Prediction Model to Analyze Source Node Localization Using Machine Learning and Fault-Tolerant in Wireless Sensor Networks

Abstract

– Recent technological developments include wireless sensor networks in modern and intelligent environments. Finding the localization of the sensor node is a problem in the research community field. Localization on a two-dimensional plane, a key focus in WSNs, is to maximize the lifespan and overall performance of sensor nodes by minimizing their energy consumption. The compiled data that base stations receive from packets.in wireless sensor networks can be used to make decisions with the help of localization. A cost-effective method of solving the problem is not the Internet of Things with GPR tracking sensor zones. There are several approaches to locating wireless sensor networks with unclear sensor locations. The main challenge lies in accurately determining the location of the base station's sensor node with a minor localization error during wireless communication. The proposed method, Distributed clustering Distance Algorithm (DCDA) using machine learning, considers the distance estimation error, location in accuracy, and fault tolerance issue with WSNs. According to the findings, the average localization error is 11% and 11.3%, respectively. For anchor nodes 20-80 and 200-450. As a result, when compared to contemporary methods of localization with centroid weighted algorithm (LCWA), Distance vector hop algorithm (DV-Hop), Coefficient for reparation algorithm (CRA), and Weighted Distributed Hyperbolic algorithm (WDHA) methods, the demonstrated Distributed clustering Distance Algorithm (DCDA) gives greater accuracy. According to the experimental results, the suggested algorithm significantly improves the number of alive nodes compared to the LBCA and G. Gupta FT algorithms. Specifically, the proposed algorithm achieves a remarkable 96% increase in active and functional nodes within the wireless sensor network.