Research On Rssi-Based Localization Technology In Wireless Networks

WSN (Wireless Sensor Network) and WLan (wireless LAN) are developing rapidly. Compared to the traditional cable connections, wireless networks have some advantages, such as flexible network composition and the freedom of nodes move, which make the location of nodes in the network more and more important. Node location information plays a supporting role for basic network construction. And the perceived location of the node can also provide location-based services. The most famous one of existing localization systems is GPS (Global Positioning System). However, GPS has its own limitation. RSSI-based localization technology is a hot topic in the research field of wireless networks.In this thesis, the research of RSSI-based localization technology is taken as follows:(1)In a wireless network, some nodes know their self-positions, while some do not. We use the model of radio signal propagation in free space, fitting relationship between RSSI and signal transmission distance. On the basis of trilateral positioning technology, we introduce the idea of probability and propose localization algorithm named LRPD (Localization algorithm based on RSSI Probability Distribution), to estimate position of the target node with the frequency of corresponding RSSI. In a real environment, we use the Intel wireless interface and Samsung tablet to take an experiment to test the performance. Besides, in some relatively sparse wireless sensor networks, we propose a mobile node moving control algorithm with the RSSI changes to determine the location of the target node.(2) Different from the open area, indoor wireless signal is easily reflected and diffracted, so the relationship between RSSI and transmission distance is difficult to calculate directly. So we mainly use fingerprinting positioning method in indoor localization. We use C++simulation experiments to test factors that may affect the positioning accuracy. According to this, we propose an improved k- nearest neighbor (IKNN) algorithm, and experiments show that the accuracy is improved by about 16%, compared to the traditional k- nearest neighbor algorithm in the experiment scenario.(3) Combining two algorithms above, we design and implement an indoor localization system with TI-CC2430 sensor nodes. Setting up virtual reference nodes and selecting the k value dynamically, it first use IKNN algorithm to determine a primary area surrounded by the reference nodes near the target one. Then, it uses LRPD algorithm in the primary area, locating the target node. In addition, a scheme using the same equipment of nodes in the wireless networks is put forward to locate the transceiver-free object, in order to detect the intrusion.