Research On Self-Localization Algorithm Of Wireless Sensor Network

Wireless sensor network (WSN) is becoming a new approach of information collection. It is widely applied in the military, the medical treatment, the environmental monitoring and so on. Since the data collection, object tracking and mesh topology managing are significant only when the positions of corresponding sensor nodes have been identified, therefore, as one of the core technologies, the nodes locating plays a pivotal role in the WSN application.In the beginning of this paper, some problems in the present localization and the categories of the Self-Location Algorithm are described, especially, the range-based algorithm and the range-free algorithm. Then the advantages and disadvantages of various range-based algorithms are analysed, and the conclusion shows that RSSI (Receive Single Strength Indicator) is more adapt to the low-cost and low-power wireless sensor field than others because of its low hardware requirements. Finally, some experiments were done with Micaz nodes which are provided by Crossbow. The conclusion of the experiments is that RSSI has higher localization precision in the short area, average localization error is nearly 20% in the two square meters. But when the distance increases, the error of the localization also increases, so it is not reliable to use RSSI when the distance is beyond eight meters.In order to realize the long-distance localization, some range-free algorithms are compared. DV-Hop localization algorithm is selected because of its precision, haleness and adaptation. But when the nodes are sparse, the error of the algorithm is comparatively high, and the precision is nearly 40% when the communication range is sixty meters. In order to solve this problem, a method of restraining collinearity degree for deploying the anchor node is adopted, and LQI is used to revise the hop information, RSSI is used to revise the distance between the node and the anchor node which communicate in single hop. It is proved that the localization algorithm is improved obviously, and the average localization precision is about 20%.