Genetic Algorithms and Simulated Annealing optimization methods in wireless sensor networks localization using artificial neural networks

Abstract

Node localization in wireless sensor networks (WSNs) is important for applications such as military surveillance, environmental monitoring, robotics, and many others. The sensor motes used in this type of application present low-power and low-cost profile. Hence, they require methods that compute their positions using indirect information such as Received Signal Strength Indicator (RSSI). This work presents Genetic Algorithms and Simulated Annealing optimization methods applied (independently) in artificial neural networks (ANNs) aiming node localization in WSNs. The RSSI measurements were used as the ANNs inputs to localize the nodes. This receiver-based approach was tested using MATLAB and Probabilistic Wireless Network Simulator (Prowler) to collect the ANNs input data, under simulated static indoor network environment. Results using the best ANN structure found after optimization using GA had a root mean square error of 0.39 meter against the 0.61 meter reached through the SA algorithm.