A Robust Localization Mechanism Using Multiple Measurements in Wireless Sensor Networks

Abstract

Localization has been one of the studies of deep learning research in WSN (Wireless Sensor Network) since its development and it is still one of the vibrant scientific research due to its potential for widely used applications. It offers unanimous services in the field of IT, Robotics, Logistics and so on. Moreover, collaboration of distributed Robotics and WSN has led to development of MSN (Mobile Sensor Network). This thesis develops the noise, high uncertainty aware and error optimization localization model moreover, the mechanism is presented and evaluated in two parts; first part of the thesis develops a DNMAL model which goals to reduce the localization error considering various device density and noise levels. Furthermore, DNMAL develops an iterative updation mechanism to minimize the LSE (Least Square Error). DNMAL observes the fast convergence which brings the fair tradeoff among computation time and localization error. Proposed DNMAL is robust in nature and thus non sensitive for initial measurement. Performance Analysis of DNMAL shows that it observes marginal error optimization in comparison with the other existing model. Furthermore, in second part of the work, this thesis presents High Uncertainty Aware-Localization Error Correction and Optimization (HUA-LECO) Model for WSNs. First of all, the foremost goal is the estimation of the mobile sensor nodes by accurately predicting their position based on the Root Mean Square Error (RMSE) when there is high uncertainty regarding the position of the sensor nodes. The second aim is the optimization of uncertainty-aware errors to precisely evaluate the location of sensor devices. Here, mobile sensor nodes are used inside WSN which continuously changes its position due to which strength of the radio signal varies with an unknown decay factor. newline