Sequential Multidimensional Scaling For Realtime Sensor Network Localization

— We investigate on the localization of nodes in a sensor network based on multidimensional scaling (MDS). The distances of node pairs are given and MDS attempts to locate the position of nodes given the distance matrix. However, conventional MDS addresses the mapping problem using eigenvector decomposition which is complicated in computation, mitigating the efficiency of this approach for real-time applications. In this paper, we introduce a sequential MDS for the fast implementations of MDS for a changing sensor network. The fast-fixed point algorithm is used to initialize the locations of nodes. When the distance matrix is varied due to the movements of nodes, sequential MDS aims at reallocating the nodes with a fast and effective update process. We validated sequential MDS with a number of experiments and shown the proposed approach is superior to conventional MDS when it deals with the location problems of moving nodes of the sensor network. Keywords-Multidimensional Scaling; Localization; Sequential Eigenvector Decomposition.