C
Mobile Networks and Applications 10, 617–625, 2005 2005 Springer Science + Business Media, Inc. Manufactured in The Netherlands.
Maximizing Transmission Time (MTT): A Distributed MAC Scheme for Enhancing Wireless LAN Performance XIN GANG WANG, GEYONG MIN, JOHN E. MELLOR and LIN GUAN Mobile Computing and Networks Research Group, Department of Computing, School of Informatics, University of Bradford, Bradford, BD7 1DP, UK
Abstract. Distributed contention-based Medium Access Control (MAC) protocols are the fundamental components for IEEE 802.11 type Wireless LANs (WLAN). The deficiency of these types of MAC protocols mainly comes from the idle slots used to contend the channel and from the transmission collisions due to the same backoff slot value being generated. Assigning the same transmission opportunity to various length packets also degrades the system performance. This study takes account of the above issues and presents a new MAC scheme called Maximizing Transmission Time (MTT) to enhance the wireless LAN performance. This scheme allows each station to transmit a burst of packets after winning a transmission opportunity instead of just one packet. This idea can reduce the average number of waiting slots and collision probability in each transmission cycle. Moreover, in order to ensure fairness among stations, a maximum transmission period is assigned to each station for controlling the length of the bursty transmission. An analytical performance model is derived for computing the throughput of the MTT scheme. The extensive simulation experiments reveal that the proposed method can enhance the wireless LANs performance significantly with high throughput, low delay and high degree of fairness. Keywords: WLAN, distributed MAC, maximizing transmission time, throughput, delay, fairness, analytical modeling
1. Introduction In WLAN, the MAC protocol is the main element that determines the efficiency in sharing the limited communication bandwidth of the wireless channel [7]. Different access methods have been proposed to control the multiple users’ access to the shared medium. Examples are FDMA, TDMA, CDMA, ALOHA and their combinations and variations [3]. One of the most popular approach is the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) protocol. It is adopted by the IEEE standards organization as the underlying protocol for the widely deployed IEEE 802.11 type WLAN [7]. In general, there are two broad categories of multiple access control protocols: contention-based multiple access protocols and reservation-based multiple access protocols [2]. Contention-based multiple access protocols are usually used in distributed network architectures (i.e. ad hoc networks). Its simplicity makes it easy to implement and it has the flexibility to allow extension of the network. Moreover, it has a low delay characteristic and is therefore suitable for bursty data traffic under low network loads. Reservation-based multiple access protocols are generally used in centralized network architectures and are suitable for systems under heavy loads. They also have the advantage of being able to support real-time traffic. However, they lack flexibility and can generate greater overhead management information than the contention approach [12]. In this paper, we focus on the study of distributed contention-based MAC protocols. The core contribution of this paper is to propose a novel MAC scheme, named MTT for the CSMA/CA based WLAN
system. This scheme enables each station to transmit a burst of packets after winning a transmission opportunity instead of just one as in the standard IEEE 802.11 DCF. Under the ideal channel condition (no hidden terminals and environment noise), the wasted system time slots mainly come from the backoff time used to contend the channel and from transmission collisions due to the same backoff slot value being generated. Assigning the same transmission opportunity to various length packets also causes unfairness problem and makes real time applications suffer from having to contend with lengthy data buckets. Increasing the number of transmission packets in each station can reduce the average waiting slots and collision probability in each virtual transmission cycle. Hence, assigning the same MTT to all stations can ensure maximum degree of fairness among them. Another contribution of this paper is to derive an analytical performance model to compute the saturation throughput of the MTT method. Under saturation assumptions this approximation model matches the numerical results obtained from simulation. The extensive performance analysis reveals that the proposed method can improve the wireless LANs efficiency significantly with high throughput, low delay and high degree of fairness. The rest of this paper is organized as follows. In Section 2, related work in the literature is reviewed. Section 3 presents the proposed scheme and its implementation in detail. In Section 4 we derive an analytical model to compute the throughput of the new scheme. Section 5 compares the results obtained from the analytical model and those from simulation experiments and then examines the performance merits of the MTT scheme. Finally, Section 6 concludes the paper.
WANG ET AL.
618
2. Related work The first distributed random multi-access protocol was called ALOHA and was proposed in the late 1960s [1,6,13]. Later, to enhance traffic performance over the multi-access broadcast channels, different schemes have been proposed and added to the ALOHA protocol. These include a slotted ALOHA and carrier sense multiple access (CSMA) mechanism to reduce collision probability; a Request-to-Send (RTS) and Clear-toSend (CTS) handshaking mechanism to avoid hidden terminal problem; a data-ACK sequence to improve the reliability and different scheduling algorithms to improve the fairness among stations ([3,9,14,15] and references therein). The widely deployed Ethernet employs a Carrier Sense Multiple Access with Collision Detection (CSMA/CD) protocol, in which collision detection is used to reduce the time spent on the transmission of collided packets. However, its wireless counterpart IEEE 802.11 WLAN does not have the physical ability to detect the collision errors and it uses CSMA/CA as the basis of MAC instead [7]. The collision avoidance mechanisms include RTS-CTS handshaking and a binary exponential backoff algorithm. Many research studies [4,5,10,15,17] concentrate on increasing the throughput performance of a distributed contention-based MAC protocol. The “maximum” throughput can be achieved by increasing the offered load. However, further increases of the offered load or the simulation time leads to an eventually significant decrease in the system throughput, owing to the well known fact that some random access schemes exhibit unstable behaviors [3,8]. Therefore, saturation throughput is used to describe the system workable throughput instead of “maximum” throughput and it refers to the maximum load that the system can carry in stable conditions. Under the saturation assumption, Bianchi [4] proposed an analytical model to compute the 802.11 DCF saturation throughputs and demonstrated its accuracy using simulation. Cali [5] analytically derived the average size of the contention window that maximizes the throughput under asymptotic conditions, which means that all the network stations always have a packet ready for transmission and it is equivalent to a saturation condition. Under the ideal channel condition, the deficiency of these MAC protocols mainly comes from the idle slots used to contend the channel and transmission collisions due to the same backoff slot value being generated [16]. A recent study [17] attempts to deal with these issues by giving a small idle backoff period for each station with successful packet transmission
and large backoff interval for the rest of the stations. When a station detects a number of consecutive idle slots it will start to reduce the backoff timer exponentially fast, compared to the linear decrease in backoff timer in the IEEE 802.11 MAC. In many of these publications fixed packet size is assumed to simplify the performance analysis. However, realistic networks support variable packet length. This is also one of the reasons for the short-term unfairness problem [10] in the broadcast channel. For instance, voice packets commonly have shorter length than data packets. According to the CSMA protocol, each packet stands the same chance to be transmitted over the channel. Therefore, the short packets can suffer from the long waiting time caused by transmitting lengthy packets. Vaida [15] attempted to solve this problem by proposing an algorithm which takes into account packet size and assigning small backoff intervals to small sizes packets. Although many of these innovative distributed contentionbased MAC schemes have been proposed, very few MAC protocols satisfy simultaneously all desirable properties such as high throughput, low delay and good fairness while maintaining the simplicity of implementation in realistic WLANs. Different from the previous work reported in the literature, this proposed new MAC scheme takes into account the abovementioned issues and improves the WLAN performance by allowing each station to transmit a burst of packets after winning a transmission opportunity instead of just one for reducing the average waiting slots. Furthermore, a maximum transmission period is assigned to each station used for controlling the length of the bursty transmission with the aim to maintain the fairness among stations. The results show that the proposed method can enhance the WLAN performance significantly with high throughput, low delay and high degree of fairness. 3. The proposed scheme Figure 1 depicts the basic operations of the IEEE 802.11 Distributed Coordination Function (DCF). A packet transmission cycle consists of backoff slots used for contending the channel and a successful packet transmission followed by an ACK packet returned from the destination station. Between the successful transmission cycles, it is possible that some packets encounter collision due to the same randomly generated backoff timer in different stations. During the collision period, the listening stations could not decode the packets and the transmission stations can not notice this fact. They can only be aware of the collision without receiving an ACK packet.
Figure 1. Basic operations of IEEE 802.11 DCF.
619
DISTRIBUTED MAC SCHEME FOR ENHANCING WIRELESS LAN PERFORMANCE
Figure 2. Illustration of the proposed MTT scheme.
We can see that when the number of active stations increases, they generate too much contention in one virtual transmission cycle. As a result, the average numbers of collisions and idle slots increase. If we assume there is a perfect scheduling algorithm, it does not have any collisions and idle slots and one successful transmitted packet is after another in a transmission cycle. Due to the unpredictable nature of traffic sources, it is unrealistic to find such an algorithm without overhead. However, we can try to maximize each station’s transmission by allowing them to transmit a packet burst, which can increase the portion of the successful transmission time in a virtual transmission cycle. The proposed scheme is based on the above idea and considers the fairness issue. Our goal is to have the winning station grab the channel for a longer period. Instead of wasting some time to assign a small back off timer for the last winning station in the standard IEEE 802.11 MAC, we give immediate access for that station after a SIFS interval. This will decrease the average number of idle slots for each contention period. As fairness is an important performance merit for a multi-access protocol [10], in order to restrict the absolute access of the winning station, a maximum transmission period is introduced for each station. If a station has some packets waiting in its transmission queue, they can be transmitted until the maximum transmission period is expired as illustrated in Figure 2. This method can provide fairness among different terminals. For example, a station running a voice application generally has small packets compared to a station transmitting large FTP data packets. By assigning a maximum transmission period to the voice applications, they are able to obtain more opportunities to share the broadcast channel. In IEEE 802.11 MAC protocol the stations have five distinct states: backoff state, transmission state, collision state, deferring state and idle state. Our proposed scheme preserves its simplicity for implementation in WLAN. The detailed algorithm is described below according to the states of the station:
r
r
r
Backoff state: Like in the standard IEEE 802.11 each station maintains a Network Allocation Vector (NAV) timer, which records the time that the channel will be controlled by other stations and it is updated by ACK, RTS/CTS or data packets. Each station first checks whether its Network Allocation Vector (NAV) has expired. If the NAV is expired and the channel has been idle for a DIFS time, the station will decrease its backoff timer (BT) by a slot time, i.e., BTnew = BTold − aSlotTime. When the BT of certain station reaches to zero, this station wins its transmission opportunity and starts to transmit a packet burst. To update the NAV in other stations, the station first checks its waiting queue and calculates its NAV. If the station has too many packets waiting and could not finish them in a Max Transmission Time (MTT), MTT is put into the DUR field. Otherwise the time needed to transmit the queuing packets is put into the DUR field, i.e. NAV = Min(MTT, Time to transmit all queuing packets). Successful transmission state: If a station has finished transmitting a packet and has been successfully indicated by receiving an ACK packet, it will check whether the sum of time spend on transmitting next packet and previous packets has exceeded the MTT. If the time is above the MTT, it means the station should finish its transmission opportunity in the current contention cycle and set its Contention Window(CW) to the minimum value, CW = minCW. A BT is randomly generated accordingly, BT = uniform(0, CW − 1) × aSlotTime, and the time spent on transmission packets is reinitialized to zero. Otherwise, the station waits for a SIFS interval and continues to transmit its next packet. Collision state: If a station fails to receive an ACK packet from the intended receiving station, it implies a collision happened. The CW is doubled and a new BT is chosen accordingly, CW = min(maxCW, CW × 2), BT = uniform(0, CW − 1) × aSlotTime. The time spent on transmission packets is reinitialized to zero.
WANG ET AL.
620
Figure 3. Morkov chain model for the backoff window size.
r
r
Deferring state: when a station is in a deferring state, it monitors the channel status. When it receives a packet from other stations, the station checks the DUR field in that packet and updates its NAV accordingly; when it monitors that the channel has been idle for a time equal to EIFS, the station calls its back-off procedure. Idle state: In the idle state, a station waits for receiving the packets generated by the upper layers.
4. The throughput analytical model In this session we will derive an analytical model for calculating the saturation throughput of the proposed scheme. Following the assumption used in [4] that at each transmission attempt the packet encounters collisions with a constant and independent probability p. Define W = C Wmin and let m represent the maximum backoff stage and we have C Wmax = 2m W. Also define Wi = 2i W, where i ∈ (0, m) is different backoff stages. Let s(t) be the stochastic process representing the backoff stage (0, . . . , m) and b(t) represent the backoff time counter for any given station at time t. The bidimensional process {s(t), b(t)} can be modelled by the discrete time Markov chain in Figure 3. The non-null one-step transition probabilities in
this Markov chain can be written as [4] P{i, k | i, k + 1} = 1 k ∈ (0, Wi − 2) i ∈ (0, m) P{0, k | i, 0} = (1 − p)/W k ∈ (0, W − 1) i ∈ (0, m) 0 0 P{i, k | i − 1, 0} = p/W k ∈ (0, W − 1) i ∈ (1, m) i i P{m, k | m, 0} = p/Wm k ∈ (0, Wm − 1). (1) Let bi,k = limt→∞ P{s(t) = i, b(t) = k}, i ∈ (0, m), k ∈ (0, Wi − 1) be the stationary distribution of the Markov chain. bi,k can be expressed as the function of b0,0 and the collision probability p. By using the normalization condition we can get [4] b0,0 =
2(1 − 2 p)(1 − p) . (1 − 2 p)(W + 1) + pW (1 − (2 p)m )
(2)
The probability that a station transmits packets in a random chosen slot time can be expressed as τ =
m i=0
=
bi,0 =
b0,0 1− p
2(1 − 2 p) . (1 − 2 p)(W + 1) + pW (1 − (2 p)m )
(3)
The collision occurs when more than one station transmit packets at a given time slots. Therefore, the collision
621
DISTRIBUTED MAC SCHEME FOR ENHANCING WIRELESS LAN PERFORMANCE
probability p can be written as
as
p = 1 − (1 − τ )n−1 .
(4)
Equations (3) and (4) represent a nonlinear system including two unknown variables τ and p, which can be solved using the numerical method. The normalized throughput S is the ratio of the average time spent on successfully transmitting payload information and the M T T transmission cycle time S=
E[payload transmitted in a M T T ] . E[length of a MTT time]
(5)
Define Ptr the probability that there is at least one transmission in the considered slot time. Since each station transmits with probability τ , Ptr can be given by Ptr = 1 − (1 − τ )n .
(6)
Let Ps be the probability that exactly the transmission of one station occurring on the channel is successful in a given time slot, conditioned on the fact that at least one station transmits packets, i.e., Ps = =
nτ (1 − τ )n−1 Ptr nτ (1 − τ )n−1 . 1 − (1 − τ )n
(7)
Note that average amount of payload information successfully transmitted in a virtual transmission cycle is Ps Ptr k E[P], given that a successful transmission occurs in a MTT cycle with probability Ps Ptr . Here k is defined as the number of packets transmitted in each MTT cycle. It can be obtained from the constraint that successful transmission time should not exceed the MTT value MTT ≥ k(H + E[P] + SIFS + δ + ACK) + DIFS + δ. (8) The average length of a virtual transmission cycle is readily obtained from the fact that, with probability 1− Ptr , the system is backing off with idle slots; with probability Ps Ptr it contains a successful transmission, and with probability Ptr (1 − Ps ) it contains a collision. Therefore, Equation 5 can be expressed
S=
Ps Ptr k E[P] , (1 − Ptr )σ + Ps Ptr Ts + Ptr (1 − Ps )Tc
(9)
where σ is the duration of an empty slot time, Ts is the average time the channel is sensed busy (i.e., the slot time lasts) because of the successful transmissions, and Tc is the average time that the channel is sensed busy by each station during a collision. Let the packet header be H = PHY hdr + MAChdr , the propagation delay be δ and E[P ∗ ] be the average length of the longest packet payload involved in a collision. As shown in Figure 4, in the MTT method we can obtain Ts and Tc as Ts = k(H + E[P] + SIFS + δ + ACK) + DIFS + δ (10) Tc = H + E[P ∗ ] + ACKtimeout + DIFS + δ.
5. Performance evaluation This section investigates the performance merits of the proposed new MAC scheme compared to those of the standard IEEE 802.11 DCF MAC using simulation experiments. We use the network simulator OPNET [11] which has the functionality of the standard DCF and can model the realistic wireless transmission environment by providing some buildin components such as background noise generator. For the purpose of this study, we extended the simulator by implementing the proposed new MAC scheme in order to model and simulate its behavior. As discussed in Section 2, we are interested in the system performance under the saturation condition and assume each station has a saturation source. This means stations immediately have a packet available for transmission after the completion of each successful transmission. The values of other parameters used to obtain numerical results, for both the analytical model and the simulations are summarized in Table 1. They specify the parameter used for the frequency hopping spread spectrum (FHSS) PHY layer and the IEEE 802.11 MAC [7]. The validity of the simulator has been demonstrated by comparing simulation results to those obtained through the analytical model in Section 5.1.
Figure 4. Ts and Tc in a virtual transmission cycle.
WANG ET AL.
622 Table 1 Simulation parameters from FHSS PHY layer. Slot time CWmin CWmax SIFS DIFS Propagation Delay Channel Bit Rate ACK Timeout MAC header PHY header ACK
50 µs 16 1024 28 µs 128 µs 1 µs 1 Mbit/s 300 µs 272 bits 128 bits 112 bits + PHY header
r
r
scribes the average queuing time for which data packets have to wait at the MAC layer. End to end delay: it records the time elapsed between the arrival of a data packet at the source station MAC layer and the reception of the packer at the destination station MAC layer. It includes media access delay, propagation delay and transmission time. Fairness index: It is designed to describe the fairness degree among the active stations to occupy the wireless channel. It is defined as: N 2 i=1 Ti FairnessIndex = N (11) N i=1 Ti2 where N is the number of stations, Ti is the throughput of station i. The higher the fairness index (up to 1) the better fairness the system can gain.
Figure 5. Saturation throughput: analysis versus simulation.
5.1. Model validation Figure 5 shows the analytical and simulation results of the normalized saturation throughput for the MTT method. Each simulation experiment was run until the system reached its stable state. Symbol “×” represents the mean value of the normalized saturation throughput obtained from simulation experiments. All simulation results in the plot are obtained with a 95% confidence interval lower than 0.02, which is represented by a bar in the graph. The figure reveals that the analytical results match those obtained from the simulation.
5.2.1. The effect of the new scheme In order to evaluate the effect of the proposed new MAC scheme, we examine the system performance under various load conditions by increasing the number of stations. The packet size follows an geometric distribution with the mean 1000 bytes. We have noted that the performance results reached here are similar when other sizes of packet are considered. The MTT is set to 8000µ seconds which is roughly enough for a station to transmit a 1000 bytes packet. The impact of MTT value is further investigated in next section. Figures 6–9 depict the results of normalized throughput, end to end delay, media access delay and fairness index versus the number of mobile stations in the system, respectively. For the purpose of comparison, the performance measurements for the standard IEEE 802.11 DCF MAC are also included. As a general trend, all figures show that the system performance degrades as the number of stations increases; i.e., the throughput and fairness index decrease and the delay increases. It is worth noting that all figures reveal the fact that the new MCA scheme
5.2. MTT analysis To investigate the performance of the MTT scheme, we focus on the evaluation of four fundamental system measurements: normalized throughput, end to end delay, media access delay and fairness index, which are presented as follows.
r
r
Normalized throughput: this typical merit describes the actual percentage of channel capacity used to deliver the data packets. It is calculated as throughput divided by the channel capacity. The higher normalized throughput the more efficient the MAC protocol is. Media access delay: it is defined as the time that the data packets experience at the MAC layer before being successfully transmitted through the wireless medium. It de-
Figure 6. Comparison of normalized saturation throughput between MTT and DCF.
623
DISTRIBUTED MAC SCHEME FOR ENHANCING WIRELESS LAN PERFORMANCE
Figure 7. Comparison of end to end delay between MTT and DCF.
Figure 10. Normalized throughput versus max transmission time.
performance by allowing stations to have longer transmission time in each contention cycle and consequently decreasing the collision probability. The advantage becomes more significant when the systems hold more stations. Figures 7–8 reveal that both end to end delay and media access delay are lower for the proposed MAC scheme compared to those for the standard IEEE 802.11 DCF MAC. Finally, Figure 9 shows that the proposed method can provide better fairness as we expected.
Figure 8. Comparison of media access delay between MTT and DCF.
5.2.2. The impact of the MTT value The value of the Max Transmission Time (MTT) plays a critical role on the performance of the proposed MAC scheme. An important and interesting question is what the optimum MTT value should be chosen in the real-world implementations. In an effort to answer this question, we vary the MTT value in simulation experiments to evaluate its impact on the system performance. As we can see in Figure 10, the normalized throughput increases as the MTT value becomes bigger. But Figure 11 shows that the fairness index drops as the MTT value increases. Under the extreme situation, the maximum throughput can be obtained by setting the MTT infinity when
Figure 9. Comparison of fairness index between MTT and DCF.
is able to provide superior performance than the standard IEEE 802.11 DCF MAC. For instance, Figure 6 shows that the normalized throughput supported by the IEEE 802.11 DCF protocol drops sharply when the number of stations is large. This is because the collision probability becomes higher as more stations compete for the shared broadcast channel. However, the proposed bursty transmission method improves network
Figure 11. Fairness index versus max transmission time.
624
the station has immediately packets waiting for transmission as soon as the completion of each successful transmission. However, for this case the fairness index reaches the worst value 1/N . N 2 Ti2 1 i=1 Ti FairnessIndex = N = = (12) 2 2 N N × Ti N i=1 Ti Figures 10 and 11 reveal an important performance result that a trade-off of the MTT value is required for maintaining network performance at a good level.
6. Conclusions WLANs are becoming increasingly prevalent for providing wireless packet service due to their desirable properties, such as mobility, installation speed, simplicity, scalability. An ideal MAC protocol for WLAN aims to provide an efficient mechanism to share limited spectrum resources with high throughput, low packet delay and good fairness among all stations. This study has presented a new scheme for the contention-based MAC protocols in WLAN. This scheme allows each station to transmit a burst of packets after winning a transmission opportunity instead of just one with the aim of reducing the average waiting slots and collision probability in each transmission cycle. Moreover, to ensure fairness among stations, a maximum transmission period is assigned to each station for controlling the length of the bursty transmission. An analytical performance model has been derived for calculating the normalized throughput of the proposed scheme. The effectiveness of the proposed new scheme is demonstrated by comparing the relative performance merits of the new scheme with the standard IEEE 802.11 DCF MAC. The extensive simulation experiments have revealed that the proposed scheme can achieve high throughput, low delay and high degree of fairness compared to the 802.11 DCF protocol while preserving the implementation simplicity.
References [1] N. Abramson, Development of the Alohanet, IEEE Transactions on Information Theory 31 (1985) 119–123. [2] N. Abramson, Multiple access in wireless digital networks, Proceedings of the IEEE, 82 (1994) 1360–1370. [3] D. Bertsekas and R. Gallager, Data Networks (Prentice-Hall International, London, 1992). [4] G. Bianchi, Performance analysis of the IEEE 802.11 distributed coordination function, IEEE Journal on Selected Areas in Communications 18 (2000) 535–547. [5] F. Cali, M. Conti, and E. Gregori, Dynamic tuning of the IEEE 802.11 protocol to achieve a theoretical throughput limit, IEEE/ACM Transactions on Networking 8 (2000) 785–799. [6] A. Chandra, V. Gummalla and J.O. Limb, Wireless medium access control protocols, IEEE Communications Surveys, Second Quarter 2000.
WANG ET AL.
[7] IEEE, Wireless LAN medium access control (MAC) and physical layer (PHY) specifications, IEEE Standards, 1999. [8] L. Kleinrock, Queueing Systems (Wiley-Interscience, New York, 1975) Vol. II. [9] L. Kleinrock and F. Tobagi, Packet switching in radio channels: Part I— Carrier sense multiple-access modes and their throughput-delay characteristics, IEEE Transactions on Communications 23 (1975) 1400– 1416. [10] C.E. Koksal, H. Kassab and H. Balakrishnan, An analysis of shortterm fairness in wireless media access protocols, in: Proceedings ACM SIGMETRICS 2000, Jun 17–21 2000, Santa Clara, CA, 2000. [11] OPNET Inc., “http://www.opnet.com.” [12] R. Rom and M. Sidi, Multiple Access Protocol: Performance and Analysis. (Springer-Verlag, New York, 1990). [13] S. Tasaka, Performance Analysis of Multiple Access Protocols (MIT Press, Cambridge, MA, London, 1986). [14] F. Tobagi and L. Kleinrock, Packet switching in radio channels: Part II— The hidden terminal problem in carrier sense multiple-access and the busy-tone solution, IEEE Transactions on Communications 23 (1975) 1417–1433. [15] N.H. Vaidya, P. Bahl and S. Gupta, Distributed fair scheduling in a wireless LAN, 6th Annual International Conference on Mobile Computing and Networking (MOBICOM 2000), Boston, MA, USA, Aug 6–11, 2000. [16] X.G. Wang, J.E. Mellor and K. Al-Begain, Discrete Event Simulation of the IEEE 802.11 MAC protocol,” Sixth United Kingdom Simulation Society Conference (UKSim2003), Emmanuel College, Cambridge, UK, 2003. [17] K. Younggoo, F. Yuguang and H. Latchman, A novel MAC protocol with fast collision resolution for wireless LANs, INFOCOM 2003. Twenty-Second Annual Joint Conference of the IEEE Computer and Communications Societies, IEEE, 2003.
Xin Gang Wang received his 1st B.Sc. degree in Computer Science from the Heilongjiang University, P.R.China, in 2001. He is currently a Ph.D. student in the computing department, University of Bradford. His research interests include performance modeling of the mobile networks. E-mail:
[email protected]
Geyong Min received the PhD degree in computing science from the University of Glasgow, United Kingdom, in 2003, and the BSc degree in computer science from Huazhong University of Science and Technology, China, in 1995. He is currently a lecturer in the Department of Computing at the University of Bradford, United Kingdom. His research interests include Performance Modelling/Evaluation, Parallel and Distributed Systems, Mobile Computing, Computer Networks, Multimedia Systems. Dr. Min is the founding co-chair of the International Workshop on Performance Modelling, Evaluation, and Optimisation of Parallel and Distributed Systems (PMEO-PDS) held in conjunction with IEEE/ACM-IPDPS. He is the guest editor of the journals Computation and Concurrency: Practice and Experience, Future Generation Computer Systems, and Supercomputing. He has served on the program committees of a number of international conferences. He is a member of the IEEE Computer Society. E-mail:
[email protected]
DISTRIBUTED MAC SCHEME FOR ENHANCING WIRELESS LAN PERFORMANCE
John Mellor has worked in the modelling and simulation of communication networks for 25 years. Early work included dynamic alternate routing and the application of learning automata to routing strategies in circuit and packet switched networks. Collaboration with a Cambridge UK company led to the development of a LAN protocol which consistently outperformed Ethernet. He was sent as a government expert to study the Manufacturing Messaging protocol in the USA and Japan. He later became a technical expert consultant on the application of European Directives within manufacturing industry. A forray into radio frequency identification tags resulted in the development of a novel protocol that was exploited by a major vehicle component manufacturer. He now finds himself involved in wireless protocols with researchers working on WiFi (802.11) and on security aspects of mobile commerce. John is leader of the Mobile Computing and
625 Networks Resaerch Group at the University of Bradford and course tutor to three innovative advanced MSc. courses in mobile computing, applications and security. E-mail:
[email protected]
Lin Guan received the B.Sc degree in computer science from Heilongjiang University, Heilongjiang, China, in 2001. She is currently a PhD student in University of Bradford. Her research interests focus on developing cost effective analytical models for the performance evaluation of congestion control algorithms for Internet traffic. E-mail:
[email protected]