Research On The Improved Range-based Algorithms Of Positioning In WSN

Wireless sensor networks (WSN) is a modern network technology synthesizes of sensing technology, embedded technology, distributed information processing technology and wireless communication technology. It can be widely used in national defense, national security, environmental science, traffic management, hazard prediction, health and medical, manufacturing, city information construction, etc. The sensor network technology will become an interface between the real world and the digital world, and will penetrate into all spheres of our life. The intensive study of this technology will not only promote the national information construction in China, but also drive the related industries and subjects, so as to bring a new growth point to national economy. As a key technology of wireless sensor network, positioning plays a central role in the sensor network operation.In this paper, we first summarize the WSN, introduces its structure and characteristic. Then we do some research on the node positioning in WSN, and mainly study the range-based algorithm in non line-of-sight (NLOS) environment. According to whether the actual measurement of the distance between the nodes in the WSN is required, The positioning algorithms are mainly divided into two sorts, one is based on the measured distance (range-based), and the other do not (range-free). Because of higher positioning accuracy, this paper focuses on the studies of the range-based algorithms. In actual systems, the signal does not transmit along line of sight (LOS) all the time. How to eliminate or mitigate the influence caused by non line-of-sight (NLOS) error is an essential issue to positioning in WSN. In recent years, people proposed several methods to address it, which can be divided into three categories:the first kind is stadia reconstruction, judge the NLOS signal first, and then apply some rule to reconstruct the measurement signal in LOS environment. The second one is weighted all signals, the rule is that NLOS signals are weighted by small weighting and LOS signals weighted by big weighting, so as to reduce the effect of NLOS. The third class is picking out the LOS signals from all received. Owned the advantages of high accuracy, the paper mainly studies this one.Among these algorithms, Residual Test (RT) algorithm proposed by Chan etc. is a typical one. This method firstly groups all the range measurements into different combinations; then computes the combination residual squares, if the system error follows Gaussian distribution, the residual squares will obey chi-square distribution of one degree; set one threshold, and calculate the number of that squared residuals are bigger than the threshold. This algorithm traverses all combinations, so it is computation consuming for the case of many Base Stations and not suitable for practical application of wireless sensor networks.To address the high computational complexity of RT algorithm, take concern of localization accuracy, we proposed algorithm IRT algorithm. The algorithm firstly checks the sets with four range measurements, and picks out the one with smallest residual square. Then iteratively combines the smallest residual square corresponding set with each range measurement excluded into several new sets, and picks out the smallest residual one until the set with smallest residual been contaminated by NLOS range measurements. Since been improved, this algorithm avoids lots of computation and reduces the influence of system error for utilizing the redundant Base Station. Furthermore, we replaced AML by LS during the verification and optimized the subsets, proposed another algorithm.In this paper, we simulated the algorithms under different environments on the MATLAB 7.0 platform. According to the simulation results, the performance of proposed algorithms is close to RT, but IRT and CATO can both reduce the computational complexity of RT's.