Distributed Scheme for Localization in Wireless Sensor Networks

Abstract

e proliferation of WSNs has been empowered by the recent advancement in Micro-Electro- Mechanical-Systems (MEMs). e applications of WSNs are enormously used to perform monitoring and sensing tasks in di erent environments like underground mines, healthcare, agriculture and smart cities. In all such applications the monitored data is signi cant only when the position of sensor node is known. Localization in WSNs is the methodology of nding geographical position of sensor nodes and it is one of the key research challenge of this eld. Geographical Positioning System (GPS) is a very straight forward solution to this problem but installing hardware of GPS in every node is not reasonable. Di erent algorithms of localization in WSNs have been introduced in literature. Generally these algorithms are classi ed in two categories: Centralized and Distributed. e thesis reviews di erent approaches of localization. An a empt is made to address localization of ad-hoc static sensor nodes. For this a distributed localization scheme is introduced that enables sensor nodes to accurately estimate their locations. Proposed algorithm uses distance information between sensor nodes and anchors. Anchors are special nodes which already know their location in advance. Anchors help other sensor nodes in their location discovery. An unconstrained non-linear objective function on measurement of errors in such scenarios is de ned. Using this unconstrained non-linear objective function, the problem of localization of sensor nodes is formulated as an unconstrained optimization problem. is unconstrained optimization problem is solved using Levenberg-Marquardt optimization algorithm which always progresses in a descent direction in order to nd local minima. e results of this technique are illustrated to con rm its advantages and these results are also compared with two of the previous techniques named localization using multi dimensional scaling and localization using neural networks.