Computationally efficient routing for large scale wireless sensor networks

Routing in a wireless sensor network field requires significant resources. Computing optimal routes using graph theoretic algorithms becomes intractable as number of nodes become very large. In this paper, authors propose an alternative using a continuous function approximation for path cost. It is based on a fluid type approximation in which the whole network is replaced by a continuum plain - where the discrete graph describing the links and their cost is replaced by a cost density over the plain. The fluid limit approach does not depend on the number of nodes and hence the complexity of finding optimal routes does not grow with the number of nodes. The approach makes use of calculus of variations. Complexity analysis shows that this leads to a large saving in communication and computational overhead associated with routing. Two practical situations for possible application of this approach are identified. The computed path cost can also serve as the theoretical lower bound of cost for paths that can be computed by any routing algorithm. Preliminary simulation results are also reported.