Research Of The Distributed Localization Algorithm Based On Wireless Sensor Network

With the development of embedded system, distributed information processing, wireless communication, wireless network technology and intelligence technology, the development and application of wireless sensor network (WSN) are promoted. Nowadays, WSN has been widely used in many military and civil fields. Since node localization technology is one of the main support technologies for WSN, how to accurately obtain the localization has become the focus of attentions. Therefore the research of WSN localization technology has important significance of theory and the practice.MDS-MAP algorithm is one of WSN localization algorithm, it uses the multi-dimensional scaling (MDS) which is a data analysis technique in statistics, and transforms the node related information into the space coordinates. The MDS-MAP algorithm may obtain high locating accuracy excluding the anchor node, thus obtains the widespread applications. But this algorithm is a centralization algorithm, the computation and communication costs are very high, which does not fit the large-scale network.This thesis designed an advanced distributed MDS-MAP algorithm (AMDS-MAP(D), Advanced MDS-MAP (Distributed)), which used the distance quantification in the localization phase, and did not need the hardware facilities. This algorithm can perform the nodes calculation locally, thereby it reduces the computation and communication costs, and is suitable to the shielded environment. In order to reduce quantization error in the process, the algorithm included the refinement algorithm, which used the position information of the neighbor node to modify and restricted the position estimate value, and enabled it to approach the real value unceasingly. At the same time, in order to overcome the simple refinement algorithm drawbacks, the concept of erroneous measure value and the bad node elimination were introduced to, further improved the refinement process. The method of clustering was also involved that the nodes in the network were divided into different clusters for local positioning, then translated the map in the local coordination into the global coordination. It has considerably reduced the time complexity of the algorithm, enhanced the scalability of positioning algorithm.In the process of node locating experiments, in order to improve the accuracy of ranging, the ranging correction model was involved. This model used a combination of correction methods, consider the hardware equipment's influence in the whole system, and the greater the number of nodes, the more accurate of the correct values. It is applied to the uncertain environment of random node.The results of this thesis show that, the AMDS-MAP(D) algorithm can realize distributed computing by involving the distance quantification and refinement algorithm, thus improved positioning accuracy. By the simulation and actual sensor nodes experiment, AMDS-MAP(D) algorithm shows good correction capabilities for the weakness of wireless signal strength, and achieves more satisfying positioning.