Cooperative Localization For Future Wireless Networks

Various applications and services based on localization have been very popular recently, as the booming development of mobile internet technologies. The wireless localization has become one of the most important applications for next generation networks. As the popularization of smart terminals, heterogeneous networks, such as mobile cellular networks and wireless local area networks, provide more potential technical supports for the wireless localization of mobile terminals. This thesis utilizes the new features of next generation networks, including the peer-to-peer communications between smart terminals, to investigate on the theory and technologies of cooperative localization for multiple terminals. Furthermore, several high performance algorithms are designed. Some detailed contributions are listed as follows.To address the low location accuracy problem in the ill-condition environments that exist some non light of sight errors or have insufficient reference nodes, this thesis utilizes the distance measurements between mobile terminals with the peer-to-peer communications to perform the multiple terminal cooperative localization. Moreover, this thesis mainly focuses on two kinds of cooperative localization including the algorithms based on probability information and the algorithms based on non-probability information.Firstly, the factor graph concept is introduced into the probability-based cooperative localization using the probability information in the measurement of wireless parameters. The frames of inter factor graph and intra factor graph are constructed. And the cooperative localization model based on factor graph is realized through the iterative propagation of belief information. Considering the diverse of information reliability between terminals and base stations, the information from cooperative nodes are weighted. Then the cooperative localization algorithm based on weighted factor graph is proposed. Simulation results indicate the low complexity, high accuracy and high robustness of the proposed algorithm.Secondly, as to the distributed and centralized methods of non-probability cooperative localization, the Cramer-Rao lower bound (CRLB) is introduced to analyses their different models and their performance lower bounds in theory. Considering that distributed algorithms normally achieve worse performance than the centralized algorithms, this thesis proposes the unique CRLB for distributed cooperative localization under the impact from position ambiguity of cooperative terminals. By comparing the performance between distributed and centralized algorithms, this thesis, for the first time, proves that distributed algorithms enjoy the identical performance lower bound with centralized algorithm after infinite iterations. Furthermore, the position ambiguity problem is considered as the main limitation for the performance gap between distributed algorithms and centralized algorithms, which is crucial to the significant improvement for the location accuracy of distributed cooperative localization.Finally, considering the position ambiguity problem of cooperative terminals, this thesis proposes the element-wise-weighted total least-squares (EW-TLS) cooperative localization algorithms. The EW-TLS algorithm is the consistent estimator for distributed localization, considering the position error of cooperative terminals and the distance measurement error simultaneously. The algorithm can achieve the optimal solution under the impact form position ambiguity. Simulation results indicate that the EW-TLS algorithm suppresses the position ambiguity impact and achieves improvement on location accuracy and robustness.