Study On Localization Algorithms For Large-Scale Wireless Sensor Networks

Wireless Sensor Network is integration of sensor techniques, nested computation techniques, distributed computation techniques and wireless communication techniques. It can be used for testing, sensing, collecting and processing information of monitored objects and transferring the processed information to users.Localization, which has been focused as a basic problem, is the precondition of many applications of sensor networks, such as target identification surveillance and tracking. In this thesis, we discuss the distributed localization algorithms in detail. These algorithms share a common, three-phase structure: (1) determine the distance from node to anchor (2) nodes compute their positions through the relationship of distance, angel or network connectivity to the anchors (3) refine the positions through an iterative procedure using neighbor's information.The research status of wireless location techniques and positioning systems for WSN are summarized based on the study of a lot of related literatures. In this thesis, the principles and characteristics of related approaches on WSN are discussed and presented. The criterion of performance evaluation and algorithms, the taxonomy for wireless sensor networks localization systems are also described in the thesis. The principles and characteristics of recent representative distributed localization algorithms including DV-distance, DV-hop, Euclidean are also discussed, presented and compared on a single simulation platform. Then, the thesis focuses on the investigation of DV-hop localization algorithm. Two improved node positioning algorithms, Dv+Lastdist and Dv-hop-MinMax, are proposed, together with the discussion of algorithm principle and realization in OMNet++ simulation environment.Simulations show the average positioning error of each algorithm in different scenarios, and the effect of different parameters on network performance is also analyzed. The main conclusion is that no single algorithm performs best, which algorithm is to be preferred depends on the conditions (range errors, connectivity, anchor fraction, etc.)In the end, it is the conclusion of the thesis and the prospect for the future research work.