A new MAC protocol for broadband wireless ... - Springer Link

Telecommun Syst (2014) 57:13–23 DOI 10.1007/s11235-013-9771-5

A new MAC protocol for broadband wireless communications and its performance evaluation Sedat Atmaca · Alper Karahan · Celal Ceken · Ismail Erturk

Published online: 25 July 2013 © Springer Science+Business Media New York 2013

Abstract This paper presents a new Time Division Multiple Access/Frequency Division Duplexing (TDMA/FDD) based Medium Access Control (MAC) protocol for broadband wireless networks, supporting Quality of Service (QoS) for real-time multimedia applications. It also gives the Call Blocking Probability (CBP), packet end-to-end delay and utilization analysis of different service classes, as they are most essential performance criterions in broadband wireless network assessment. The Connection Admission Control (CAC) mechanism in the proposed MAC efficiently organizes the bandwidth allocation for different service classes by means of a fairness based scheduling algorithm. In addition, the simulation model of the proposed MAC scheme is realized by using OPNET Modeler network simulator. The results of the analytical calculations for the CBPs are compared to those of the simulation of the proposed MAC, thus validity of the MAC protocol is proved. Keywords QoS · MAC · Bandwidth allocation · Call blocking probability · Performance evaluation

S. Atmaca (B) · A. Karahan · C. Ceken Technical Education Faculty, Electronics and Computer Education Department, Kocaeli University, 41380 Kocaeli, Turkey e-mail: [email protected] A. Karahan e-mail: [email protected] C. Ceken e-mail: [email protected] I. Erturk Faculty of Engineering, Computer Engineering Department, Turgut Ozal University, 06010 Ankara, Turkey e-mail: [email protected]

1 Introduction The importance of wireless communications, especially the multimedia applications, has gained much importance due to the developments in high performance microelectronic circuits and devices. The current trend in wireless networks is to provide multiple traffic types such as voice, video and data with the required QoS [1, 2]. In these networks, the allocation of available bandwidth efficiently and reliably to offer a guaranteed QoS is one of the most important issues due to the requirement of a well managed QoS-sensitive MAC protocol [3]. The traditional wireless communications systems are unable to manage efficiently and effectively multiple traffic types, as these systems were designed only for voice communication. In multi-service networks, where different applications with different requirements in terms of bandwidth, delay or jitter use the common network simultaneously, the bandwidth allocation problem becomes even more important. In order to maintain multimedia traffics in a wireless networks, a well organized QoS aware MAC protocol is required [4]. There are various MAC protocols provisioning QoS support for multiple type traffics in wireless networks [5, 6]. These MAC protocols are based on Frequency Division Multiple Access (FDMA), Time Division Multiple Access (TDMA) and Code Division Multiple Access (CDMA). In this study, TDMA is chosen as a MAC technique, as it is more appropriate for flexible bandwidth allocation. In a TDMA based MAC protocol, which is generally designed for connection oriented services, the channel is divided into the frames and frames are divided into equal-length time slots. In a network where different service classes are used, TDMA has the flexibility to assign one or more transmission slots to a user according to the user’s bandwidth requirements.

14

CBP is one of the key performance metrics for QoS assessment in multi-service networks since it determines if a user can get required bandwidth from the Base Station (BS). Thus, CBP performance of a multi-service network is always taken into the account while designing a multiservice networks. CBP analysis of multi-service networks have extensively been studied by many researchers [7–10]. Multi-service networks support call demands having a different number of the Bandwidth Units (BUs) which is defined as the integer number of call requests offered to the system [11]. However in most of the papers published, CBPs have been analyzed with small size system capacity. But, in our work, the network with large system capacity which is more appropriate for next generation wireless communication is considered. In [12], CBP calculation in a single link loss model where calls of each service-class come from fairly small finite sources and compete for the available link bandwidth under the complete sharing policy is presented. This work utilizes relatively small system capacity when compared with system capacity of our proposed network model. The analysis of CBP calculation in multi-rate models is presented in [13]. This paper proposes product form solution for the state probabilities. However, product form solution becomes intractable when the system capacity and service classes increase. Therefore, a recursive solution for the state probabilities is proposed in [14]. It simplifies the calculations for a large number of service classes and system capacity. In [15], a new methodology for call performance evaluation of multi-class services is presented and the CBP for each service class is calculated at the MAC layer. This work also uses small system capacity and the number of BUs demanded from the network is fairly high. Thus, it is easy to calculate CBPs for different service classes in such a system. In this paper, we first propose a new TDMA/FDD based MAC protocol, utilizing large system capacity and supporting broadband wireless multimedia applications. Our proposed MAC uses complete sharing admission policy in which all connections are accepted if resources are available [16]. It differs from the previous works [17] in that the frame size is adjusted to utterly allow using the complete sharing policy so that the required number of time slots can be allocated regardless of their place in the frame. We then present analytical models to calculate the CBP of different service classes used in the MAC. We also present the simulation results of the End-to-End Packet Delay (EED) and channel utilization for different service classes of the proposed MAC. Our EED delay results satisfy the QoS requirements of the different service classes. In addition, our channel utilization results are applicable for broadband wireless communications. Lastly, simulation models of the proposed MAC are realized by using OPNET Modeler simulation tools and CBP results of analytical model are validated by the simulation results.

S. Atmaca et al.

The rest of the paper is organized as follows: fundamentals of the QoS support for multi-class wireless broadband networks are explained in Sect. 2. The proposed MAC protocol and its simulation models are presented in detail in Sect. 3. Analytical models of proposed MAC protocol are obtained in Sect. 4. Simulation scenarios of the proposed MAC protocol are given in Sect. 5. The comparison of the analytical and simulation results of the proposed MAC Protocol is offered in Sect. 6. Conclusions and comments are given in Sect. 7.

2 Preliminaries of QoS support for heterogeneous data traffics in wireless broadband networks In recent years, there has been a tremendous growth in development of QoS support for real-time multimedia traffics in wireless systems [18–20]. Although there are a variety of technologies that can provide these services, such as thirdgeneration cellular networks and Wireless Local Area Networks (WLANs), Wireless Asynchronous Transfer Mode (WATM) has some advantage of providing end-to-end multimedia capabilities with guaranteed QoS. WATM provides several significant advantages including; (1) flexible bandwidth allocation and service type selection for multimedia applications, (2) efficient multiplexing of bursty data and multimedia traffic sources, (3) a wireless platform with multimedia support for any QoS needs in a much better way [17, 21]. As WATM technology can support bandwidth on demand, it is capable of combining many different kinds of services into one medium [6]. There are four well-known different service classes in ATM systems defined as follows; • • • •

Constant Bit Rate (CBR), Variable Bit Rate (VBR), Available Bit Rate (ABR), Unspecified Bit Rate (UBR).

Real-time digital audio and video applications require CBR and VBR (real-time) service categories, respectively. For CBR services, the Peak Cell Rate (PCR) descriptor must be provided to the ATM network, while for VBR services, Sustainable Cell Rates (SCR) is mandatory. ABR service class is specified by the Minimum Cell Rate (MCR) descriptor. Applications where guaranteed packet delivery is not required (e.g., Internet traffic and e-mail) are provided over a UBR service. Therefore, any cells sent on UBR connections have to wait until there is free bandwidth available and not being used by either CBR or VBR traffic [17].

A new MAC protocol for broadband wireless communications and its performance evaluation

15

Fig. 1 Frame structure of the proposed TDMA/FDD MAC protocol

3 Proposed MAC protocol supporting real-time multimedia applications and its simulation model In this paper, a wireless network, supporting multimedia traffic applications, that uses the TDMA/FDD MAC protocol is considered. In the proposed MAC, the TDMA/FDD approach presented in [17] has partly been used. TDMA is used as a multiplexing technique due to its superiority and suitability for real time multimedia applications. In TDMA technique, radio spectrum is divided into frames and frames are divided into time slots that can be assigned to different connections. Users access the entire radio spectrum only in its own dedicated time slot/s (previously allocated) to send/receive its data. FDD is adopted as a duplexing technique in the proposed MAC, thus, two different carrier frequencies are used for the uplink and downlink channels. The Wireless Terminals (WTs) are not able to communicate directly one another in the proposed method with the centralized architecture. The WTs, in order to communicate with any other, initially ask for a transmission channel from the BS. According to the QoS requirements of a given connection request, the BS assigns the available time slots using a Dynamic Slot Allocation Table (DSAT). The number of the time slots which is going to be allocated for the connection request is determined by bandwidth requirements. The BS constitutes DSAT by managing its scheduling the new QoS-aware algorithms based on ATM service classes for real-time multimedia traffics. A connection with required QoS guarantee can only be established if the number of slots is sufficient to support the type of service class requirement. This proposed scheme differs from the previous works in especially that the frame size is adjusted to utterly allow using the complete sharing policy so that the required number of time slots can be allocated regardless of where they are placed in the frame. The frame structure of the proposed MAC, which is the basis for developing QoS and DSAT algorithms, is shown in Fig. 1. It is mainly composed of two types of slots described

as control and data slots. Control slots are used for transmission of the control packets between BS and WTs while data slots are used for transmission of data packets between WTs via BS. The proposed MAC is divided into two main complementary parts, one operating at the WTs and the other operating at the BS. QoS supported data transmissions can be provided when these two MAC parts work together. WTs and BSs can be fixed or mobile in WATM architecture. In the development phase of the proposed MAC, the networking environment is projected as including fixed (stationary) WTs and a BS, both with running parts of the new QoS algorithms concurrently. The following subsections give details of all functions and algorithms in both parts together with their modeling and simulation environment in the OPNET Modeler. 3.1 Wireless terminal MAC model The WTs designed for the proposed network principally perform three main processes to implement its functions and algorithms. The first is to request a connection establishment/termination from the BS. The second is to obtain its own time slot/s from the BS. And the last is to send multimedia data in the allocated time slots, supported with the required level of QoS. Before sending data to any other WT in the network, WTs have to inform the BS about their bandwidth requirement for the connection establishment [22]. WTs accomplish this by creating a control packet called con_req_packet containing source terminal address, and service requirements (SLSx) of the wireless applications (Fig. 2a). It is then sent to the BS in the first available control slot. If the number of slots is not adequate to provide the type of service class requirement of the WT, a connection cannot be established and the WT may later reattempt to connect again by sending another connection request packet to the BS. After a connection is established, the wireless application traffic in data packet is sent to the destination WT in the time slots allocated to this connection (Fig. 2b). An ATM cell

16

S. Atmaca et al.

Fig. 2 (a) Connection request packet (con_req_packet), (b) Data packet (data_packet adapted from ATM cell), (c) Connection terminate packet (con_end_packet)

Fig. 3 Wireless terminal state transition diagram

comprises 53 bytes, consisting of a 5-byte header and a 48byte information field (PAYLOAD). A 2-byte error correction field (Cyclic Redundancy Check, CRC) which is used for detection and correction of the possible cell transmission errors is also added to all ATM cells defined in detail in [23] and [24]. In the last phase of the communications, the source WT creates a control packet called as con_end_packet to terminate the connection (Fig. 2c), and sends it to the BS in the first available control slot. The state diagram of a proposed WT MAC model is illustrated in Fig. 3. In the ‘Initialization’ state machine, state variables are defined and their first values are loaded. In the ‘Idle’ state machine, any interrupts from other layers are waited. The ‘Connection Request’ state machine creates the connection request packet containing QoS parameters and sends it to the BS. In the ‘Packet from Source’ state machine, data in cells received from source are sent to the destination in the time slots allocated to this terminal. The ‘Packet from Antenna’ state machine obtains connection time slots assigned by the BS using its QoS-aware DSAT scheduling algorithms. The ‘Connection Termination’ state machine creates a connection termination packet and sends it to the BS. The ‘Send Data Packet’ state machine sends the data packet to the upper layer.

3.2 Base station MAC model The core functions of the BS MAC are realized using three fundamental processes. The first is to assign a time slot/s for a connection request using DSAT scheduling algorithms, the second is to forward any arrived data packets to their destinations and the last is to terminate any active connection. The most important and primary function of the proposed BS MAC is the efficient management of the DSAT. The BS state diagram of the proposed QoS-aware MAC is illustrated in Fig. 4. ‘Packet from Antenna’ state machine delivers arriving packets to the next state machine according to the packet formats. The ‘Connection Request’ state machine fairly executes a scheduling algorithm that manages the DSAT considering QoS requirements of the connection requests. The DSAT is created and its default values are loaded by the ‘Initialization’ state. The ‘Idle’ state always waits for interrupts. The ‘Connection Termination’ and ‘Send Data Packet’ state machines handle connection termination requests and data packet deliveries, respectively. ‘CBR’, ‘VBR’, ‘ABR’ and ‘UBR’ state machines are used for time slot allocation for these services according to their bandwidth requirements.


The DSAT and its scheduling algorithm in the BS are the most important parts of the proposed MAC protocol. Providing QoS guaranteed services for bursty traffics such as real-time multimedia applications is achieved by efficiently managing the DSAT. Figure 5 shows the structure of the proposed DSAT. Slot Number, Terminal ID, App Num, Priority and Service Class rows of the table represent slot number, address of the WT using this slot, application number, priority of the slot (−1: default, HIGH (H): guaranteed, LOW (L): non-guaranteed) and service class of the connection (−1: default, 0: CBR, 1: VBR, 2: ABR, 3: UBR), respectively. For a new connection establishment, the number of slots required for this connection and their guarantee level in the DSAT are determined using QoS parameters and traffic descriptors used in the standard ATM connections [25, 26]. According to the Peak Cell Rate (PCR), Sustainable Cell Rate (SCR) and Minimum Cell Rate (MCR) of CBR, VBR

Fig. 4 Base station state transition diagram

Fig. 5 Time slot structure of the DSAT

17

and ABR traffics respectively, priority fields of guaranteed slots are set to HIGH. Using the other QoS parameters (Cell Transfer Delay (CTD) and Cell Delay Variation (CDV)) and traffic descriptors, the priority field is set to LOW in the DSAT for the non-guaranteed slots required for the VBR, ABR and UBR traffics. Any slot whose priority field is LOW may be reassigned for a new connection whose slot priority field is required to be HIGH (e.g., for a CBR traffic). Priority field of UBR traffic slots are always set to LOW (except the one whose priority field is set to HIGH that is required for connection establishment and to sustain the connection) and, if necessary, these slots can be used by any guaranteed CBR, VBR or ABR traffics since the UBR service does not provide any QoS guarantees (i.e., best effort services). Furthermore, the first slot of a frame is used for connection request and connection termination packets (Fig. 5). When a WT has a control packet to send, it uses the first available control slot to send it. Connection establishment scheme of the different service classes is described as follows. • Establishment of a CBR connection: For a CBR connection request, the number of HIGH priority slots (guaranteed slot) is determined using the PCR. In the BS, firstly the empty slots (i.e. with no priority = −1) are assigned to HIGH priority for this connection demand. If the number of empty slots in the DSAT is not sufficient then nonguaranteed slots (i.e. with LOW Priority), which have already been allocated for UBR, ABR and VBR connections, are assigned in this given order. Finally, if the number of HIGH priority slots is sufficient to provide the CBR service class requirement of the WT, the connection is established and both the requesting WT and others are informed about its reserved slots. Otherwise, connection cannot be established and the WT must reattempt to set up a connection after a certain time. • Establishment of a VBR connection: For a VBR connection request, the number of HIGH priority slots is determined using the SCR, and the number of LOW priority slots (non-guaranteed slot) is determined using the PCR. If the number of HIGH priority slots is sufficient to provide the VBR service class requirement of the WT, the connection is established and if there are still empty available slots, LOW priority slots are assigned to this connection. If the number of HIGH priority slots is not sufficient, connection cannot be established and the terminal reattempt to set up a connection after a certain time.

18

S. Atmaca et al.

• Establishment of an ABR connection: For an ABR connection request, the number of HIGH priority slots is determined using MCR parameter whereas the number of non-guaranteed slots is determined by PCR. If the number of HIGH priority slots is sufficient to provide the ABR service class requirement of the WT, the connection is established and if there are still empty available slots, LOW priority slots are assigned to this connection. If the number of HIGH priority slots is not sufficient, connection cannot be established and the terminal must reattempt to set up a connection after a certain time. • Establishment of an UBR connection: For a UBR connection request, any available empty slots, unused or remaining from other service classes, can be used with no QoS guarantees. The PCR parameter of the UBR traffic represents the maximum data rate that can be supported in such a given connection and it is used to determine the number of LOW priority slots.

4 Analytical model for the call blocking probability of the proposed MAC protocol In our TDMA/FDD MAC protocol, the determination of the call blocking probabilities of K different service classes is performed by adopting the CBP formulas of [12], because it is suitable for large system capacity. In the MAC designed, each service class k has a finite source population Nk and requests for constant rate of bk BUs from a shared link. The system capacity (SC) of the shared link is SC bandwidth unit. αk = vk /μk , is the number of traffic attempts by the source of service class k, where 1/vk is the expected inactive period during which a class-k user does not make a new call attempt after the completion or loss of its preceding call; and 1/μk is the mean holding time of a class-k call. Thus, CBP of service-class k, denoted as Bk , is calculated by the formula described below, SC

Bk =

G−1 G(c)

where G and G(c) are the normalization constant and the stationary probability in that order, and they are calculated as follows, SC

G(c)

(2)

c=0

⎧ 1, for c = 0 ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ 1 K N α b [c/bk ] β m−1 G(c − mb ), k k=1 k k k m=1 k G(c) = c ⎪ for c = 1, . . . , SC ⎪ ⎪ ⎪ ⎪ ⎩ 0, otherwise

In this paper, simulation scenarios utilizing the proposed MAC protocol have been developed with OPNET Modeler [27]. The wireless communication channel is modeled in OPNET by 13 pipeline stages including transmission power, antenna gains, path-loss, background noise, interference, signal-to-noise ratio, propagation delay and transmission delay [28]. A realistic wireless scenario has been created by using this powerful simulation environment. The free space propagation model for the path loss (PL ), which represents the signal attenuation (in dB) between the effective transmitted power and received power, has been modeled as: Pt λ2 (4) = −10 log PL (dB) = 10 log Pr (4π)2 d 2 where λ is the carrier signal wavelength and d is the propagation distance between transmitter and receiver antennas. Power of the packet received at each node is calculated as: Pr =

Pt Gt Gr λ2 = Pt Gt Gr PL (4π)2 d 2

(5)

where Pt is the signal power at the transmitting antenna, Gt and Gr are the gains of the transmitting and receiving antennas, respectively. The Signal to Interference plus Noise Ratio (SINR) is calculated as: SINR =

N0 +

Pr n

j =1 Ij

(6)

where Pr is the power of received packet, Ij represents the power received from j th interferer, n is the number of interferers and N0 is the thermal noise. The amount of thermal noise to be found in a bandwidth of 1 Hz in any devices is computed by N0 = k · T · B

(7)

where k is the Boltzmann’s constant (1)

c=SC−bk+1

G=

5 Simulation scenarios of the proposed MAC protocol

(3)

where βk = −αk and [c/bk ] is the largest integer not greater than c/bk .

k = 1.38 · 10−23 J K−1

(8)

T is temperature in Kelvin, and B is the channel bandwidth in Hertz. There are some differences between power levels of the two simultaneously arriving packets. In order to distinguish the desired packet from the others, power reception threshold is used to get the more powerful packet. OPNET pipeline stages determine whether a packet is valid or collided by SINR computation. Packets that do not satisfy the SINR threshold for the wireless channel are ignored. The wireless channel gain is determined by path loss and does not vary with time, as nodes are considered immobile. The BS and the stationary (fixed wireless) WTs, uniformly distributed in the example networking scenario, employ the proposed MAC protocol in order to communicate


19

Fig. 6 An example simulation scenario utilizing our proposed MAC

Table 1 Simulation parameters

Frequency band

Uplink = 3 GHz, Downlink = 4 GHz

Transmitter power

BS = 100 mW, WTs = 100 mW

Uplink/Downlink bit rate

25 Mbit/s

Modulation Type

QPSK

Channel model

Free space propagation model (LoS)

The number of time slot

300

The frame length

6 ms

CBR parameters

PCR = 256 kbit/s, CTD = 10 ms, CDV = 1 ms

VBR parameters

SCR = 2000 kbit/s, PCR = 5000 kbit/s, CTD = 5 ms, CDV = 1 ms

ABR parameters

MCR = 720 kbit/s, PCR = 1000 kbit/s

UBR parameters

PCR = 50 kbit/s

SINR threshold value

27 dB

with each other in the same wireless environment. In simulation models, omni-directional antenna is considered to be used in WTs and BS of proposed MAC. In the example model shown in Fig. 6, there are 120 WTs on which multimedia data traffics are generated, transferred and received. After a regular connection establishment, a WT sets up a timer for a random ON state between [0–360] seconds using exponential distribution and during this period it transmits its data traffic. As soon as the ON state period expires, the WT terminates the connection. In a similar manner, after a regular connection termination, the WT sets up a timer for an OFF state between [0–360] seconds using exponential distribution, and during this period

it does not introduce any data traffic. The multimedia data traffic introduced to the network by any WT is randomly destined to another WT. The simulations are repeated for both different number of wireless users and increasing offered loads. The simulation parameters used are given in Table 1.

6 Analytical and simulation results of the proposed MAC protocol Simulation results of the proposed MAC model are presented with a varying infinite source population, followed

20

S. Atmaca et al.

Fig. 7 Call Blocking Probability results of the proposed TDMA/FDD MAC

by performance comparisons and analysis. The performance metrics concerned in this research work are call blocking probability, mean EED and channel utilization. CBP is the ratio of the number of blocked calls to the total number of connection requests. Blocked calls are defined as the calls that are refused by the MAC due to inadequate number of bandwidth units. EED is the time between the transmission of the first bit of a packet from the source and the reception of the last bit of the packet by the destination after the connection is established. In our system, the channel utilization is defined as ratio of the achieved throughput in bps to the total channel capacity. The CBP and delay results with their explanations of the proposed MAC protocol are given in detail in the following subsections.

6.1 Call blocking probability results The simulation results of the call blocking probabilities as function of number of WTs for each service classes of the proposed MAC protocol have been presented in Fig. 7a, b, c, d and validated by comparing to the results of analytical models. The results of the simulation model for the CBP match with the analytical results and confirm high accuracy of the proposed model. As the number of WTs increases in Fig. 7a, the CBP for the VBR service classes is obtained as 90 % for 96 WTs. When the number of WTs exceeds 96, almost all of the network resources are utilized, and nearly all VBR calls are then blocked. In Fig. 7b, the CBP for the ABR service classes is obtained as 90 % for 288 WTs, and then almost all of the network resources are exploited, re-


21

Fig. 8 Delay results of the proposed TDMA/FDD MAC simulation model

sulting in nearly all ABR calls blocked. In Fig. 7c, the CBP for CBR service classes is obtained as 70 % for 480 WTs. For CBR service classes, there is approximately 4 % difference between the simulation and analytical results, which is stemming from the assumptions adopted in the analytical models. For instance modeling the control slots are ignored in the analytical model and this results in a slight increase in the bandwidth utilization for high number of WTs. In Fig. 7d, the CBP for the all service classes is obtained for comparison reasons. This figure shows that the WTs which require high bandwidth experience high call blocking probability. Exploiting the DSAT and its scheduling algorithm, the total bandwidth is efficiently and dynamically shared among the WTs according to their bandwidth requests and QoS parameters. Thus, for example, when 96 WTs are uniformly

served by VBR, ABR and CBR connections, CBPs have been obtained as 90 %, 55 % and 27 %, respectively. These results conclude that the WTs which require VBR connections with high and fluctuating bandwidth experience high call blocking probability. 6.2 End-to-end packet delay results In Fig. 8, mean EED results of the all service classes (i.e., supported with CBR, VBR, ABR and UBR service classes) as a function of the offered load for a specified number of users in the simulation model are presented. Considering the different natures of the real-time multimedia application traffics, different mean EEDs have been obtained from each model as a consequence of employing the new QoS-aware DSAT scheduling algorithms utilized in the MAC. Mean

22

S. Atmaca et al.

7 Conclusions and comments

Fig. 9 Channel utilization results of the proposed TDMA/FDD MAC simulation model

EED graphics especially for the CBR and VBR data traffics (less than 4.5 ms and 3.5 ms respectively well below the predefined 10 ms and 5 ms thresholds) verify that the required level of QoS is guaranteed as the offered load per user varies from 20 kbits/s to 300 kbits/s for CBR and 200 kbits/s to 2400 kbits/s for VBR. Moreover, as the offered load increases, the proposed MAC continues providing its guaranteed service support for CBR and VBR data traffics resulting in acceptable EEDs. The ABR service class also experiences relatively low EED results when the offered load is increased from 60 kbits/s to 900 kbits/s. The UBR data traffics practice relatively low EEDs when the offered load is between 10 kbits/s and 50 kbits/s. When the offered load exceeds 50 kbits/s, extremely high EEDs are obtained. The mean EED results obtained from each connection with different service classes in the proposed MAC are below the acceptable delays, predefined in the simulation parameters in Table 1, even when the offered load is maximum. These EED results show that our proposed MAC enables transmission of the different real-time multimedia applications. 6.3 Channel utilization results In Fig. 9, the channel utilization as a function of normalized offered network load is presented. In the simulation scenario, all service classes are uniformly distributed among the wireless terminals and the loads of all traffic types offered to the network are equal. The channel utilization of all service classes increases when the offered load is increased. However, the utilization of each service class reaches its maximum value for different offered load values since each one utilizes different amount of the total bandwidth according to their QoS requirements. When the offered load exceeds 2.2, maximum overall channel utilization is obtained as 0.84 which is suitable for broadband wireless communications.

This paper introduces a new TDMA/FDD based MAC protocol for wireless data communications, providing QoS guarantees for real-time multimedia applications and presents detailed CBP analyses of different service classes. The simulation model of the proposed MAC scheme is realized by using OPNET Modeler network simulator. The CBP results of the proposed MAC have been validated by comparing to the analytical model results. The results of the simulation model for the CBP match with the analytical results and confirm high accuracy of the proposed model. The comparative results conclude that the proposed MAC is capable of supporting broadband multimedia applications. The work is to be furthered to include mobility issues and to provide priority mechanisms for the blocked connections carrying especially CBR and VBR traffics.

References 1. Akyildiz, I. F., Janise, M. J., Martorell, L. C., Puigjaner, R., & Yesha, Y. (1999). Medium access control protocols for multimedia traffic in wireless networks. IEEE Network, 2–11, July/August. 2. Hsu, I., & Walrand, J. (1997). Admission control and resource management for multi-service ATM networks. Telecommunications Systems, 7, 185–207. 3. Paladina, L., Paone, M., & Puliafito, A. (2007). Efficient CAC in broadband wireless access networks based on hierarchical structures. IEEE Mobile Adhoc and Sensor Systems, 1–6. 4. Pliakas, T., Kormentzas, G., & Skianis, C. (2007). Scalable video streaming traffic delivery in IP/UMTS networking environments. Journal of Multimedia, 2, 37–46. 5. Sanchez, J., Martinez, R., & Marcellin, M. W. (1997). A survey of MAC protocols proposed for wireless ATM. IEEE Network, 52– 62. 6. Mikkonen, J., & Nenonen, L. (1998). Wireless ATM MAC performance evaluation, a case study: HIPERLAN type 1 vs. modified MDR. Mobile Networks and Applications, 247–259. 7. Ross, K. W. (1995). Multiservice loss models for broadband telecommunication networks. Berlin: Springer. 8. Delbrouck, L. E. N. (1983). On the steady state distribution in a service facility with different peakedness factors and capacity requirements. IEEE Transactions on Communications, 31(11), 1209–1211. 9. Kaufman, J. (1981). Blocking in a shared resource environment. IEEE Transactions on Communications, 29, 1474–1481. 10. Zeng, H., Fang, Y., & Chlamtac, I. (2002). Call blocking performance study for PCS networks under more realistic mobility assumptions. Telecommunications Systems, 19, 125–146. 11. Ramaswamy, S., & Gburzynski, P. (1999). A dynamic call admission scheme for VBR traffic in ATM networks. Telecommunications Systems, 11, 373–394. 12. Moscholios, I. D., Logothetis, M. D., & Nikolaropoulos, P. I. (2005). Engset multi-rate state-dependent loss models. Performance Evaluation, 59, 247–277. 13. Aein, J. M. (1978). A multi-user-class, blocked-calls-cleared, demand access model. IEEE Transactions on Communications, 26(3). 14. Roberts, J. (1981). A service system with heterogeneous user requirements—application to multi-service telecommunications


15.

16.

17.

18.

19.

20.

21.

22.

23. 24. 25.

26. 27. 28.

systems. In G. Pujolle (Ed.), Proceedings of performance of data communications systems and their applications (pp. 423–431). Amsterdam: North Holland. Mertzanis, I., Sfikas, G., Tafazolli, R., & Evans, B. G. (1999). Satellite–ATM networking and call performance evaluation for multimedia broadband services. International Journal of Satellite Communications, 17, 107–127. Stanisic, V., & Devetsikiotis, M. (2005). An analysis of bandwidth allocation strategies in multiservice networks. In IEEE international conference on communications (Vol. 3, pp. 1529–1533). Ceken, C., Erturk, I., & Bayilmis, C. (2006). A new MAC protocol design based on TDMA/FDD for QoS support in WATM networks. Computer Standards & Interfaces, 28(4), 451–466. Pliakas, T., Kormentzas, G., & Skianis, C. (2008). End-to-end QoS issues of MPEG-4 FGS video streaming traffic delivery in an IP/DVB/UMTS network. European Journal of Operational Research, 191, 1089–1100. Alagoz, F. (2002). Approximations on the aggregate MPEG video traffic and their impact on admission control. Turkish Journal of Electrical Engineering and Computer Sciences, 10, 73–84. Raychaudhuri, D. (1996). Wireless ATM networks: architecture, system design and prototyping. IEEE Personal Communications, Special Issue on Wireless ATM, 42–49. Racz, S., Telek, M., & Fodor, G. (2001). Link capacity sharing between guaranteed and best effort services on an ATM transmission link under GoS constraints. Telecommunications Systems, 17(1– 2), 93–114. Atmaca, S., Ceken, C., & Erturk, I. (2009). A new QoS-aware TDMA/FDD MAC protocol with multi-beam directional antennas. Computer Standards & Interfaces, 31(4), 816–829. Stallings, W. (2005). Wireless communications & networks (2nd ed.). Upper Saddle River: Pearson Prentice Hall. Ayanoglu, E. (1999). Wireless broadband and ATM systems. Computer Networks, 31, 395–409. Ayanoglu, E., Eng, K. Y., & Karol, M. J. (1996). Wireless ATM: limits, challenges, and proposals. IEEE Personal Communications Magazine, 3, 18–34. ATM Forum, Traffic management specification, version 4.0, April 1996. OPNET Technologies Inc. (2011). Optimum network simulation and engineering tool, available from http://www.opnet.com. Rappaport, T. S. (2002). Wireless communications: principles and practice (2nd ed.). Upper Saddle River: Prentice Hall. Sedat Atmaca received his Ph.D. degree from Kocaeli University, Turkey, in 2009. His active research interests include wireless communications, smart antennas, wireless sensor networks, high speed communication protocols, cognitive radio networks and modeling, simulation and analyzing of telecommunications networks.

23 Alper Karahan received the B.S. and M.Sc. degrees from Kocaeli University, Turkey in 2006 and 2010, respectively. His research interests are computer networks modeling and analysis, wireless data communications, wireless sensor networks, WBAN, LANs/WAN, QoS and MACs.

Celal Ceken received his Ph.D. degree in 2004 from Kocaeli University, Kocaeli, Turkey. Since 1999 he has been with the Electronics and Computer Education Department of the Kocaeli University. He is interested in many fields such as wireless communications, computer networks, web based programming and database systems. His current research interests include wireless MAC protocols, high speed wireless communication protocols, next generation wireless systems, cognitive radio, wireless sensor/actuator networks and wireless networked control systems. Ismail Erturk received his M.Sc. and Ph.D. degrees from University of Sussex, UK in 1996 and 2000, respectively. His research interests include WSNs, MAC, network security, steganography, ATM, internetworking, CAN, QoS and real-time multimedia applications.