Cooperative Wireless Localization In Heterogeneous Network

With the evolution of mobile communication network, heterogeneous cooperation has become a significant feature in the fifth generation (5G) mobile communication. Heterogeneous and cooperative network (HCN) not only satisfies the sharply increasing data requirement, but brings the unprecedented opportunities and challenges to the wireless localization. In order to achieve the high-accuracy localization in 5G mobile communication, this dissertation makes an in-depth localization research on the topology requirement, performance extreme analysis, high-accuracy and low complexity algorithm design, and the position-based applications, under the heterogeous cooperation. The main contributions of this dissertation are summarized as follows.(1) Heterogeneously cooperative localization based on rigid graph (RG) theory and factor graph (FG) theory:On the basis of RG theory, we derive the necessary condition for unique localization of heterogeneous network (HN) with topology-class and local-topology information; The minimum localization error, i.e., Cramer-Rao Lower Bound (CRLB), in HN is calculated by Fisher matrix; Based on FG theory, we propose a topology-robust cooperative localization algorithm. The simulation results verify that the localization accuracy of the proposed algorithm improve 27.1% than that of the typical Taylor-series expansion algorithm (TSE) and its convergence probability reaches 1.7 times of TSE, in rigid network topology and larger measurement error (ME).(2) Heterogeneously cooperative localization with the position errors of reference nodes (PERN):In order to analyze the nonlinear relationship between PERN and ME following independent but non-identical distribution, we perform the localization by calculating the Frobenius norm of perturbation matrix and minimizing local cost function. The simulation results show that the localization accuracy approximates the CRLB; Moreover, PERN and ME can be considered as two independent vectors. By analyzing the statistic property of the sum vector, we propose an error-vector-sum weighted FG-based localization algorithm (EVSWFG). The simulation results verify that the localization accuracy of EVSWFG is higher than other typical localization algorithms by at least 36.2% and its convergence probability reaches at least 1.75 times of other typical localization algorithms, in rigid network topology and larger PERN.(3) The selection of reference node in HCN:In a typical HCN, e.g. device-to-device communication underlaying cellular network, we propose an optimal multi-cell joint reference-node selection algorithm and a suboptimal single-cell reference-node selection algorithm by Perron-Frobenious Theorem and eigenvalue analysis, in order to perform the reference node selection given a high network coverage; Moreover, we derive the close-form expression of HCN coverage probability by the stochastic geometry theory and analyze the influence of reference node density on network performance.(4) The application of the position-based multi-hop transmission:In order to solve the problems of high exchange overhead in heterogeous cooperation, we propose multi-hop uncoordinated cooperative transmission; By the stochastic geometry theory, the network coverage probability and the expected number of transmissions for three typical scenarios are derived respectively; Then, the optimal transmission function is obtained using Lagrangian multiplier method and simulations; On the basis of three scenario analysis, we can conclude that transmission collision, resulting from PERN, is one of the main casuses to affect the network performance.Thus, the above work can provid a theoretic guide and a performance evaluation standard for achieving high-accuracy localization in 5G mobile communication. Moreover, the application of the position-based multi-hop transmission gives a new approach to solving the high exchange overhead in 5G network.