Research On Localization Algorithms In Wireless Sensor Network Based On Multidimensional Scaling

Wireless sensor network is a new network technology that combines wireless sensor, wireless communication and embedded system technology. It has been caused great concern after being put forward. Wireless sensor networks have many applications such as environmental monitoring, disaster prediction and rescue, military field, medical and health fields, intelligent home, intelligent plant, remote monitoring. After monitoring a certain issue, we are more interesting in the location of that issue, thus, the information of location is the most important matter of wireless sensors in data collection, the monitoring without any information of location is meaningless and useless. As a result, the localization of the nodes in wireless sensor network is a research focus.In this article, I have researched many location algorithms and deeply researched the technology of multidimensional scaling (MDS). According to some weakness of MDS technologies, new algorithm has been proposed.This article firstly introduces the knowledge of wireless sensor network and makes a classification of current algorithms in wireless sensor network, and also this article have introduced some typical localization algorithms. There are many localization algorithms currently; the algorithms of range-free technology are more popular because of its low power and low cost consumption. The technology of multidimensional scaling (MDS) is one of them, MDS localization algorithm can work in the environment of range-free and fewer anchors and gets good precision, but it is a centralized localization algorithm and the node in the network should be uniform distributed.After having deeply researched MDS localization algorithm and according to some weakness of MDS, this article proposed an improved MDS (AMDS-MAP) localization algorithm. AMDS-MAP localization algorithm uses power-control technology distinguishing a one-hop node, thus, the distances between nodes can be more truly to their real distances. And also, AMDS-MAP localization algorithm uses MDS technology to a single known-node and anchor nodes rather than the whole network, thus, every node can be located by itself, that makes AMD-MAP localization algorithm a distributed algorithm. As a result, AMDS-MAP localization algorithm is better than MDS-MAP localization algorithm in precision and also it is a distributed algorithm, which makes AMDS-MAP much more useful in large wireless sensor network.Secondly, according to the weakness that both AMDS-MAP and MDS-MAP cannot work well in the environment which nodes are uneven distribution such as C-sharp network, this article proposes an new clustering algorithm that based on anchor and I have researched how to get better precision by placing the anchors. By placing the anchors at which there are more nodes as well as anchors are uniform distributed, we can get better precision.Thirdly, this article has deeply researched Monte Carlo Localization(MCL)algorithm and proposes an new algorithm that combines MCL and distribute MDS-MAP(MCL-MDS). MCL-MDS localization algorithm firstly uses distribute MDS-MAP to locate the node, then the consequence should be tested by the principle of filtering in MCL, we use that consequence if it pass the filter and we use MCL to locate the node if it do not pass the filter. MCL-MDS localization algorithm can combine the advantage of both MCL and AMDS-MAP, thus, it can work in any environment and get better performance.