Virginia Tech, Center for Wireless Telecommunications (CWT), Blacksburg, VA 24061. Abstract. In an effort ... we see immediate benefits as well as our cognitive.
COGNITIVE RADIO APPLICATIONS TO DYNAMIC SPECTRUM ALLOCATION: A DISCUSSION AND AN ILLUSTRATIVE EXAMPLE David Maldonado, Bin Le, Akilah Hugine, Thomas W. Rondeau, Charles W. Bostian (Advisor: Dr. Charles W. Bostian) Virginia Tech, Center for Wireless Telecommunications (CWT), Blacksburg, VA 24061 Abstract In an effort to improve radio spectrum management and promote a more efficient use of it, regulatory bodies are currently trying to adopt a new spectrum access model. At the same time, cognitive radio technology has received a lot of interest as a possible enabling technology. In this paper, we provide a brief description of the broad impact of cognitive radios in different markets. At Virginia Tech’s Center for Wireless Telecommunications (CWT), we have designed a biologically inspired cognitive engine with dynamic spectrum access (DSA) as one of its intended applications. An experimental software simulation shows a 20 dB SINR improvement using cognitive techniques in an interference environment over that provided by current IEEE 802.11a service PHY standard.
Introduction The vision of cognitive radio has generated a lot of discussion and increasing attention from the wireless communications community in recent years. With the ability to learn from and adapt to both their surrounding environment and user needs, cognitive radios offer a great number of benefits in almost all markets of interest: military, government, public safety, and commercial. As a result of our work at Virginia Tech’s Center for Wireless Telecommunications, we have designed and are in the process of implementing a generalized cognitive engine/radio applicable to all these areas. In this paper, we will provide an overview of cognitive radio technology and describe the application areas where we see immediate benefits as well as our cognitive engine approach. We will then provide simulation results illustrating one of the most obvious and immediately beneficial cognitive radio applications: dynamic spectrum sharing. Although there are many techniques used to share spectrum and improve capacity in wireless channels, we will experimentally demonstrate that by sensing the environment and making real-time decisions on frequency, bandwidth, and waveform, we can achieve an increase of 20 dB signal to interference and noise ratio (SINR) in wireless LANs over the standard techniques. This improvement is achieved using a very early-stage
Maldonado-Febus
cognitive radio and indicates how much better performance a fully-developed solution can obtain. Cognitive Radios: Moving Towards a Definition Intelligent communication devices or agents have existed for quite some time as futuristic ideas found primarily in science-fiction novels and movies. The reader may recall literary examples of intelligent machines like Asimov’s three laws of robotics, the AI computers in Gibson’s Neuromancer, or Marvin, the paranoid android, in Adams’ The Hitchhiker’s Guide to the Galaxy. They have certainly caught our attention not only by having the ability to communicate but by all the information that they make available to the users. As wireless communication devices move towards a more softwarebased and flexible hardware architecture (software defined radio (SDR) technology), they are becoming capable of awareness and more intelligent operation, bringing them closer to those mentioned above. The evolution has already taken communication devices from fixed radios to adaptive-aware ones, and we enjoy the benefits when we use a modern cellular telephone. Cognitive radios (CR) are the next evolution of such devices through the addition of a layer of intelligence that provides the ability to better satisfy user and network needs. The cognitive radio initiatives have gained a lot of momentum after Joseph Mitola revived the concept of an intelligent communication device and presented a general vision of what such a device could deliver [1]. Cognitive radio technology uses real-time knowledge of its environment to adapt its behavior dynamically with the intent to enhance its operation. The knowledge of the operational environment is referred to as situation awareness and will include, but is not limited to, information about the physical environment, RF channel, radio resources, and user/application requirements. Although neither a formal universal definition of CR nor a clear understanding about what level of sophisticated intelligence is needed to categorize a radio as cognitive exist, there is a consensus that a cognitive radio will include an informed decision making process that functions without any human intervention. Cognitive Radios Application Areas Government and Regulatory Bodies Interest
1
With the promises of an intelligent and aware device, a wide range of applications have emerged, from Dynamic Spectrum Access (DSA) to interoperability solutions to the idea of a universal portable communicator; all of these target markets ranging from the military to the commercial ones. DSA is currently being considered as the prime candidate for the first practical application of cognitive radio technology. The impact and possible importance of this idea is felt throughout the United States agencies responsible for spectrum management. The Federal Communications Commission (FCC), the National Telecommunications and Information Administration (NTIA) and the Department of State have expressed interest in what CR technology has to offer and how it would affect their current regulatory scheme. In particular, the FCC has launched a set of initiatives to facilitate the development and deployment of this technology as listed in Table 1. One of their most recent actions will allow the use of cognitive radios/cognitive applications to be incorporated into certain TV bands [2]. The CR impact also extends beyond our geographical border as other countries and international agencies such as the International Telecommunications Union (ITU) are looking to adopt a similar cognitive radio approach to increase spectrum utilization. We acknowledge that although dynamic spectrum access looks to be very promising, the complexity required to achieve it could be overwhelmingly difficult. TABLE I
TABLE I FCC COGNITIVE RADIO AND RELATED DOCUMENTS Topic Receiver Standards Interference Temperature Cognitive Radio License-exempt Operation in the TV Broadcast Bands Additional Spectrum for Licenseexempt devices below 900 MHz and in the 3 GHz Band Cognitive Radio Report & Order
Document Number ET Docket No. 03-65 ET Docket No. 03-237 ET Docket No. 03-108 ET Docket No. 04-186 ET Docket No. 02-380
ET Docket No. 05-57
Military The military community has recognized the benefits that this new radio technology offers. With frequency agility and/or flexibility, the ability to enhance interoperability between different radio standards, and the capability to sense the presence of interferers, CR has become a must-have technology. The military obviously has a great interest in
Maldonado-Febus
remaining unseen by the enemy and in protecting communication transmissions. Because a cognitive radio is aware of its environment, it can both recognize enemy communication devices and discover paths of opportunity to transmit its own information in a clear, uninterrupted, and unseen way. Cognitive radios will also have a major role to play in interoperability of military systems. By recognizing other communication devices, the cognitive radio can adjust itself to communicate with those devices. This will allow many different legacy radios to interoperate. The U.S. Department of Defense (DoD) has devoted a great amount of effort to CR in recent years and has established programs such as SPEAKeasy radio system, Joint Tactical Radio System (JTRS), and neXt Generation (XG) to further explore the possibilities of the creation of an intelligent communication agent. The XG program was established to create a new generation of spectrum access technology [3]. Although XG’s primary focus was dynamic spectrum access, it recognized the need for more intelligent and autonomous devices such as cognitive radios to achieve the degree of needed performance. Besides XG, other DoD programs like UNMAN are exploring the use of CRs as a primary communication device. Public Safety Public Safety and emergency response is another area in which cognitive radio has gained a lot of attention. For years public safety agencies have desperately needed additional spectrum allocation to ease frequency congestion and enhance interoperability. These particular problems can be mitigated through the use of cognitive radio technology. For emergency and public-service providers, a major part of this concept is spectrum sharing, which can help in maintaining call priority and response time. The National Institute of Justice issued a call for proposals in which they seek to find a technology that can not only provide them with a solution to the interoperability issue they currently face but with an ubiquitous system able to handle communication needs yet to come [4]. Cognitive radios can play an important part in improving interoperability by enabling devices to bridge communications between jurisdictions using different frequencies and modulation formats. For example, the technology will allow a police department from Maryland to communicate effectively with a police department from Washington D.C. even if they use two different radio systems. Cognitive radio technology has caught the attention of the U.S. Department of Justice, which has employed the National Public Safety Telecommunications Council (NPSTC) to help in the effort of researching cognitive
2
radio to aid with public safety communication issues [5]. The NPSTC effort focuses on finding ways to expand spectrum allocation and reuse, which is eminent in their research on promoting the availability of 700 MHz, 800 MHz, and 4.9 GHz, spectrum for public safety and Homeland Security nationwide. Cognitive radios can prove to be more effective by utilizing some of the existing spectrum that is not widely used. Our group at Virginia Tech is developing a CR solution in which the cognitive radio will be able to scan the spectrum and model the public safety communications environment including the types of users and their location for facilitating communications and interoperability.
Broader Impacts and Commercial Use Spectrum sharing, as part of the DSA, is one of those applications which cognitive radio technology offers to render a great benefit. One of the biggest challenges of the implementation of CR technology lies within the policies and regulations regime. The accuracy or predictability of the radio’s operation and possible available spectrum poses a great deal of concern. Even more, any CR spectrum access technique will have to understand the difference between scarcity and allocation; two separate but related issues. CR offers the opportunity to take the conventional static approach to frequency assignment and turn it into a dynamic one. As the number of mobile devices and the user’s demand for spectrum increases in any specific system, the more difficult it becomes to satisfy everyone’s needs. Detecting the presence of a device and identifying its RF footprint and contribution to the overall interference will provide the necessary information required to share spectrum intelligently through dynamic allocation and radio reconfiguration. This issue becomes more critical if the devices were granted access to a broader spectrum. Cognitive radio requires a rethinking of traditional practices of spectrum allocation, particularly the idea that the only way to avoid interference is to assign each user exclusive use of a radio channel. Interference is not an inherent property of spectrum—it is a property of the devices that make up the radio environment. Whether a new radio system interferes with an existing one depends entirely on the technical characteristics of both
Maldonado-Febus
systems. Consequently, the extent to which there appears to be a spectrum shortage largely depends not on how many frequencies are available but on the technologies that can be deployed. There are two main schools of thought about exploiting the spectrum: service-oriented and technology-oriented. Service-oriented spectrum sharing utilizes current technologies like CDMA, WiFi and possible UWB to form a policy-agile local access network; while technology-oriented spectrum sharing focuses on developing a whole new infrastructure using adaptation algorithms and advanced radios (e.g. cognitive radio). With the proliferation of wireless technologies in the ISM band, especially after the success of wireless local area networks (WLAN) like 802.11, interference is becoming increasingly problematic. In urban environments, the ISM band is already showing the symptoms of spectrum scarcity as the demand for its use continues to increase and performance degradation becomes the norm. Although some technologies are currently using some adaptive techniques (802.11g uses channel identification, dynamic frequency selection, and adaptive modulation) to obtain higher data throughput, they are still governed by a standard that limits their full potential. By taking the CR approach into a challenging RF environment like the ISM band, where inherently the devices need to accept any interference while reducing the possibility of interfering with others, it provides a framework for us to quantify current system performance and look at the improvements that the CRs can provide. It is within bands that are heavily utilized that we see the greater need for spectrum efficiency improvement and where the promises offered by CR technology could render its greater benefits. Leveraging on the success of wireless technologies such as 802.11 and new advances in emerging ones like 802.16 and .22 could help translate current CR research directly into commercial benefits. This could enhance the possibility of the provision for commercial off-the-shelf products for both military and public safety use. The realization of CR is still in the development stages but not far from reality because it leverages the flexible SDR architecture, which is a heavily researched topic [6]. SDRs offer a flexible reconfigurable platform needed for CR implementation, but they currently face issues of flexibility, speed, power consumption, size, and price, just to mention a few, which limits their availability. Cognitive radio technology is an important innovation for the future of communications and likely to be a part of the new wireless standards, becoming almost a necessity for situations with large traffic and interoperability concerns. Dynamic channel allocation (DCA), also called dynamic spectrum allocation (DSA) [7], is one such technology-
3
oriented sharing technique for hybrid networks. In [7], the results show that even a simplified algorithm can produce gains for the radio networks in the dynamic allocation scheme.
are well suited to solving multi-objective optimization and decision problems, which is why we have chosen to work with a Multi-Objective Genetic Algorithm (MOGA) in order to control the radio’s adaptive process.
Dynamic spectrum sharing between systems using different technologies is attracting increasing interest. Traditional static frequency planning is not spectrum efficient when the network becomes highly heterogeneous.
The wireless system genetic algorithm (WSGA) is a MOGA designed for the control of a radio by modeling the physical radio system as a biological organism and optimizing its performance through genetic and evolutionary processes. In the WSGA, radio behavior is interpreted as a set of PHY and MAC layer operation parameters defined by traits encapsulated in the genes of a chromosome. Other general radio functional parameters (such as payload size, antenna configuration, voice coding, encryption, equalization, retransmission requests, and spreading technique/code) are also identified as possible genes in the chromosome definition to allow for future growth through each layer of the radio communication’s stack. The chromosome shown in Figure 1 represents the PHY-layer traits currently of consequence to the WSGA due to current hardware limitations and the current state of the simulation.
An Intelligent Interface: CWT Cognitive Engine In spectrum occupancy, we can all dream of a day when there are no restrictions, physical or regulatory, on what frequency we can use to communicate, from DC to 100 GHz. While this dream seems to provide nearly unlimited spectrum and capacity, we have two major problems: first, the physical realization of these devices is still impractical, and second, we must have ways to efficiently and reliably use the spectrum provided. Spectrum usage is currently managed through static or predefined techniques. Even spread spectrum techniques offer great improvement in capacity, but have their limits, and frequency agile radios, while effective, usually have limited capabilities (i.e., a limited number of channels and predefined bandwidth). Cognitive radio technology provides radios with the ability to alter many of its parameters and make intelligent decisions regarding the adaptation to take. Instead of predefined channels and bandwidths, cognitive radios can choose their own center frequency and alter the bandwidth according to the available spectrum and network needs. Through this intelligent decision making process, a cognitive radio provides the solution to the regulatory issue mentioned above: they will intelligently cooperate to make the best use of spectrum both in the presence of other cognitive radios and legacy, non-cognitive, radios. In order to communicate successfully, the radio must first be configured to fit the specific channel condition, such as a cellular fading channel or an interference-prone unlicensed channel; second, the radio must support user required service types, like voice or data; third, sitting on top on everything the radio does are the regulatory requirements the radio must obey to operate legally in any band and geographic location. To combine all these issues effectively and provide the best performance tradeoff, the radio needs to be aware of its environment; in other words, the radio needs a cognitive engine to analyze the physical link, user demands, and regulatory regimes, and it must balance multiple objectives and constraints. Genetic algorithms (GA)
Maldonado-Febus
The WSGA analyzes the chromosome’s fitness through a set of fitness functions defined by performance evaluations of the current radio channel. Each fitness function is weighted to represent the relative importance the user has associated with each objective. The Pareto front therefore moves so that the optimal solution provides the most efficient performance for the user’s QoS requirements under combinatorial constraints. Here, efficiency and optimization mean providing a QoS without over-maximizing, which may waste radio resources such as spectrum and power. For example, a user sending email does not need a 100 Mbps link with a 30 dB carrier to noise ratio. The fitness evaluation functions are designed to reflect the current link quality of both PHY and MAC layer, which currently include the mean transmitting power, data rate, BER, packet error rate (PER), spectral efficiency, bandwidth, interference avoidance, packet latency and packet jitter. One of the most powerful attributes of the WSGA is the dynamic fitness definition and evaluation, where not only is the weighting of each function adjustable, but any fitness function may or may not be used as required by the current link conditions and user requirements. All functions are dynamically linked from a database so that they can be dynamically added and weighted into the fitness evaluation for a specific link condition and performance objectives. Such dynamic adjustment of the fitness evaluation is directed by some higher-layer intelligence such as the learning machine in the cognitive engine which conducts the evaluation of the overall radio system and network performance. (for more specifics on CE let’s refer them to Tom’s SDR_04)
4
In general, fitness functions associate to specific channel conditions. For example, the fitness function for determining a BER is channel specific. Likewise, the function weights associate to a specific user’s desire for network performance, e.g. a desire to improve the BER while sacrificing the data rate. Although these associations hold for a generalpurpose analysis of the WSGA, the user can influence the functions, such as the data rate objective, and the channel can also influence the weights for each function, such as increasing the BER weighting to compensate for a particularly poor channel. We use a relative tournament selection method similar to [8], except that the fitness of the winner from a single comparison is scaled by the weight associated with that fitness function. After all the single comparisons in all dimensions, the winning member is the one with the highest fitness, and that one survives to the next population. While this does not guarantee that all winners are the best, or nondominated, members of the population (only better relative to its combatant), it maintains species diversity within the population while still pushing towards a Pareto front. Diversity in the population allows different solutions to be tried and helps prevent the algorithm from getting stuck in a local optimum. We did not apply Horn’s [Horn] method of fighting two individuals against a subset of the population because his method, while he claims it produces better results, calls for a larger population and more fitness comparisons. This becomes computationally intensive and is undesirable in a real-time optimization system. Crossover and mutation are simple implementations of these mechanisms. Crossover is performed on a single point chosen as a uniform random number with a static probability of crossover occurring. Mutation is also a single point operation chosen from a uniform random number with a static probability of crossover occurring. Future enhancements to the WSGA call for an adaptive adjustment of crossover and mutation probabilities as well as the population size during the optimization process for higher convergence efficiency and accuracy. The ability to apply constraints to the optimization problem as shown in (1) gives us the opportunity to incorporate regulatory and physical restrictions during chromosome evolution. If a trait determined by the chromosome exceeds the limits of the radio’s capabilities, like finding a center frequency outside the tunable range of the radio, or breaks the law, like transmitting too much power in a specific band, then
Maldonado-Febus
the WSGA penalizes those chromosomes. We have chosen to use a penalty approach similar to that outlined in [Fonseca-I] by setting the fitness evaluation of that chromosome to zero, basically nullifying its chance to survive to the next generation. The final issue to realize the operation of the WSGA is the exchange of optimized chromosomes between radios wishing to communicate such that all networked radios evolve to have better traits. Provided a communications link already exists, all radios on the network running the WSGA will share the chromosomes and the most fit chromosome is elected as the winner (as though this were the last generation of the genetic algorithm). The radios will then all switch their parameters together. If no communications path is present, a control channel can be set up to allow temporary communication between the radios in order to exchange chromosomes and reconfigure themselves. We recognize that this is not the most complete or satisfactory solution for all situations, and we will need to adopt some protocol to establish the connections and exchange the adaptation information between all radios on the network. Cognitive Radio Example: Improvements on Spectrum Utilization The simulation presented in this paper provides an insight to the benefits that cognitive radio technology can achieve, even with preliminary and limited cognitive algorithms. The experiment sets a narrow focus to the unlicensed 5.8 GHz ISM (industrial, scientific, and medical) band to compare the system performance in terms of spectrum utilization of IEEE 802.11a/g physical layers [8] [9] to a cognitive radio version of such a WLAN radio. In this experiment, we created a standard OFDM PHY layer of 802.11a/g WLAN and then simulated the spectrum utilization with both a static channel assignment policy and with dynamic spectrum allocation using cognitive algorithms. Simulation Design An 802.11a/g OFDM WLAN PHY-layer communications system is established to simulate the performance improvement in terms of link QoS and efficiency of spectrum resource management when a layer of spectrum cognition is applied over current static (non-cognitive) PHY-layer standard. System infrastructure The designed 802.11a/g WLAN system has an ad-hoc network topology as illustrated in Figure 1. Since only spectrum efficiency is be investigated to show the benefits of cognitive implementation, only PHY-layer (spectrum
5
behavior) is simulated. Therefore, the access-pointto-client link quality of service (QoS) is the primary parameter to evaluate the system capacity of the occupied channel (or the sub-network in terms of the point-to-multi-point scheme).
Figure 1. Conceptual diagram of experiment. For an OFDM PHY layer, the true network capacity is largely dependent on 802.11-type MAC-layer with CSMA/CA link control [10] [11] with a certain wireless channel model. The channel capacity is dependent on the SINR, a relationship we will further develop in the final paper, and so we focus on the link SINR simulation on the spectrum behavior by adding cognitive spectrum management to the current WLAN standard. From the simulation it is clear that current 802.11a/g standard has very limited spectrum efficiency due to static OFDM channel assignment. By adding a layer of cognition to observe the spectrum condition and adaptively allocate the needed spectrum, two key performances are greatly improved. First, the link QoS is improved due to optimized channel selection; second, the utilization of the spectrum, as a global resource, is balanced throughout the network due to interference avoidance at each cognitive network node. Transmitter power adaptation is to be added into link cognition capability as the next step of the simulation, and significant averaged power reduction is expected as a result of the adaptive interference avoidance.
Figure 2. Sample experiment of the geography of user and interferers and the observed spectrum of SINR. In the static channel assignment simulation, the channels of APs and users are fixed regardless of the current channel condition. To illustrate the advantage of cognitive spectrum allocation, both the AP and user need only to have the capability of sensing the signal in the operational band, and a simple optimization algorithm is applied to pick the static channel defined by the 802.11a standard. Even with such limited cognition for the spectrum, the performance is greatly improved both in link quality and overall spectrum efficiency. Figure 2 shows a sample of the simulation environment. First, it shows the placement of the WLAN systems with the circle represents the user and the asterisk the access point desired for communications. All other access points and spectrum users are interferers represented by triangles. The figure also shows the spectrum use for the available channels. It is immediately obvious from this figure that channel 4 will provide the highest SINR.
The simulation system is developed using Matlab. The OFDM radio link is simplified by assuming a AWGN wireless channel. The established ad-hoc network consists of multiple access points (APs) with point-to-multipoint link configuration.
Maldonado-Febus
6
In the final version of this paper, we will expand the simulation results, specifically by adding MAC-layer concepts to provide results in terms of channel capacity, and we have presented here only preliminary results based on our cognitive radio algorithm work.
35 Cognitive Radio Standard
Number of Access Points
30 25 20 15
Acknowledgements
10 5
This work was supported by the National Science Foundation under grants 9983463 and DGE-9987586 and in part by the Virginia Space Grant Consortium.
0 -5 -10 -15
References 0
5
10
15
20
25
30
Average SINR
Figure 3. Improvement in SINR between CR and standard technique. Simulation process and results CR-OFDM-PHY model: the AP's channel can be dynamically changed according to location and interference level, and the subscribers can pick an AP based on sensed SINR and load condition of each AP. The averaged subscriber link SINR is the parameter to evaluate the system performance, and the standard deviation of SINR for all the network nodes is the parameter to evaluate the balance or efficiency of the global spectrum management. The statistical results are obtained through a Monte Carlo simulation. Adaptation by the cognitive radio for link spectrum optimization is performed each iteration. The results of this preliminary experiment show that improvement in the ISM band’s SINR is increased by 20 dB using our cognitive radio techniques. As stated before, the performance of network capacity and throughput can be estimated through analytical calculation [10] [11]. Note that this simulation is constrained to 802.11a/g-defined physical only thus the link performance improvement is quite limited. We plan on a standard-free WLAN physical layer simulation to experience a much better performance from an “unleashed” cognitive radio network. Conclusions Cognitive radio techniques offer a promising approach to dynamic spectrum allocation. A simple example shows a 20 dB SINR improvement for a wireless LAN, using cognitive techniques in an interference environment over that provided by current IEEE 802.11a service.
Maldonado-Febus
[1] J. Mitola, Cognitive Radio: An Integrated Agent Architecture for Software Defined Radio, Ph.D. dissertation, Royal Institute of Technology (KTH), 2000. [2] FCC, “Facilitating Opportunities for Flexible, Efficient, and Reliable spectrum Use Employing Cognitive Radio Technologies,” FCC Document ET Docket No. 03-108, Dec. 2003. [3] http://www.darpa.mil/ato/programs/xg [4] The SAFECOM Program, Department of Homeland Security, “Statement of Requirements for Public Safety Wireless Communications and Interoperability,” Version 1.0. March 10, 2004. http://www.safecomprogram.gov/SAFECOM/interoperabi lity/default.htm [5] J. Powell, "Cognitive and Software Radio: A Public Safety Regulatory Perspective," report to NPSTC meeting, June 2004. [6] J.H. Reed, Software Radio: A Modern Approach to Radio Engineering, Englewood Cliffs, NJ: PrenticeHall, 2004. [7] P. Leaves, et al., “Dynamic spectrum allocation in a multi-radio environment: concept and algorithm, 3G Mobile Communication Technologies,” IEE Second International Conference on 3G Mobile Communications, pp. 53 – 57, 2001. [8] IEEE Std 802.11a-1999. Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications: High Speed Physical layer in the 5GHz Band. [9] IEEE Std 802.11g/D1.1-2001, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) speei5cations: specifications: Further HigherSpeed Physical Layer Extension in the 2.4 GHz Band. [10] P. Gupta and P. R. Kumar, “The capacity of wireless networks,” IEEE Trans. Information Theory, vol. 46, pp. 388 – 404, Mar. 2000. [11] O. Tickoo and B. Sikdar, “On the impact of IEEE 802.11 MAC on traffic characteristics,”, IEEE Trans. Selected Areas in Communications, vol. 21, pp. 189 – 203, Feb. 2003.
7
