Research And Application Of New Wireless Localization Technology In Ubiquitous Networks

The ubiquitous network technology such as wireless local area network(WLAN)、wireless sensor network(WSN) and ad-hoc networks has undergone rapid development in recent years. These network architectures have made it feasible for the widely application of many location-based services (intelligent transportation, remote surveillance, intelligent home applications etc). This paper mainly focuses on the research on new wireless localization technology in classical WLAN and WSN scenarios and we also developed an indoor positioning software for realistic application. The detailed work of this dissertation is summarized as follows:To deal with the low positioning precision caused by the limited number of access points (APs) in WLAN, a new compressive sensing (CS) based location fingerprinting algorithm is proposed for indoor localization. Then the location fingerprints space denosing and filtering algorithms are further proposed to solve the restricted isometry property constraints of CS theory in positioning application. The simulations and experiments indicate the CS based location fingerprinting algorithms can effectively reduce the average localization error compared with the state of art location fingerprinting algorithms. In addition, according to the theoretical CRLB analysis and experiment verification on the positioning error, an APs deployment optimization strategy based on the genetic algorithm is provided to improve the localization performance from the perspective of network architecture. Its effectiveness has also demonstrated in the simulations.Based on the extensive study on location fingerprinting positioning in WLAN, a location positioning system is developed for localization in large scale indoor environments. It utilizes the intelligent mobile terminal to collect the Wifi signal as the input of positioning system. Then the positioning software platform can effectively output the2-dimension coordinate of the targets and achieves an average positioning error of2.86m for realistic application. Moreover, based on extensive experiments in kinds of indoor environments, the function of this positioning system has also been extended by specific modules to solve certain problems in the positioning process.At last, two new convex programming based node localization algorithms are also proposed in WSN. The W-SDP algorithm is suitable for high-precision node localization in small scale WSN while the W-SOCP is used for high-density and large scale WSN. By exploiting the prior information of the distance measurement error between different node pairs, the objective function of the node localization problem is optimized by designed weighting coefficients. Moreover, for certain kinds of nodes, some two-hop distance constraints are also added in corresponding objective function. The simulations demonstrate that without increasing computational complexity, the W-SDP and W-SOCP can effectively reduce the node localization error than existing SDP and SOCP node localization algorithms, respectively.