Research On Three-Dimensional Localization Algorithms For Wireless Sensor Networks

Wireless Sensor Networks refer to the self-organizing multi-hop networks, in which the designated monitoring area is composed by large number of the cheap micro-sensor nodes and are formed by wireless communication. The aiming to collaboratively perceive, collect and dispose the information from the perceived objects in the network overlay area, and then send to the observer. Recently, with the high speed development of the sensor technology, Micro Electro-Mechanical technology, wireless communication technology and modern network, the Wireless Sensor Networks are already become one of the most important technology now. They have broad application prospects, and have already attracted more and more attention, it is awarded that it will be the one of bring a great changed in human life of the top ten key technology in the future. Wireless Sensor Networks can be widely used in environment monitoring, military reconnaissance, catastrophe early warming, traffic guidance and many other fields.In Wireless Sensor Networks, a large number of the sensor nodes are usually deployed randomly in the monitoring area. The precise position of the nodes in the network may be unknown in advance. But in many applications of the WSN, we need to know the exact position where the event occurs, it is important to know the position of the sensor nodes. Therefore, determining the precise positions of the sensor nodes is the prerequisite and basis for a variety of applications. At present, most research about sensor nodes localization are based on the two-dimensional area, but in the practical applications, sensor nodes are usually deployed randomly in the three-dimensional space in which need the localization algorithm could locate the sensor nodes. The major research of this thesis focuses on the nodes positioning in 3D wireless sensor networks.First this paper reviews the basic concept, the characteristics and key technology of the WSN, introduces the evaluation indicator and classification of the node location technology, shows some method to solve node location, and conclude some performance indicator.In consideration of the contradiction between the communication lost and the positioning accuracy, a novel based on grid-divisional and power classification convergent iteration algorithm has been proposed for three-dimensional wireless sensor networks (GDPCI) after researching several typical three-dimensional nodes positioning with meshing the thought of convergent iteration. This algorithm does not need to know the distance between nodes, with the power classification strategy, the anchor node divides the coverage area into several sphere shell area, the unknown nodes only need to receive the messages which come from the anchor nodes broadcasting, and the messages include the position coordinates and the level of the broadcasting power of the anchor nodes. When the unknown nodes which are in the communication range of the anchor nodes receive and store the beacon information, and then with the thought of the convergent iteration, each anchor node proceed the power grading according to the space which have already determined. Thus, the intersections of the centroid which come from different anchor nodes will be the estimated position of the unknown node. That can make the space which includes the unknown nodes become smaller, and efficiently increase the localization accuracy.Compared with the existing three-dimensional positioning algorithms, GDPCI algorithm has less computation cost and does not need extra hardware support, and because the second power grading only for the space where has already determined, the communication cost as well as the localization error of the algorithm has become lower, and also extend the life of the network, meanwhile, it is more robust to the network topology, and is very suitable for the application of low-power wireless sensor networks. The simulation result demonstrate that in the three-dimensional space,20 anchor nodes and 200 unknown nodes are deployed randomly after the level of power grading more than 6, the average positioning error is only 14%, and more up to 99% of the unknown nodes can be located.