| Wireless sensor nodes are symbiotic when deployed in an activity region and heavily rely on each other for successful transmission of data. Therefore, failure of some nodes can possibly partition the network. Since unmanned operation is one of the primary properties of these nodes, there needs to be a way in which these nodes can collectively maintain connectivity and resolve any reachability problem that may be faced when a node fails. Different approaches have been proposed in the literature to address this issue. However, most published approaches are based on a single underlying principle of trying to replace the failed node without considering the possible fact that the location of the failed node might not have been optimal and which could have been the reason for its failure. These approaches also tend to trigger a cascaded relocation of many nodes resulting in increased overhead. This thesis presents a novel solution that pursues rearrangement of nodes to restore connectivity while limiting the scope of the recovery to the vicinity of the failed node. Whenever a node fails, its neighbors reposition themselves to not only re-establish data paths but also ensure that they are connected in the best possible way. The connectivity restoration is modeled as a variant of the Steiner tree formation problem and solved using novel heuristics. The proposed approach is validated through simulation and is shown to outperform existing schemes. |
