NLOS Mitigation Methods Based On WSN

In recent years, wireless sensor networks (WSNs), as a new developed technology, is becoming a research focus in international academic and industry field. It has a wild application prospective in military, environment, industry and many other fields. The localization of WSN nodes is one of the key technologies and application base of WSNs. In practical application, not all signals propagate under line of sight (LOS) situations. Signals scatter or reflect when they reached obstacles. Hence, the propagation delay appears, leading to the result of non-line of sight (NLOS) errors coming along. The localization accuracy is influenced by NLOS errors, and the performance of the whole system is decreased. Therefore, some WSN localization algorithms are proposed and deeply analyzed, aiming at the elimination of NLOS errors, in this thesis.First, comprehensive analysis of the basic theories, measurements, localization algorithms and NLOS models of WSN NLOS localization is done in this thesis. Comparison of the advantages and disadvantages of existed algorithms is carried out. Recent basic methods of NLOS error elimination are concluded, in order to make a base for proposed localization algorithm with better accuracy.Secondly, the reasons why NLOS errors exist and their features are analyzed. NLS, LLS, Taylor-LS, and CLS are carefully analyzed in NLOS localization, in the aim of finding the optimized form of LS in NLOS localization, which is used in localization algorithms proposed in this thesis.On the base of analyzing classical Rwgh localization algorithm, an innovative localization algorithm which is able effectively eliminate NLOS errors is proposed. PSO algorithm is adopted in the algorithm to solve the problem that LS is easily affected by the selection of beacon nodes. The temporary coordinates estimated by the LS are optimized with PSO algorithm. Aiming at the large calculation amount problem of traditional EM algorithm, a combination pattern with the number of measurement reducing is proposed, which select the least residual error to iteration calculate. Simulation results show that the NLOS errors are effectively reduced using the proposed algorithm, and the calculation amount problem of Rwgh algorithm is overcome even without the statistic features of NLOS errors.According to the mentioned research works, a WSN node localization algorithm based on GMM is proposed. The distance estimations are calculated using GMM with the distance measurements in NLOS environment. In order to acquire the optimized result, the ultimate distance estimation is calculated with the proposed novel EM algorithm optimized with PSO. The serious distortions are excluded by the proposed algorithm so that the NLOS effectiveness is reduced. The proposed algorithm is also a simple and practical method to localize WSN nodes. It is proved that the proposed algorithm is able to improve the localization accuracy and also has good robustness.Through the analysis of the reasons why NLOS errors appear and classical NLOS error elimination algorithms, a series of effective algorithms is proposed in this thesis. The good robustness and high localization accuracy are proved by simulation experiments.