Studies Of Distributed Space Time Coding And Beamforming Techniques In Cooperative Networks

Through sharing antennas of neighboring users in a cooperative manner to constructa virtual multiple-antennas environment, the spatial diversity can be achieved in a single-antenna cooperative system without increasing the hardware complexity. Since the coop-erative technique can efficiently combat the fading effect and improve the reliability ofcommunication systems, it becomes an important research area for the wireless commu-nications. Amplify-and-forward protocol not only has low complexity but also achievegood spatial diversity by employing distributed space time coding (DSTC) and beam-forming techniques at relay nodes, which attracts much attention these days. This dis-sertation firstly focuses on one-way relay network (OWRN) without channel state infor-mation (CSI) at the nodes and studies the DSTC scheme for asynchronous systems underfrequency-selective channels, then focuses on OWRN with CSI at the nodes and proposesthe robust beamforming algorithm under flat fading channels as well as the beamformingdesign under frequency-selective channels, and finally studies the beamfoming techniquein two-way relay network (TWRN) with CSI at the nodes.Employing DSTC at relay nodes can achieve high bandwidth efficiency along withfull diversity. However, the time offsets among the relay nodes and multipath fadingbetween nodes will destroy the structure of DSTC, which may drastically undermine thediversity potential of cooperative networks. This dissertation proposes a novel DSTCtransmissionschemebasedonOFDMtechnique,whichcanobtainthefullspatialdiversityeven with the existence of timing errors and multipath fading. Moreover, all DSTCs forsynchronouscooperativesystemscanbeappliedtotheproposedscheme. Furthermore,forthe systems with two relays, we employ subcarrier grouping and precoding at the sourcenode in the proposed scheme to achieve full spatial diversity along with full multipathdiversity.In practice, the CSI is usually imperfect due to factors like estimation error, feedbackdelay, etc. The performance of the designs based on the assumption of perfect CSI willbe degraded in the presence of CSI imperfections. Based on worst-case design, a robustbeamforming design problem can be formulated, which can guarantee good performancefor all the channel realization in the uncertainty set but is difficult to solve due to the in-finity of constraints. This dissertation proposes robust beamforming algorithm based onthe worst-case design corresponding to the criterion of maximizing the SNR at the des- tination, subject to TPC and PPC. Based on extended S-lemma and Schur complementtheorem, it is shown that this originally intractable problem can be transformed to a quasiconvexproblem, which can be solved efficiently by solving a convex problem with SOCand LMI constraint at each step. Simulation results show that the proposed algorithm canefficiently resist the performance loss in terms of outage probability.Employing OFDM transmission, cooperative systems can compensate for the multipatheffects effectively. Applying beamforming weights in the time domain (TD) by usingcyclic BF filters requires lower complexity and less feedback from the destination to performBF. However, the existing time domain beamforming algorithm can not mitigate theperformance degradation caused by deep fading at some subcarrier. To cope with thisproblem, the dissertation designs the BF vectors by maximizing the minimum SNR overall subcarriers at the destination, first TPC and then PPC. The design under this criterioncan guarantee good performance at all subcarrier but the constraints of this design problemhas fractional quadratic functions which are non-convex. To solve this problem efficiently,the dissertation firstly using the semidefinite relaxation (SDR) technique to approximatelysolve this problem by a convex semi-definite program (SDP) problem, then obtain thebeamforming vector by randomization technique. The SDR approach converts the vectoroptimization to matrix optimization, which incurs high computational complexity whenthe TD filter length is long. To address the complexity concern, an iterative method is proposedby solving a SOCP problem at each iteration. This algorithm directly optimize theTD filter coefficient and has lower complexity. Simulation results show that SDR methodperforms better than the iterative method when the filter length is relatively short. When alonger TD filter is employed, the iterative method outperforms SDR method. Finally, westudy the influence of initial approximation to the performance and proposes a searchingalgorithm for the appropriate initial approximation. Simulation results illustrate that it canprovide an SNR improvement compared to the iterative method under TPC.Generally, TWRN can provide spectral efficiency improvement compared to OWRN.However, most of the beamforming designs are focusing on good SNR performance,which can not guarantee great performance in terms of sum-rate. This dissertation studiesthe beamforming design to maximize the sum-rate under total relay power constraint(TRPC). We firstly focus on reciprocal channels and propose a beamforming design basedon branch and bound algorithm and prove that it can obtain a global optimal solution. Simulationresults show that it outperforms existing methods. Further, a suboptimal algorithmwith lower complexity is proposed, where only one variable need to be optimized. It is illustrated that the suboptimal solution performs closely to the branch and bound algorithmwhen the number of relay nodes is small. Finally, the branch and bound algorithmand suboptimal algorithm is applied to the design problem under nonreciprocal channels.Simulation results show that the branch and bound algorithm still have the best performance.The suboptimal solution performs better than existing algorithm when the powerof relay nodes is relatively low.