The relay MIMO(Multiple-Input Miltiple-Output) technique, which is a combination of MIMO and relay, can further improve the spectral efficiency, link reliability and coverage of cells. It is efficient to provide hotspot coverages for future wireless communications. The multicell MIMO technique can effectively combat intercell interference(ICI),increase the spectral efficiency of the system and cell-edge users and expand the coverage of systems by allowing the coordination among multiple MIMO nodes via backhaul links. And it is an effective measure to provide wide-area coverages for future wireless communications. The above two enhanced coverage techniques have become more and more attractive for the wireless communications. With the availability of Channel State Information(CSI) at the transmitter, the performance of relay/ multicell MIMO systems can be improved significantly. However, due to the following factors such as time-varying of the wireless channel, accuracy of the channel estimation and reliability of the feedback channel, the transmitter can not obtain the ideal CSI. Therefore, it is not only necessary,but also has practical application values for having an in-depth study on key techniques for relay/multicell MIMO systems with imperfect CSI. The concrete contents and the most contributions are as follows.For the limited feedback joint precoding design problem of Frequency Division Duplex(FDD) single-user spatial multiplexing Amplify-and-Forward(AF) MIMO relay systems, a set of practical solutions, including the performace optimization criterion, the codeword selection scheme and the codebook construction method, are proposed. Firstly,the proposed criterion based on maximization of minimum post-processing Signal-toInterference and Noise Ratio(SINR) can guarantee the Bit Error Ratio(BER) performance. Based on the max-min SINR criterion, the optimal and the suboptimal joint precoding structures are proposed. And then a practical limited feedback joint precoding design scheme is proposed, which is based on the suboptimal structure. For the high computation problem of calculating the minimum post-processing SINR directly, a lower bound on this SINR is derived. Based on this SINR lower bound, a joint codeword selection(JCS) scheme at the destination, which relies on the two hop channels, is proposed.The JCS scheme has the optimal performance. For the high computation and feedback latency problems of the JCS scheme, an independent codeword selection(ICS) scheme,whose codeword selections can be carried out concurrently at the relay and destination, is proposed. The ICS scheme can significantly reduce the feedback latency and computational complexity, while displaying a BER performance close to that of the JCS scheme.To improve the system BER performance further, a codebook construction method based on the Generalized Lloyd Algorithm(GLA) is proposed, which has better chordal distance properties than that based on the Grassmannian Subspace Packing(GSP).For the robust joint transceiver design problem of Time Division Duplex(TDD) multiuser AF MIMO relay downlink systems, considering the practical case that the Base Station(BS) only knows the first hop CSI while the relay knows the CSI of two hops, a suboptimal SVD-SMSE joint robust design scheme is proposed, which can effectively combat the channel estimation error and the channel reciprocity delay effects. With regard to the BS precoder designing, a beamforming scheme based on the first hop channel Singular Value Decomposition(SVD) and a uniform power allocation scheme are proposed, which can avoid the feedback overhead of the second hop CSI. For the relay forwarding matrix and user equalizers, a joint robust design problem, which is based on minimizing the Sum Mean Square Error(SMSE) of all users with a constraint on the relay total transmit power,is established. By using KKT(Karush-Kuhn-Tucker) conditions, a closed-form solution to this problem is achieved. Simulation results verify the effectiveness and robustness of the proposed suboptimal scheme.For the robust relay beamforming design problem of TDD distributed AF MIMO relay aided multiuser peer-to-peer communication systems, based on the SINR performance measure, two optimal robust design schemes are first proposed, which can effectively combat the stochastic CSI errors. One is aimed at balancing the SINR at destinations under the relay total transmit power constraint concering the fairnesss, the other is to minimize the relay total transmit power with the guarantee of Qo S(Quality of Service) requirements(SINRs exceed preassigned thresholds) at destinations concerning the efficiency. Based on the Semidefinite Relaxation(SDR) technique, the above NP-hard problems can be converted into the Semidefinite Program(SDP) problems which can be solved efficiently through the interior-point method. Considering the complexity of the SDP problems is proportional to the fourth power of the number of relay antennas, by applying the ZeroForcing(ZF) constraint into the beamforming vector design, two suboptimal robust design schemes are further proposed, which are corresponding to the above optimal designs and can significantly reduce the computational complexity. The robustness and effectiveness of all proposed designs are demonstrated through simulations. Moreover, simulation results provide guidelines for choosing the concrete robust beamforming design, which is determined by the number of relay antennas and the size of CSI errors.For FDD limited feedback clustered coordinated multicell transmission systems, a joint user feedback bit allocation(FBA) and BS power allocation(PA) strategy is proposed, which can solve the problem that the conventional FBA strategies adopted only the equal PA and thus could not get the macro-diversity. Considering that the average sum-rate expression is so amenable to analysis, a modified performance measure — the virtual sum rate is adopted, which can reduce the difficulty of the derivation and analysis. Based on the derived virtual SINR lower bound, a joint FBA and PA problem, which is based on maximizing the virtual sum rate lower bound under constraints on the total number of feedback bits per user and the total transmit power per BS, is set up. For the tight coupling problem between the FBA and the PA, a solution with coordinated ascent searching method is proposed to solve the FBA subproblem and the PA subproblem alternatively. For the two subproblems, a waterfilling closed-form solution and an iterative waterfilling PA algorithm are presented, respectively. Furthermore, the alternating iterative algorithm is proposed and its convergence analysis is presented. Simultion results show that the proposed strategy performs better than the conventional FBA strategies with only equal power assumptions. |