Performance Evaluation of Routing Metrics for Community Wireless Mesh Networks

With the growth of different types of Internet traffic there is a compelling need to provide better quality of service, especially, over the increasing number of wireless networks. Expected Transmission Count (ETX) is a high throughput route selection metric that measures link loss ratios. ETX of a path reflects the total number of packet transmissions (including retransmission) required to successfully deliver a data packet along that path. Expected Transmission Time (ETT) is an improvement of ETX. ETT of a path is a measure of the transmission time needed to successfully deliver a packet along the path. ETT measures the loss ratio and the bandwidth of the link. Both, ETX and ETT, in comparison to hop count, provide better route selection for routing protocols widely used in Wireless Mesh Networks (WMNs). Using minimum hop count to find the shortest path has been shown to be inadequate for WMNs, as the selected routes often include the weakest links. This thesis presents a performance evaluation comparing hop count, ETX and ETT when used with the Optimized Link State Routing version 2 (OLSRv2) protocol. This study is based on the wireless mesh topology of a suburban residential area in New Zealand, and analyses the performance of three common Internet traffic types in terms of throughput, end-to-end delay, jitter and packet loss ratio, and presents findings that are closer to the perspective of what an enduser experiences. Also, a grid network of 121 nodes was used to analyze how the metrics choose paths, the performance changes (for different path lengths) and other conditions that affect the performance of the three metrics.