Analytical modeling of medium access control in finite-load and saturated wireless LANs

The IEEE 802.11 standard provides the specifications for a Wireless Local-Area Network (WLAN) technology, commonly known as Wi-Fi. IEEE 802.11 uses an Ethernet-like contention-window mechanism to resolve the multiple-access problem to the wireless channel. Essentially, a wireless station doubles its contention window size after detecting a frame collision. Although this method is effective in reducing collisions on the channel, it increases packet overhead which results in reducing the throughput. In general, the performance of the WLAN varies widely depending on the number of customers in the network coverage area and the shape of the traffic on the channel. A few analytical models have been proposed over the past few years to understand the behavior of WLANs. Although insightful, most of these models were based on a highly-simplified "saturated" networks model in which all wireless stations behave like traffic source with infinite number of frames to transmit. The saturation assumption is useful in that it leads to simple steady-state models with fixed transition probabilities. However, it is not realistic to assume that wireless stations are always attempting to transmit frames, in real networks. This thesis is concerned with the development of improved analytical models for nonsaturated, or finite-load, WLANs and also propose enhancements to existing saturation WLAN models. In particular, we have developed two analytical models for WLANs with finite load. One model for the standard IEEE 802.11 WLAN and the second for the quality-of-service enabled WLAN described by the IEEE 802.11e standard. The proposed models capture the probabilistic nature of wireless networks and the interdependencies among wireless stations. In our analysis, we rely on novel schemes that use coupled station-view and network-view models to compute the overall throughput, collision probability, and delay in a WLAN. When compared to other recent work, our results prove to be the most accurate. We complement our work on finite-load models by presenting a more accurate model for IEEE 802.11e WLANs operating under saturation, and propose a few adaptive contention-window algorithms for maximizing the frame transmission rate and consequently the saturation throughput. We show through simulation that the proposed algorithms increase the throughput by several multiples.Applied Science, Faculty ofElectrical and Computer Engineering, Department ofGraduat