A case for mobility- and traffic-driven routing algorithms for wireless access mesh networks

We present an approach for developing routing algorithms where we first attempt to understand the network characteristics (i.e. network connectivity, mobility and rate of change of the topology) and expected traffic patterns of a specified mobile scenario before proceeding to design an algorithm that optimizing routing performance for the common cases. We apply our approach to Triton, a proposed 802.16-based (WiMax) maritime wireless access mesh network. Our trace-based analysis shows that in our environment, if stationary nodes are preferred in the route selection, the rate of change for the routes to the gateway nodes is reduced by 23.3% and the average duration for which routes between a node and a gateway remains valid is increased by 31%. We also show that it is important to take the expected traffic patterns into account when designing routing for a specific system. By ignoring changes to the network topology that do not affect the anticipated traffic, we can significantly reduce the overheads associated with the repair of the routing tables.