| Wireless Sensor Network (WSN) is a new generation network, which integrates sensor technology, embedded technology, distributed processing technology and wireless communication technology, and has been applied in many fields, such as military, environmental monitoring, construction, industrial and agri-cultural fields, etc. Location information of the sensor network is very important. For example, when the WSN monitors a fire, what people first want to know is the location of the fire. Currently, research of nodes localization in WSN usually focuses on two-dimensional case, but in reality, the nodes are generally ar-ranged in three-dimensional space, among which there are quite a number of applications are carried out in indoor environment. Therefore, it is necessary to study the unknown node localization of three-dimensional indoor environment.This thesis analyzes researches on three-dimensional localization of wireless sensor network made at home and abroad, and then creates a specific scenario, that is, a three-dimensional space of a room with limited length, width and height. We want to study how to achieve high-precision localization based on distance measuring with RSSI. The main research can be summarized as follows:1. The number and placement methods of reference nodes. The number and placement methods of reference nodes have a great influence on the accuracy of localization. First, we prove a theorem, which describes the relationship between the placement of reference nodes and whether the node can be located. In order to solve the low-precision in the z direction, the decomposition of once locating processing is exe-cuted, and then we can get several positions of an unknown node for midvalue averaging filter. Thus we will obtain the location of the unknown node, and avoid some single cases which can not be located, and improve the localization accuracy. Based on the previous work, for this scenario, the reference nodes placement approach is designed.2. Reference nodes selection algorithm. In locating processing, we can use many different combina-tions of reference nodes, but due to ranging error, we can not get the same results by using any combination of reference nodes. According to the matrix theory, we make a reference nodes selection algorithm, which can minimize the influence of distance measurement error. 3. Application of error processing theory to WSN localization. In the RSSI and coordinate data of a locating processing, there may exist some error values which are far off the normal ones, as called "gross error". Gross error should be removed. The national standard "GB/T 4883-2008 Statistical Interpretation of Data Detection and Treatment of Outliers in Normal Sample" clearly provides the judgments and treatment of outliers (namely "gross errors"). In this thesis, in order to improve the accuracy of localization, we in-troduce the error processing theory into WSN locating processing.4. Using twice-filtering methods to improve the locating accuracy further. Digital filter can effectively inhibit the components of interference signal, and eliminate random error, making the data as close as pos-sible to the true values. Due to the instability and randomness of RSSI, by measuring a number of RSSI values and then do a midvalue averaging filtering to improve the range accuracy; for the obtained position-ing coordinates of multiple estimates in the course of localization, we use a midvalue averaging filter again to improve the accuracy of estimated position.5. Studies the influence of ranging number to the locating result in Line-of-Sight (LoS) and Non-Line-of-Sight (NLoS) environments separately.Simulation results show that the proposed three-dimensional localization algorithm for indoor envi-ronment has less error. Compares to the currently available localization algorithm which only using RSSI, relative to the size of the room, positioning error is reduced by 7%, and can be used in other positioning scenarios. |
PDF Full Text Request