Principle Component Analysis-Gradient Descent localization algorithm for Wireless Sensor Network

Abstract

Due to the latest progress in wireless communication, wireless sensor nodes (WSN) are being used in many real world applications. To find the location of WSN, when we have only proximity information between the nodes, is the difficult task. However, we have massive number of localization algorithms, each of them using different approach to get the final location. For robust fashion and noise sparse network, Principle component analysis (PCA), is one of that techniques which work good even with less number of anchor nodes. The proposed work addresses the problem of location discovery of nodes based on the combination of PCA and Gradient Descent (GD) method. In this paper we use PCA to find the initial estimated position of nodes and proposed GD as a refinement algorithm for final position.