Performance Modeling and Analysis of Wireless Local Area Networks with Bursty Traffic
Thesis or dissertation
University of Exeter
The explosive increase in the use of mobile digital devices has posed great challenges in the design and implementation of Wireless Local Area Networks (WLANs). Ever-increasing demands for high-speed and ubiquitous digital communication have made WLANs an essential feature of everyday life. With audio and video forming the highest percentage of traffic generated by multimedia applications, a huge demand is placed for high speed WLANs that provide high Quality-of-Service (QoS) and can satisfy end user’s needs at a relatively low cost. Providing video and audio contents to end users at a satisfactory level with various channel quality and current battery capacities requires thorough studies on the properties of such traffic. In this regard, Medium Access Control (MAC) protocol of the 802.11 standard plays a vital role in the management and coordination of shared channel access and data transmission. Therefore, this research focuses on developing new efficient analytical models that evaluate the performance of WLANs and the MAC protocol in the presence of bursty, correlated and heterogeneous multimedia traffic using Batch Markovian Arrival Process (BMAP). BMAP can model the correlation between different packet size distributions and traffic rates while accurately modelling aggregated traffic which often possesses negative statistical properties. The research starts with developing an accurate traffic generator using BMAP to capture the existing correlations in multimedia traffics. For validation, the developed traffic generator is used as an arrival process to a queueing model and is analyzed based on average queue length and mean waiting time. The performance of BMAP/M/1 queue is studied under various number of states and maximum batch sizes of BMAP. The results clearly indicate that any increase in the number of states of the underlying Markov Chain of BMAP or maximum batch size, lead to higher burstiness and correlation of the arrival process, prompting the speed of the queue towards saturation. The developed traffic generator is then used to model traffic sources in IEEE 802.11 WLANs, measuring important QoS metrics of throughput, end-to-end delay, frame loss probability and energy consumption. Performance comparisons are conducted on WLANs under the influence of multimedia traffics modelled as BMAP, Markov Modulated Poisson Process and Poisson Process. The results clearly indicate that bursty traffics generated by BMAP demote network performance faster than other traffic sources under moderate to high loads. The model is also used to study WLANs with unsaturated, heterogeneous and bursty traffic sources. The effects of traffic load and network size on the performance of WLANs are investigated to demonstrate the importance of burstiness and heterogeneity of traffic on accurate evaluation of MAC protocol in wireless multimedia networks. The results of the thesis highlight the importance of taking into account the true characteristics of multimedia traffics for accurate evaluation of the MAC protocol in the design and analysis of wireless multimedia networks and technologies.
PhD in Computer Science