Studies On The Performance, Capacity, And Resource Allocation Strategy In IEEE 802.11e Networks

Wireless communication systems have been successfully implemented worldwide. Various applications are supported by different kinds of wireless networks such as celluer networks, wireless local area networks, and wireless ad hoc networks. As a result of the evolution of modern communication techniques, mobile packet data service and mobile multimedia service become the main trends of the future development of wireless communication systems. Compared with wired networks, wireless networks provide less quality-of-service (QoS) guarantees to multimedia services. This deficiency becomes an obstacle for the development of wireless communication systems. To solve this problem, it is necessary to enhance and improve the performance of MAC layer protocols, which define methods of sharing the wirless medium and have critical influence on the performance of wireless networks.IEEE 802.11 is the de facto standard in wireless local area networks and is regarded as the principle MAC protocol in both ad hoc networks and wireless sensor networks. 802.11e is proposed to support QoS in future wireless networks. Thus, improving and optimizing the performance of 802.11e has great practical importance. With the trendency of increasing applications and growing network scale, it is critical to measure, improve, and optimize the performance of 802.11e and effectively to conduct resource management for providing QoS guarantee. Modeling network protocols is a main method in solving those problems. This thesis analyzes the peroformance of IEEE 802.11e and studies related problems such as the influence of both protocol parameters and medium access procedure on MAC performance and how to optimize the MAC protocol and allocate bandwidth among different access categories.We study the performance of IEEE 802.11e EDCA in both saturated and non-saturated condition, and propose two analytical models. The performance model of EDCA in saturated networks provides basis for configrating protocol parameter, optimizing protocol performance and allocating wireless bandwidth. Compared with previous models, the presented saturation model gives a clear and more accurate depiction of 802.11e service differentiation mechanisms including contention window (CW), arbitration inter-frame space (AIFS) and internal collision handler mechanism. Based on the analytical model in saturated networks, we analyze the performance of IEEE 802.11e under different traffic loads. The variation of performance metrics such as throughput and delay at different phase is studied. Different from previous studies which assume all access categories have the same packet length, we study the senario where different access category has different packet length. Extensive simulation and analysis results are conducted to validate the proposed models.With the proposed performance analytical model, we discuss the maximal protocol capacity of IEEE 802.11e. We prove that there exists an optimal operating point in IEEE 802.11e. The optimal operating point can be expressed in terms of meidum idle probability. By properly choosing protocol parameters, the wireless bandwidth can be efficiently utilized. We also study how to achieve proportion weighted fairness by selecting parameters. It is proved that with the proposed paramter configuration algorithms the maximal protocol capacity can be achieved under a fairness constraint.To provide QoS guarantee, it is necessary to allocate bandwidth among all access categories with bandwidth requirement constraint. Bandwidth allocation is related with bandwidth prediction and call admission control, which are important parts of QoS guarantee mechanisms. Previous studies focus in saturated networks. However, networks can be saturated, semi-saturated, or non-saturated due to different traffic loads. Different network states increase the difficulty of bandwidth prediction. Furthermore, how to determine the bandwidth loss due to both collisions and medium idle is also important for bandwidth prediction. This thesis presents a bandwidth prediction algorithm for IEEE 802.11e based on the saturated performance model and maximal protocol capacity. The state of each node can be determined by comparing its arriving traffic loads with its throughput. With the maximal protocol capacity and optimal operation point of 802.11e network, we determine the bandwidth loss due to collision and idle medium, and simplify the bandwidth prediction algorithm by setting the network under the optimal operation point. With the proposed prediction method, the corresponding call admission control algorithm and rate control algorithm are also presented.