Evaluating Metrics for Optimal Path Selection in Large Wireless Community Networks

Abstract

Routing metrics commonly used nowadays in large wireless community networks (WCN), do not seem appropriate since they are generally not based on the real link capacity but on choosing the shortest path. Routing protocols such as OSPF or BGP use static metrics, while wireless links should be evaluated frequently because their bandwidth varies dynamically. Ad hoc routing protocols use more complex metrics, however they are better suited for sporadic or mobile nodes than for static nodes placed in house roofs to create the backbone infrastructure. In this work, we compare several metrics which might be of interest in realistic WCN scenarios. To do this, we have performed routing studies over real topologies extracted from a specific WCN, namely “guifi.net”. The routes obtained using different metrics are compared in terms of length, capacity, and alternative paths. We show that, when using metrics based on real link bandwidth, the amount of routes with low bandwidth could be halved and the capacity of the network could be increased up to a 25%.