Data aggregation in multihop wireless mesh sensor Neural Networks

Abstract

Wireless sensor networks have been found to be useful for detection of events like volcanic eruption, landslide, and agricultural crop stress. The problem of in-network data aggregation for binary event detection has been well studied in literature for multihop wireless sensor networks with tree topology. In this paper, we propose a novel Neural Network based distributed detection approach for multihop wireless sensor networks with mesh topology. As compared to the tree topology, the unidirectional mesh topology resembles more closely to a neural network wherein each sensor node has been modeled as a neuron in the neural network. An exhaustive analysis has been presented to compare the proposed approach with the existing approaches for tree topology along with their modified counterparts for the mesh topology. The simulation results depicts a substantial gain in the detection accuracy by the proposed neural network based data aggregation approach in wireless mesh sensor networks as compared to the other existing methods for tree and mesh topologies of the wireless sensor networks.