Nonlinear autoregressive neural network with exogenous input for an energy efficient non-cooperative target tracking in wireless sensor network

Abstract

The prediction algorithms have been studied as a part of target tracking applications for many years. The prediction algorithm helps to select appropriate nodes to achieve precise target locations while tracking. The only group of sensor nodes nearer the predicted location is activated to save network energy. The inaccurate prediction algorithm may hamper energy consumption by activating inappropriate nodes resulting in a target loss. We propose a nonlinear autoregressive neural network with exogenous input (NARX)-based target-tracking algorithm that improves tracking accuracy and energy efficiency. The proposed algorithm uses vehicle location time series and exogenous vehicle velocity time series as inputs and exerts accurate prediction location for given non-cooperative manoeuvring targets. The proposed algorithm is evaluated in terms of average prediction error, total network energy used, and the count of a target loss with state of art. The experiment outcome proves that the proposed novel NARX-based tracking algorithm outperforms and saves up to 26% of network energy with up to 83% reduction in tracking error compared to existing target tracking algorithms.