Estimation of Maximum Achievable End-to-End Throughput in IEEE 802.11 Based Wireless Mesh Networks

Wireless mesh networks can be quickly deployed in various situations to provide temporary to permanent wireless network coverage. To assess the feasibility and reliability of a given end-to-end communication need, it is essential for communication end points to accurately estimate their achievable end-to-end throughput. Several capacity, end-to-end throughput, and available bandwidth estimation techniques have been studied in the past for wired and wireless networks. The contention among wireless nodes arising due to the IEEE 802.11 medium access control protocol's channel access mechanism renders the estimation of such network attributes challenging in multi-hop networks. This paper evaluates one state-of-the-art capacity estimation approach for ad hoc wireless networks and shows that it in fact measures achievable throughput instead of capacity and its estimated achievable throughput is not realizable. An analysis of end-to-end delays of injected probe packets is presented to show the effects of medium access contention and network queuing on the delays and estimated achievable throughput subject to different network traffic patterns and multi-hop collisions. Based on the observations, an alternative less intrusive delay distribution based achievable throughput estimation solution is proposed. With ns-2 simulation, the scheme was shown to accurately estimate the achievable throughput under various cross traffic conditions.