High-precision localisation algorithm in wireless sensor networks

This paper is concerned with the nodes localisation approaches in wireless sensor networks by adding two new variables, proportionality coefficient and deviation. This idea is different from the generally used least square nodes localisation methods which only consider coordinates of nodes. It is shown that if the measurement error between a blind node and each of its adjacency nodes is a linear combination of a fixed proportion and fixed deviation. The real position of blind nodes can be determined by filtering the errors. Even though the proportionality coefficient and deviation are uncertain, if the changes generated by random disturbances are small enough, the localisation errors can be reduced greatly by using the proposed method. The computational time is shorter than existing ones since only the time of solving linear equations is needed. If the errors need to reduce by iterations, a stable result can be obtained after ten iterations. It can be seen that only the measurement errors and the stability of the measurement are required but the communications cost is not required. Moreover, the proposed method possesses the conditions of parallel algorithm, i.e., it is available for multi-nodes.