Mobile node localization method based on KF-LSSVR algorithm

Abstract

Abstract As the mobile node localization algorithm in three-dimensional environment cannot meet the demand of actual application, a hybrid mobile node localization algorithm for a wireless sensor network (WSN) in three-dimensional environment is proposed in this paper, which is based on the least squares support vector regression (LSSVR) and Kalman filter (KF). The proposed algorithm firstly constructs the LSSVR localization model by sampling measurement area and training sample sets. Then, the KF model is used to iterate and correct the measured distance in order to obtain the distance between the unknown node and each anchor node. Finally, the LSSVR localization model is employed to obtain the estimated location of the unknown node. The experiments were conducted and the experimental results were analyzed according to ranging errors, anchor node density, communication radius, moving speed, and node localization errors. Simulation results show that the proposed algorithm using a joint KF and LSSVR algorithm is superior to the KF algorithm and the LSSVR algorithm, and it can reduce the localization errors and improve the localization accuracy.

Keywords

• Least squares support vector regression
• Kalman filter
• WSN
• Mobile node
• Three-dimensional localization