| Wireless Sensor Network (WSN) incorporates multiple technologies. It is a network of numbers of low cost micro-sensor anchors disposed in a certain monitored area, with the ability to sense, coordinate, calculate, self-organize and communicate wirelessly.WSN can be widely applied in many areas, including military, the monitoring of environment, complicated machine and the safety of industrial park, etc. All these applications need the support of the technology of target localization. During the detecting of WSN, information of position which acquires nodes is great importance, and without location information, all the efforts will become useless. What's more, the localization of nodes is also the premise of some technologies, such as target tracking and geographical routing. Above all, it is absolutely vital to research the technology of nodes localization.And according to whether to measure the actual distance of the nodes during the localization, the technique of nodes localization includes two kinds of localization algorithms: range-based localization algorithm and range-free localization algorithm. Some research indicate that the former one has a high precision and a high hardware requirement, while the latter one has a low hardware requirement and a small cost for the nodes and it can be applied in large scaled network of sensors, but it is not very precise. So the range-free localization algorithm is studied in this thesis.Major tasks of the thesis are as follows:(1) We do a deep research on the technique of nodes localization of WSN, and also make a detailed description, classification, analysis and comparison of the staple algorithms at present to sum up their advantages and disadvantages.(2) We give a priority to the study of APIT localization algorithm and analyze its disadvantages. APIT localization algorithm is a kind of range-free localization and it is more precise in a fixed isomorphic dense networks. However, it has a bigger localization error and a lower coverage rate in a random distributed network, which is because in a random distributed network of sensors, the nodes are not well distributed, maybe there are lots of nodes in some places, and little in other places, even no one. So the locations of many nodes can not be determined, and has a great influence on its precision. For these problems, a localization method based on mobile anchor node is proposed in this thesis, whose idea is that some mobile anchor nodes are introduced into the sensor network, and, during the initialization of the network, try to make the nodes be well distributed through a certain mobile coverage algorithm, then the localization performance of the unknown nodes can be promoted.(3) In most cases, the localization performance can be influenced by the number of the anchor nodes. And a fair number of anchor nodes can increase precision and coverage rate. So an algorithm based on a network of isomerism sensors to get the optimal number of nodes is presented in the thesis. Theoretically speaking, the number of needed anchor nodes and unknown nodes, which can cover the whole monitored area in a fixed area, can be worked out through this algorithm in advance.(4) For the ineffective nodes (only one or two adjacent nodes) which can not be localized by APIT algorithm, a specific method is provided in the thesis. Furthermore, we also use RSSI quantized model to amend the precision of APIT algorithm, and upgrade some new anchor nodes, such as convert the unknown nodes which can be localized at the first stage to the new anchor nodes, and then to participate in the next localization.(5) At last, we verify and discuss the overall performance of the improved localization algorithm. Based on the Matlab 7.0 simulation platform, we do a simulated comparison between the improved algorithm and the relevant algorithms. Experiments show that, compared with the same kind of algorithms, the algorithm based on mobile anchor node proposed in the thesis which has a higher precision and a bigger coverage rate. |
PDF Full Text Request