A Parametric Study on Gauss-Newton Iterations for Wireless Sensor Network Localization

High performance computing has opened up the possibility of increased accuracy in mobile devices such as wireless sensor networks, thanks to increased availability of processing power and network access. In spite of the advances made, the low power footprint acts as a constraint on which is a challenge. In order to address this challenge, optimization of the algorithm can provide a solution. In this study, we present a wireless sensor network and demonstrate how Gauss-Newton algorithm iterations can be balanced for optimum accuracy and lower computational effort towards a final goal of improving localization accuracy. Special attention is made to the effect of network size and Gaussian error on the Root Mean Square Error (RMSE) and computational time with respect to differing iterations. The results can be utilized in order to make better use of low speed parallel and distributed microprocessors for high performance wireless sensor network localization.