Wireless Sensor Networks (WSN) has found a great variety of applications ranging from military surveillance to health monitoring. It is important to get correct location information to enable WSN applications such as target tracking and environment monitoring. Various techniques have proposed to realize the localization, yet the self-organization character of WSN makes the process of implementation vulnerable in hostile environment. The traditional localization technique can not meet the need of accuracy.This paper proposes a Statistical Information Grid (STING) multi-resolution clustering scheme, based on the traditional techniques of localization. The main content of the multi-resolution-clustering-based secure localization is as follows: In the radius of WSN communication, deploy the anchor nodes (including the malicious nodes) and the unknown nodes according to a certain proportion. Enumerating all anchor nodes that in the range of unknown nodes, two for each time. If the circles whose radiuses are the distance between anchor node and the unknown node and centers are the position of the anchor node, intersect at some point, recording the points. Organizing the field as multi-level spaces, the division of grid cell is based on distribution of intersect points and the grain of localization. The high-level grid cell is computed based on the low-level grid cell. The procedure of clustering is after the division of grid cells, and the grid cell which includes maximum of the intersect points is selected. Compute the centroid of the gird using the minimum mean square estimation (MMSE) to refine the result. Different levels would get various grained results of localization, and finally realize the secure and multi-resolution localization.The experiment indicates that the secure localization scheme could filter out the malicious nodes in a robust and efficient way. The algorithm not only identifies the node's position, but also has lower time complexity and higher security. |