Multidimensional Scaling-Based Localization Algorithm for Wireless Sensor Network with Geometric Correction

Position is an extremely important parameter for each node in most of applications using wireless sensor work (WSN). In this paper, a new distributed localization (DMDS-GC) algorithm is proposed based on the multidimensional scaling (MDS) technique and the geometric correction technique. In DMDS-GC, the entire network is split into several localization clusters (LCs) with sufficient common nodes between neighbor clusters by a heuristic search algorithm. Then, based on the geometric relationships between nodes, the shortest path distance matrix is corrected in each LC. During the computation and optimization of relative local maps for LCs, the SMACOF algorithm, which is an improved version of MDS, is implemented. Finally, with the help of common nodes, the global map of entire network is achieved through coordinate transformation and mergence of LCs. Simulation results show that the DMDS-GC algorithm has better performances on localization precision, communication load, energy efficiency and robustness to range error, which can meet the needs of nodes running in WSN.