Research On Localization In Dynamic Sensor Networks

Dynamic sensor networks as a kind of wireless sensor networks, in recent years have been widely used in many fields, such as environmental monitoring, animal tracking, health, mine rescue, all of these applications are dependent on the mobility of the nodes. The mobile node can not only improve the event monitoring, but also enhance the flexibility, self-organization and adaptability in wireless sensor networks. For the dynamic sensor networks, the node location information is the premise of its application. Compared with node localization in static sensor networks, the node localization in dynamic sensor networks is more complicated due to the mobility of nodes. How to achieve the localization with low power consumption, low complexity and high precision in dynamic conditions is one of the challenges facing the localization in wireless sensor networks.The node localization technologies in dynamic sensor networks are the main topic in this dissertation. The dissertation explores the localization technologies in different dynamic sensor networks models and proposes some localization schemes. And then the dissertation presents a comparative simulation study of node mobility models in the performance evaluation of localization. The main work and innovation are as follows:Firstly, aiming at the network model which consists of static unknown nodes and mobile anchors, the dissertation presents a localization scheme with mobile anchors based on virtual beacon selection. How to reduce the error accumulation and minimize localization error in the presence of ranging error is considered in the scheme. The scheme gives the quantitative analysis on the error introduced by the relative position between the nodes from the perspective of geometry. The theorem is proved which the localization error is minimal when the position of the three virtual beacons into an equilateral triangle. In order to obtain more accurate location information, the unknown node chooses the appropriate virtual beacons to compute its estimated position according to this theorem.Secondly, the dissertation focuses on the path planning of mobile anchor and proposes a dynamic path planning method which is adaptive for network topologies. In order to make the mobile anchor moves according to the distribution of nodes in the networks, the moving direction of anchor is divided. The unknown nodes communicate with the anchor to obtain the number of the neighbor nodes. The anchor chooses the direction who owns the maximum number of neighbor nodes and moves. The simulation results show that the dynamic path planning method proposed in the dissertation can effectively avoid traversing the region without any node, particularly suitable for the networks with the non-uniformly distributed nodes.Thirdly, the dissertation explores the localization technologies aiming at the network model which consists of mobile nodes and anchors. Monte Carlo localization algorithm is an important method to solve this kind of problem. However, a large number of samples are sampled to achieve the positioning. This not only reduced the efficiency of the algorithm, but also consumed a great deal of energy of nodes. Aiming at these shortcomings, the dissertation proposes an adaptive sampling improved Monte Carlo algorithm. In the algorithm, the sample area is built on the one-hop and two-hop anchors of the nodes. And then the Kullback-Leibler distance as error limits is adopted to measure the error between the true value and estimation of the posterior probability density distribution of the node position. According to the sampling area size, the maximum sampling attempts based on the error limit is calculated. After uniformly sampling and filtering in sample area, the weights are given to the samples according to the one-hop neighbors of the nodes, and the estimated position is the weighted means of the samples. The simulation results show that the algorithm can reduce the number of sampling and the computational expense in case of ensuring the accuracy.Due to the network size and cost constraints, it is difficult to obtain the moving tracks of the nodes from the actual scene. Thus, at present the research on the node localization algorithm in dynamic sensor networks still adopts the simulation technologies. The node mobility model is the basis for simulation on node localization algorithm in dynamic sensor networks. Lastly, the dissertation presents a comparative simulation study of node mobility models in the performance evaluation of localization, and puts forward the viewpoint that the application scenarios of the localization and the basic characteristics of node mobility models should be taken into account in the simulation research of the node localization in order to accurately evaluate the performance of the algorithm. Moreover, the dissertation gives the analysis and explanation by examples.