Study On Node Localization In Wireless Sensor Network Based On Unmanned Aerial Vehicle

Wireless sensor networks (WSNs) have a very promising prospective in many areas which need information acquiring. The location of where the information acquired is very important. GPS modules are always failure to work indoors. Because of this, beacon nodes are not usable for localization of indoor WSNs. Aiming at the mentioned problem, a localization mechanism is proposed in this thesis based on an unmanned aerial vehicle (UAV) and a mobile anchor node carried with it. This mechanism is able to balance localization efficiency and accuracy.First, aiming at acquiring the location of WSN nodes without beacon nodes indoors, a Linear Auto-localization algorithm is adopted to calculate the distances between all unknown-nodes. In order to solve the high measurement error problem of using RSSI, a method using grouping strategy and data smoothing if proposed in this thesis. It is able to efficiently reduce measurement errors and sampling frequency. The effectiveness and computational accuracy of LAL algorithm are guaranteed by this strategy. By simulation experiment, the improved LAL algorithm has the ability of controlling the transfer of the errors, and builds the foundation of the follow-up researches.Second, according to the proposed localization system, a path planning method is proposed aiming at reducing the localization errors. This method has a good performance in controlling the flying cost and the localization accuracy. This method, based on Voronoi diagram and the analysis of the Dilution of Precision (GDOP), optimizes the selection of path point. And then, the Genetic algorithm is adopted to solve the traveling salesman problem, in order to acquire the final optimized path of the UAV. Through simulation experiment, by comparing with two classical traversing methods, the proposed method has the advantages in controlling both the flying cost of UAV and the localization accuracy of unknown-nodes.Third, the idea of solving a non-linear target tracking problem with Unscented Kalman Filter algorithm is introduced to the accurate localization problem. And UKF is improved by combining with the maximum likelihood (ML) estimation. The distances between mobile beacon node and unknown nodes are measured by RSSI method. And using the location estimated by ML, the initial status values are sent to UKF to calculate the ultimate estimated location. The simulation experiment shows that the proposed ML-UKF algorithm has very good performance in both localization accuracy and adaptability.In order so solve the problem that the lack of effective location information of beacon nodes indoors without GPS, a WSN localization mechanism with the assistance of UAV is proposed. The distance between unknown nodes is estimated using modified LAL. Based on the analysis of GDOP, a path planning method aiming at reducing the localization errors is proposed. And finally, the UKF algorithm is improved with ML in calculating the ultimate location of unknown nodes. All the proposed methods are proved by simulation to be practical and effective. The localization accuracy is guaranteed as well.