The Research And Implementation Of Secure Localization Algorithms In Wirless Sensor Networks

Location information of sensor nodes not only plays a fundamental role in key network mechanisms including routing, task scheduling and topology control in sensor networks, but also serves as key information of environmental surveillance, traffic control, home medical care and many other applications. However, due to the inherent vulnerability of sensor networks, existing algorithms using both beacon-based and beacon-free methods introduce a non-negligible error. Meanwhile, malicious nodes may disrupt normal operations by tamping with their own locations.In this thesis, we propose a localization scheme based on bounding box, which uses the Received Signal Strength Indicator(RSSI)to measure distance. In order to mitigate the estimation error brought by non-unique intersection points problem, our proposal adopts squares instead of circles in triangle inequality localization method. Based on the algorithm aforementioned, an anomaly detection approach is further proposed to enhance the security of our scheme. During the training phase, the nodes in different neighbor regions of a selected node is repeatedly counted, and collected as sample sets for Orthogonalized Gnanadesikan-Kettenring estimation. During the detection phase, every node in the networks calculates the Mahalanobis distance between the number of neighbors and the estimated value to determine whether its own location is normal. Simulation results demonstrate that the proposed algorithm maintains a relatively good performance even under a harsh attack and the false positive rate remains stable when the number of malicious nodes dramatically increases. A geographic information system aided localization platform is then constructed with the functions of data collection, storage, management, and remote monitor via Internet. Eventually, the location-anomalies detection algorithm is implemented in Matlab and Java,and incorporated into the GIS-aided localization platform. Software test result shows that our system has the advantage of reliability, scalability and friendly geometry-user interface (GUI).