436 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 57, NO. 1,JANUARY 2008 Formulation of Distributed Coordination Function of IEEE 802.11 for Asynchronous Networks: Mixed Data Rate and Packet Size Mustafa Ergen and Pravin Varaiya, Life Fellow, IEEE Abstract—In carrier-sense-multiple-access/collision-avoidance networks in which stations have different data rates, some stations are penalized because slow stations receive more time to transmit. Thus, a single low data-rate station unfairly brings down the throughput of the high data-rate stations. We introduce a simple and standard-compliant algorithm to fairly utilize the channel. We first provide a formulation for the throughput with mixed data-rate connections. To alleviate the low performance of the high data-rate stations, we introduce a mechanism that implements an adaptive scheme to adjust the packet size according to the data rate. With this scheme, stations occupy the channel for equal amounts of time. We then extend the scheme to a frame- aggregation scheme to show how different packet sizes affect performance. Index Terms—Distributed coordination function, fairness, frame aggregation, IEEE 802.11, Markov model, QoS, wireless voice over Internet protocol (VoIP). I. I NTRODUCTION W IRELESS communications is evolving to a stage when each device will be a node in a mobile network with multiple interfaces. This also brings the growth of multimedia applications that impose requirements on communication para- meters [1]. As a result, wireless networking is moving toward asynchronous connectivity. We consider a scenario with several IEEE 802.11 [2]–[5] compliant nodes located near each other, which can transmit at different data rates and packet sizes. Each node may initiate packet transmission with variable physical transmission rates depending on its connection quality. Sometimes, different IEEE standards may share the same spectrum with the same access mechanism as we see with IEEE 802.11b [3] and 802.11g [5] networks. Similarly, a station’s packet size can also change during the connection, depending on the type of flow. We are interested in obtaining analytical formulas for throughput in such a scenario. We suppose that every flow of a station is Manuscript received March 16, 2006; revised July 28, 2006, December 12, 2006, and February 5, 2007. This work was supported in part by the ARO- MURI UCSC-WN11NF-05-1-0246-VA-09/05 and in part by National Semi- conductor. The review of this paper was coordinated by Prof. X. Shen. M. Ergen is with Wichorus, San Jose, CA 95134 USA. (e-mail: ergen@ eecs.berkeley.edu). P. Varaiya is with the Department of Electrical Engineering and Com- puter Sciences, University of California Berkeley, Berkeley, CA 94720 USA (e-mail: varaiya@eecs.berkeley.edu). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TVT.2007.901887 Fig. 1. Station’s data rates throughout the simulation. saturated, i.e., there is always a packet to transmit when a station has the right to transmit [6], [7]. Our approach begins by introducing the Markov chain model from [8], which, in turn, is an enhancement of Bianchi’s model [6]. The performance differences between these two models are discussed in the next section; however, our analytical formula- tion can also be applied to Bianchi’s model. Case Study: We simulate a scenario in OPNET in order to revisit the anomaly presented in [9]. The network has five stations with an 11-Mb/s data rate, except for station 1, which changes its rate over time, as shown in Fig. 1. The upper plot of Fig. 2 shows the total throughput of the network; the lower plot shows the activity in station 1’s channels. As shown, there are four channels, and each corresponds to one rate in IEEE 802.11b [3]. Station 1 severely impacts the network, as shown in the upper plot of Fig. 2. For example, between 100 and 200 s (and, again, between 400 and 600 s), station 1 with 1-Mb/s data rate decreases the total throughput by more than half. Fig. 3 shows the plots of two stations, from which one may infer that the total throughput is equally divided and that stations with higher data rate experience the same throughput as the slow station. Recent papers [9]–[12] have reported this behavior in the IEEE 802.11 carrier-sense-multiple-access/collision-avoidance (CSMA/CA) networks. 0018-9545/$25.00 © 2008 IEEE Authorized licensed use limited to: Univ of Calif Berkeley. Downloaded on March 11, 2009 at 19:52 from IEEE Xplore. Restrictions apply.