Research On Statistical Methods For Robust Localization In Wireless Sensor Networks

The node localization is a very important foundation technology in the field of wireless sensor networks, because so many applications of wireless sensor networks (WSN) are dependent on the location of the node. However, the networks are generally deployed in a hostile environment. The nodes are vulnerable to a variety of targeted attacks, resulting in a positioning failure. Once fail to localization, it will bring serious consequences, such as paralysis of the entire sensor network or monitoring to get data failure. Therefore, the secure localization in wireless sensor networks is a challenging topic which has wide application prospects.Security positioning mechanism can be resolved in varying degrees, but the positioning of security issues still has some shortcomings:weak to resist the attack, the higher the malicious beacons ratio of involved in positioning attack, the greater the positioning errors; if this ratio is greater than a certain percentage, algorithm will failure completely. To address these issues, we focus on the conditions of secure localization algorithm, and then give a safety localization error range. We propose an adaptive switching between least square method and least mean square method which has the computational advantages. Not only has the advantage of the least squares method of calculation (when the attack exists), and also has the advantage of switching to the calculation of the robust model.