Research On WSN Localization Based On MDS And Distance Correction Technique

Wireless sensor network (WSN), which is the integration of sensor techniques, MEMS techniques and nework communication techniques, is an innovative technique of information acquisition and processing and has a wide application foreground. In some key techniques including positioning technique, however, there are still many problems remained. The presenting and solving of these problems are essencially significant to the theory and application research of WSN.Sensor localization is crucial for the configuration and applications of WSN. In this papar, through analyzing and generalizing the present state and perspectives of localization techniques of WSN, the range-based localization algorithm was chosen as cut-in point. First, two distance correction methods were proposed, which are quadrilateral method and adjust weight method. Assuming that the measured distances between neighbor sensors were available, quadrilateral method used the geometrical relationship of sensors to calculate the distance estimation between multi-hop sensors and then corrected the distance estimation through eliminating the interference of redundant solution and range error. Adjust weight method tended to converts the relationship between the real distance and the shortest path distance of anchor sensors to the adjust weight and then corrected the distances between sensors which were within certain radius of anchor sensors through the shortest path distance table.Through simulation, the adjust wight method with better performance was chosen to be applied in the phase of distance correction in the later localization algorithm. Then, a novel distributed localization (MDC) algorithm which is based on multidimensional scaling and distance correction was proposed in this paper. It perfectly inherits the dominance of the existing MDS localization algorithm as well as covers its shortage. In MDC, several local positioning regions were firstly constructed by a search algorithm, and the distances were corrected in each local positioning region. After that, an iterative process was applied conbining the classical MDS algorithm and SMACOF algorithm to compute the local relative coordinates of sensors. Finally, the global positions of sensors were obtained through merging relative maps of all local positioning regions using the information of anchor sensors. Simulation results show that the MDC has better performances on positioning precision and robustness to range error than the existing MDS-MAP and MDS-MAP (P),which can meet the requirements of applications about sensor localization in WSN.