Dynamic Service Discovery as an Optimization Problem in Wireless Sensor Networks

In a ubiquitous environment, e.g. smart home, a user is surrounded by a network of sensor nodes all around and also on his body or clothing. We modeled such a sensor network as a services network where the nodes exchange services for collaborating and smart decision taking. As the user moves around performing activities, the surrounding network also changes as wireless internode connections are made or broken. The challenge is to re-discover the network quickly and transparently to the user. We used a two-step approach. First, Proximal Neighborhood Discovery identified nodes that formed the network. Second, Optimal Service Discovery determined, for each such network node, who were the best service provider nodes from the same network. We modeled this as an optimization problem and solved using a new and efficient algorithm. We implemented the algorithm using nesC and simulated using TOSSIM interference-model. The results showed appreciable improvements over conventional approaches.