Three-dimensional WSN Node Localization Method Based on LSSVR Optimized by DE Algorithm

Abstract

Aiming at the problem that the three-dimensional (3D) node localization precision is affected by the model parameters of least squares support vector regression (LSSVR) node localization method for wireless sensor network, a localization method based on LSSVR optimized by differential evolution (DE) algorithm is proposed in this paper. Firstly, the 3D node localization model is built through LSSVR and the kernel function parameter and the regularization parameter are optimized by DE algorithm. Then, the fitness function of DE algorithm is constructed according to the mean square error of a number of virtual nodes from the predicted position and their actual position, and the global optimal parameters of LSSVR are acquired through limited modeling parameters iterative searching method. Finally, the LSSVR optimized by DE algorithm is used to realize the node localization. The simulation results show that the localization accuracy of the proposed method is superior to that of least square (LS) and LSSVR method. Keywords: Differential evolution algorithm, least square support vector regression, three-dimensional node localization, wireless sensor network.