Component Based Self-Healing Approach for Fault-Tolerant Data Aggregation in WSN

Abstract

Node faults in WSNs are often attributed to battery energy drain outs. While data aggregation is widely acknowledged as an energy conservation tool; the aggregator nodes are susceptible to battery energy drain out faults. When an aggregator/ intermediary node in a data aggregation tree fails, the child nodes of the faulty node get detached from the aggregation tree and form disconnected nodes, termed as Affected Nodes (ANs) and/or subtrees. To tolerate such faults, we use the concepts of component-based graph theory and consider the subtrees as components that need to be connected back to the root-component tree. Based on this, we develop two component based self-healing fault-tolerant algorithms SCR and SCR-DTRA, that exploit the inherent redundancy in WSNs to determine alternate paths from the affected components to the root node. Unlike most other approaches, the hallmark of the component-based algorithms is that they preserve precedence relations in the data aggregation hierarchy. The results of our simulations show that the two algorithms succeed in the recovery of more than 90% of ANs and subtrees affected due to faults. In terms of longevity, the two algorithms, last more than twice the number of rounds compared to other algorithms used for comparison.