A benchmark for channel assignment algorithms in wireless testbeds

Channel assignment for multi-radio wireless mesh networks has been proven efficient to increase the network performance by decreasing the interference of simultaneous transmissions. Many algorithms have already been proposed, however, a meaningful comparison of their performance is difficult. The main reasons are that different experimentation environments and performance metrics are used for their evaluation. Thus, finding a universal methodology for the performance evaluation that ensures comparability is complicated. In this paper, we close this gap with a methodology for the performance evaluation of channel assignment algorithms in wireless testbeds. We developed domain-specific performance metrics that express the decrease of network-wide interference. Using these metrics, we developed benchmarking scenarios for performance measurements in wireless testbeds. We present the benchmarking results of a distributed link-based channel assignment algorithm and compare the results to a random-based algorithm and a single channel network. The evaluation was run on the DES-Testbed, a 128 node multi-radio testbed at the Freie Universität Berlin.