Efficient localization scheme based on coverage overlapping in wireless sensor networks

Abstract

Localization is an important and active topic in wireless sensor networks. This paper proposes a range free localization scheme based on coverage overlapping. The nodes knowing their locations are called reference nodes and the nodes without the knowledge of their locations are called unknown nodes. In addition, for a reference node, the area within its communication range is called the coverage region. In the proposed scheme, an unknown node identifies the coverage regions in which it is located by collecting beacon information from its neighboring reference nodes. The overlapped area of all the coverage regions should be the area where the unknown node is located in. An overlap point is defined to represent the overlapped area of two coverage regions. Besides, an overlap degree, which is the number of the coverage regions that cover this overlap point, is associated with each overlap point. All the overlap points that are not located in the Estimative Rectangle (ER), which can be derived by applying the CPE algorithm, are discarded. The remaining overlap points are sorted according to their overlap degrees. Moreover, among them, the ones with the maximum overlap degree are averaged and the averaged point is the estimated location of the unknown node. With the proposed mechanism, the reference nodes that cannot contribute to the localization accuracy are excluded, and hence better localization accuracy can be achieved. Experimental results show that the proposed scheme outperforms the related works, including Centroid and CPE algorithms, in term of better localization accuracy.

An efficient wireless access point selection algorithm for location determination based on RSSI interval overlap degree determination

“…The concept of the overlap degree has been used for localization in [20]. It is used to represent the number of regions that cover the overlap point.…”

“…K is the total number of RPs which are the nearest RPs to TP. In this paper, K is 4 as advised by [18][19][20][21] when KNN is used. In general, if K equals 3 or 4 will yield good results.…”