Odometry Based Localization Technique Research For Mobile Wireless Sensor Networks

In recent years, more and more people are concerned about the wireless sensor networks with the rapid development of micro system technology and wireless communication technology. In most applications, the data collected by the sensor nodes makes sense only when it is associated with the location information. Therefore, how to locate the sensor nodes is not only one of the core problems, but also a research focus in wireless sensor networks.Based on the extensive research of the existing wireless sensor nodes localization methods, we made use of the velocity information of the nodes, which can significantly improve the localization accuracy and overcome the impact of non-line of sight as well, even though a lot of obstacles in the working environment. So we proposed three node localization methods based on the mobile features in this paper.(1)OBTL(Odometry Based TOA Localization) obtained the reference position making use of the mobile node's velocity information, and then selected the most accurate ultrasonic distance measurement between the mobile node and the beacon nodes based on the reference position. At last by utilizing simple ratio computation, it got the accurate localization result according to the most accurate ultrasonic distance measurement which was considered as not NLOS affecting measurement by the dynamic threshold. The simulation results showed that OBTL was able to overcome the impact of local NLOS, compared to LMedS its accuracy improved about 89.4% while the computation time cost was only 6.1% of LMedS.(2)OBMCL(Odometry Based Monte Carlo Localization) utilized the mobile node's velocity information, which overcame the independent of the history limitation in the existing Monte Carlo localization, first it improved the prediction accuracy of the particles, and then linked the connectivity information at different time to update the weight of particles more accurately during updating phase and resampling phase. The simulation results showed that its accuracy improved more than 69.1% compared to MCL and DVhop. (3)uNMCL(under NLOS Monte Carlo Localization) made pre-assessment of the connectivity information by the node's velocity information in order to exclude the impact of the NLOS. Then it updated the particles'weight by the correct connectivity information which could improve the localization accuracy and immunity. Experiment results showed that its accuracy improved 72.3% than DVhop under NLOS environment.The simulation showed that the three methods proposed in this paper had low computation complexity with high localization accuracy, which were very suitable for mobile sensor nodes with limited resource.