Metrics for Characterizing the Quality of Service of Multicommodity Flows in Multi-Channel Multi-Radio Multi-Rate Wireless Mesh Networks

— This paper proposes important performance metrics for characterizing the quality of service of multicommodity ows in multi-channel, multi-radio, multi-rate (MC-MR) wireless mesh networks. The term ”multicommodity” implies different types of applications like audio, video, text etc. being serviced simultaneously by the network. Traditionally, the capacity of the links in a mesh network is assumed to be deterministic. However, this is practically not the case as links are often subject to environmental variations and failures. In such cases, data communication can occur at variable rates and as a result providing proper QoS to applications becomes a challenge. For a particular multicommodity ow demand, we analytically model and evaluate the following three metrics. 1. The expected multicommodity ow achieved by the network 2. The expected multicommodity ow that is achieved by the network under given cost constraints and 3. The quantitative value of the importance of each link in the network with respect to the multicommodity ow demand. In addition, we also state the lower bounds for the rst two metrics. A multicommodity ow demand is represented by a rate (capacity) vector of k applications being simultaneously served by the network. Our methodologies for metric evaluation are based on concepts from the theory of probabilistic network ows and run in polynomial time. We corroborate our theory with simulations. Our metrics will be of great signi cance to network designers and application service providers who could use them to judge the suitability of applications for a given network and/or spend time and effort in maintaining certain high importance links for better QoS support. To the best of our knowledge, this is the rst work to address the performance of multicommodity ows in MC-MR wireless meshes.