SECURE LOCALIZATION ALGORITHM TO DETECT SYBIL ATTACK IN WIRELESS SENSOR NETWORKS

Abstract

Wireless sensor networks (WSNs) have revolutionized surveillance and monitoring applications by offering remote control and regulation capabilities. In most applications, networks utilize mobile nodes and rely on localization techniques to track the nodes' positions and movements. However, ensuring the security of the entire network poses a critical challenge. A single malicious node pretending to be another can wreak havoc and compromise the entire system. To tackle the presence of malicious nodes, this research presents a novel secure localization algorithm designed to estimate the positions of unknown mobile sensors in the presence of multiple coordinated Sybil nodes, while also detecting these malicious nodes. The research aims to provide a robust system that can withstand the rigors of real-world applications. The algorithm accomplishes this by initially evaluating the network's geometric characteristics and unknown node location by time of arrival (TOA) measurements, followed by iterative detection employing the Generalized Likelihood Ratio Test as a mathematical framework. Next, infectious nodes are eliminated from the network, and estimation is performed utilizing the Geman McClure cost function. The final estimation guarantees resilience, facilitating precise localization even in noisy environments and in scenarios where not all malicious nodes are detected. The algorithm's performance is assessed by analyzing Root Mean Square Error (RMSE) and the probability of correct detections for different network states and considering two types of attacker models, those capable of exclusively performing enlargement or reduction attacks, as well as attackers capable of executing both attacks. The algorithm demonstrates a high probability of detection, surpassing 0.95, for attack intensities greater than 15m, while achieving a lower Root Mean Square Error (RMSE) compared to localization systems reported in the literature. The proposed algorithm's convergence is assessed by comparing it with existing literature, thereby affirming its practicality in WSN environments. The system can be improved by extending the capability to detect and prevent attacks other than enlargement and reduction attacks and improving the detection rate at low attack intensity.