Research On Three-dimensional Global Localization Algorithm In Wireless Sensor Networks

As the application of Wireless Sensor Network (WSN) is developing increasingly, the position of the sensors becomes more and more important as the basic of many applications of Wireless Sensor Network. According to the characteristic of the sensors which has limited energy, storage and computation ability and etc, it's more and more important to get the sensor's position quickly and accurately. Until recently, there are two categories of localization algorithm in WSN: range-based algorithm and range-free algorithm. This paper will do research and improving on the two categories of localization algorithm separately.Firstly, this paper studies the range-based localization algorithm. Although the existed range-based localization algorithms improve the precision and computation cost, they can hardly do this at the same time, when they improve the precision of the localization, they will have a high computation cost. The linear optimization can achieve small calculating amount at the cost of reducing the positioning accuracy, and the traditional unconstrained nonlinear optimization has a better performance in accuracy but always demands large calculating amount. Basing on the characteristic of sensors, this paper proposes a modified nonlinear optimization method. Simulation results show that the two algorithms proposed in the paper are efficient to relieve the contradiction between calculating amount and localization accuracy by improving the traditional algorithms. This method has the advantages of high localization accuracy, relatively low calculating amount, and without requiring extra communicational cost. At the same time, considering that the centralized algorithm has the disadvantage of high computation cost and energy cost for some sensors, this paper studies the distributed algorithm in localization of WSN. By introducing the particle swarm algorithm, this paper proposes a Multi-population Parallel Particle Swarm Optimization Algorithm (MPPSO). Through deploying particle swarm in several beacons, we can reduce computation cost of sensors to be localized and achieve the parallelism of the algorithm, meanwhile, the local optimization problem can effectively being avoided by using the scheme of multi-population co-evolution. Simulation results show when ranging error is less than 30 percentages, MPPSO can improve location precision by 8% to 40% against some traditional algorithms.Secondly, considering the hostile environment of the sensors, it is very important to acquire the node's position without GPS. We propose one novel anchor-free algorithm based on cluster technique, node's energy, connective degree, and the geometric limit principles of triangle inequality is also considered to heuristically build clusters, then fuse the clusters into one cluster and get one uniform global coordinate system. This algorithm effectively solves the position problem without any anchor node, conquer the cumulative error problem in traditional anchor-free algorithm, improve the localization precision and save the sensors energy. Simulation results show that after improving the fusion section, our algorithm can improve localization precision by 30% to 70% comparing to the traditional algorithm ABC.