Abstract: Wireless Sensor Networks (WSNs) have the prospect to become the most crucial technology of the future. Based on the applications, there is a need to locate the physical location of sensor node to improve the performance. This is known as localization problem. Some traditional localization algorithms are used but still convergence problem exists. So, to solve the above problems and obtain an efficient location identification, a system has been designed using machine learning and swarm intelligence. In this research, a Relevance Vector Machine (RVM) with Glow-worm Swarm behaviour based optimization Algorithm (GSA) is proposed for efficient localization. Here, the trilateration, triangulation and Maximum Likelihood (ML) based location discovery process is focused. For high accurate localization, the proposed system considers the node density factor. In this process, the node is in the overlapping region of circles considered as trilateration problem and it is solved by RVM. The RVM is mainly used for splitting the anchor and overlapping region node and similarly to find the weight for those nodes, so that, the processing time is reduced. After finding the innermost intersection of a point, the GSA is used to update the archive based on the distance and geometric topology constraints. The evaluation of proposed RVM-GSA localization is compared with Average Weight Based Centroid Localization (AWBCL) algorithm with the help of MATLAB tool. The obtained result shows that the proposed RVM-GSA algorithm is a promising scheme that can minimize the localization problem.
M. Arun and P. Manimegalai, 2018. An Efficient Localization based on Relevance Vector Machine with Glow-Worm Swarm Optimization for Wireless Sensor Networks. Journal of Engineering and Applied Sciences, 13: 406-414.