Experimental Evaluation of BMX6 Routing Metrics in a 802.11an Wireless-Community Mesh Network

Modeling the performance of Wireless Mesh Networks (WMN) is a task that has received a lot of attention and has produced a large body of related literature. Most of the times, such literature is based on analytic assumptions or computer simulations, and results not always match reality. In this paper we use data collected over a one-month period in a 50 nodes wireless community network in Barcelona to compare the experimental throughput we measured over multi-hop paths with the capacity estimated using the well known conflict graph model introduced in [1]. Our experiments show that the model generally overestimates the available capacity, despite the availability of precise information on the underlying network graph. We also use the data to test the performance of the BMX6 routing protocol that runs on the network nodes and show that in the large majority of cases the path chosen by BMX6 corresponds to the path with the highest available capacity, which is a key feature to guarantee the maximum exploitation of the network resources.