Topology Optimization and Network Deployment Algorithm in WSNs for Mobile Agent-based Applications

A Wireless Sensor Network (WSN) is typically depicted as a distributed resource constrained platform for supporting novel pervasive applications. A possible approach is to develop applications over WSNs by resorting to mobile agent frameworks that are aimed to minimize resource costs and memory allocation. In this context, network topology is a major issue to be considered for resource optimization: for example, assuming that the underlying network topology for monitoring applications is a spanning tree, it is straightforward to see that converge-cast data flow patterns do not match with the any-cast requirement for agent mobility. Optimal spanning trees are expected to be short, balanced with a uniform density of agent instances. This paper focuses on the definition of a cost function, the related metrics and the supporting protocol to setup optimized spanning trees. The optimization procedure is able to account also for typical topology rearrangements induced by node faults in a WSN. Simulations have been performed and encouraging performance results are reported.