitive MAC for dynamic spectrum access. In Proceedings of the IEEE DySPAN Conference, pages 224–232, Nov. 2005. [47] T. S. Rappaport. Wireless Communications-Principles and Practice. PrenticeHall Press, 2001. 2nd edition.
128 [48] M. Riback, J. Medbo, J. Berg, F. Harrysson, and H. Asplund. Carrier frequency effects on path loss. In Proceedings of Vehicular Technology Conference (VTC), pages 2717–2721, May 2006. [49] Rutgers Winlab. Cognitive radio platform. http://www.winlab.rutgers.edu/pub/docs/focus/Cognitive-Hw.html. [50] A. Sabharwal, A. Khoshnevis, and E. Knightly. Opportunistic spectral usage: Bounds and a multi-band CSMA/CA protocol. IEEE/ACM Transactions on Networking, 15(3):533–545, June 2007. [51] A. Sabharwal, A. Khoshnevis, and E. Knightly. Opportunistic spectral usage: Bounds and a multi-band CSMA/CA protocol. IEEE/ACM Transactions on Networking, 15(3):533–545, 2007. [52] H. B. Salameh, M. Krunz, and O. Younis. Distance- and traffic-aware channel assignment in cognitive radio networks. In Proceedings of the IEEE SECON Conference, June 2008. [53] H. B. Salameh, M. Krunz, and O. Younis. MAC protocol for opportunistic cognitive radio networks with soft guarantees. IEEE Transactions on Mobile Computing, 2009. To appear (http://www.ece.arizona.edu/∼krunz/Publications.htm/). [54] H. B. Salameh, T. Shu, and M. Krunz. Adaptive cross-layer MAC design for improved energy efficiency in multi-channel wireless sensor networks. Ad Hoc Networks, 5(6):844–854, 2007. [55] S. Sankaranarayanan, P. Papadimitratos, A. Mishra, and S. Hershey. A bandwidth sharing approach to improve licensed spectrum utilization. In Proceedings of the IEEE DySPAN Conference, pages 279–288, Nov. 2005. [56] R. Sedgewick. Algorithms in C, Part 5: Graph Algorithms. Addison-Welsy, 2002. 3rd edition. [57] S. M. Selby. Basics Mathematical Tables. The Chemical Rubber Co. 17th edition. [58] S. Shankar. Squeezing the most out of cognitive radio: A joint MAC/PHY perspective. In proceeding of Acoustics, Speech and Signal Processing (ASSP07), pages 1361–1364, April 2007. [59] T. Shu, S. Cui, and M. Krunz. Medium access control for multi-channel parallel transmission in cognitive radio networks. In Proceedings of the IEEE GLOBECOM Conference, Nov. 2006.
129 [60] J. So and N. Vaidya. Multi-channel MAC for ad hoc networks: Handling multi-channel hidden terminals using a single transceiver. In Proc. of the ACM International Symposium on Mobile and Ad-Hoc Networking and Computing (MobiHoc), pages 222 – 233, Tokyo, Japan, May 2004. [61] E. Sousa. Performance of a spread spectrum packet radio network link in a poisson field of interferers. IEEE Trans. Information Theory, 38:1743–1754, Nov. 1992. [62] E. Sousa and J. Silvester. Optimum transmission ranges in a direct-sequence spread-spectrum multihop packet radio network. IEEE Journal on Selected Areas in Communications, 8(5):762–771, June 1990. [63] C. Steger, P. Radosavljevic, and J. P. Frantz. Performance of IEEE 802.11b wireless LAN in an emulated mobile channel. In Proceedings of IEEE Fall Vehicular Technology Conference (VTC), April 2003. [64] Y. Xing, R. Chandramouli, S. Mangold, and S. Shankar. Dynamic spectrum access in open spectrum wireless networks. IEEE Journal on Selected Areas in Communications, 24(3):626– 637, 2006. [65] Y. Xing, C. N. Mathur, M. A. Haleem, R. Chandramouli, and K. P. Subbalakshmi. Dynamic spectrum access with QoS and interference temperature constraints. IEEE Transactions on Mobile Computing, 6(4):423–433, 2007. [66] M. Youssef and A. Agrawala. The Horus WLAN location determination system. In MobiSys’05: Proceedings of the 3rd international conference on Mobile systems, applications, and services, pages 205–218, 2005. [67] C. Yu, K. G. Shin, and L. Song. Link-layer salvaging for making routing progress in mobile ad hoc networks. In Proc. of the ACM International Symposium on Mobile and Ad-Hoc Networking and Computing (MobiHoc), pages 242–254, May 2005. [68] Y. Yuan, P. Bahl, R. Chandra, P. Chou, J. Ferrell, T. Moscibroda, S. Narlanka, and Y. Wu. Knows: Kognitive networking over white spaces. In Proceedings of the IEEE DySPAN Conference, pages 416–427, April 2007. [69] Y. Yuan, P. Bahl, R. Chandra, T. Moscibroda, and Y. Wu. Allocating dynamic time-spectrum blocks in cognitive radio networks. In Proc. of the ACM International Symposium on Mobile and Ad-Hoc Networking and Computing (MobiHoc), Sept. 2007. [70] S. Zekavat and X. Li. User-central wireless system: Ultimate dynamic channel allocation. In Proceedings of the IEEE DySPAN Conference, pages 82–87, Nov. 2005.
130 [71] J. Zhao, H. Zheng, and G.-H. Yang. Distributed coordination in dynamic spectrum allocation networks. In Proceedings of the IEEE DySPAN Conference, pages 259–268, Nov. 2005.
