Outlier node localisation in sensor networks based on double layer modified unscented Kalman filter

摘要

Traditional sensor network abnormal node localisation has some problems, such as low positioning accuracy, high positioning time cost and so on. The range of motion of abnormal nodes is determined by controlling the distance of abnormal nodes; According to the fading degree of abnormal nodes, the unstructured feature extraction model is constructed, and the minimum mean square error of abnormal nodes is calculated. According to the median change of neighbour nodes, the movement situation of abnormal nodes is analysed. With the help of the upper unscented Kalman to determine the nonlinear state space of abnormal nodes, the abnormal nodes in sensor networks are located, and then the positioning error is corrected by the lower unscented Kalman to realise the abnormal node location. The results show that the highest accuracy of the proposed method is about 96, and the shortest positioning time is about 1.1 s.