Research On Key Techniques Of Linear Non-Regenerative Two-Way Relaying Cooperative Communications

Compared with one-way relaying (OWR), two-way relaying (TWR) can efficientlyutilize the spectrum and can improve the data rates. It is growing as a new researchregime in the cooperative communications. Relaying schemes which are based on thelinear non-regenerative (LNR) protocols are simple to design, and algorithms of thiskind are with lower-complexity, faster computation speed and less priority knowledgeof the systems than other kinds of algorithms. So, the LNR protocols are attracting moreand more attentions. Recently, three basic system models are investigated in the TWRcommunications based on LNR, which are the centralized two-way relaying networks(CTWRNs), the distributed two-way relaying networks (DTWRNs) and the multi-pairtwo-way relaying networks (MPTWRNs), respectively. In this thesis, the research focuson these three models, and the relaying schemes and their corresponding signalprocessing techniques are studied carefully.A general system model of the TWR communications system based on the LNRprotocol is presented. The signal transmission and processing flow in the two-wayrelaying networks (TWRNs) are depicted and modeled in details. Some general systemparameters and performance metrics are defined. This part constructs a basement of thewhole work.Relaying schemes which are designed based on the beamfoming techniques arestudied in the CTWRNs. Firstly, the existing literatures are concluded and somerepresentable design methods are listed. Secondly, by assuming the channel stateinformation (CSI) can be accurately acquired by all the communication nodes (source,destination, and relay), beamforming parameters of all the nodes are jointly designed byminimizing the sum of mean square errors (MSMSE) of the source and the destination.As the optimal MSMSE design method is with high complexity, the beamfomingparameters of each node are analyzed mathematically and physically, some usefulconclusions are achieved. Moreover, a sub-optimal MSMSE beamforming designmethod is proposed, by which, the confliction between the system performance and thealgorithm complexity is properly solved. The mathematical performances of the twoalgorithms are also compared, which shows that the sub-optimal method is muchsimpler and faster to compute. Simulation results indicate that the optimal MSMSEdesign method has a best system performance in terms of average system bit error ratesand space diversity gain, but when there are3or more than3antennas at the relay node,the performance gap between the optimal and sub-optimal methods can be neglected. So,we can say that the sub-optimal MSMSE method can be applied widely in the future.Besides above all, a general signal processing flow is proposed based on the signalprocessing property of each communication node, and based on which, the beamforming parameter design methods of each node are discussed, respectively. Anew modularization beamforming design method of the LNR-based TWRcommunications is supplied.Three main problems are investigated about the DTWRNs. Firstly, a TWR systemonly with single antenna nodes is considered. Beamforming parameters are designed intwo power control scenarios, which are the relay nodes total power control and relaynodes individual power control, respectively. In the latter power control scenario, thecomplexity of the MSMSE-based optimal design method is growing dramatically versusthe relay number, and in order to avoid this, a sub-optimal method is proposed. In thesub-optimal design, the beamforming parameter of each relay node is converted to acomplex power control factor. Due to the factors rely on each other, an iterativealgorithm is carried out to numerically solve for the best solutions. Secondly, when thecommunication nodes (source nodes and destination nodes) are equipped with multipleantennas, in order to efficiently utilize the space diversity gain, the MSMSE-basedbeamforming parameters of all nodes are jointly designed under two power controlscenarios, which are as same as mentioned above. Then, the jointly beamforming designproblem in the MIMO DTWRNs is solved. Analysis results show that, when the relaytotal power is limited, the solutions of the relay beamforming parameters are just asimple transformation of the solutions in the CTWRNs, and when the relay individualpower is limited, the beamforming design problem can be equally converted to abeamforming design where all the communication nodes are with single antenna.Simulation and analysis results show that the proposed beamforming design methodrequire little control channel resources, and the design methods are easy to apply. Whenthe relay total power is limited, the MSMSE-based methods have the best performanceand when the relay individual power is limited, the sub-optimal methods also have goodperformance and also have low complexity. At the end, when the source and destinationnodes are equipped with multiple antennas, the opportunistic relaying schemes in theDTWRNs are investigated. Three opportunistic relaying schemes, which are named asantenna-relay selection based opportunistic relaying scheme (ARSORS), maximumratio beamfoming based opportunistic relaying scheme (MRBORS) and equal gainbeamforming-based opportunistic relaying scheme (EGBORS), respectively, areproposed. Through these schemes, when the source and destination are with multipleantennas, the opportunistic relaying problems are solved. Full diversity gain can beachieved by the ARSORS which is mathematically certificated. Simulations show thatthe MRBORS has best performance in terms of outage probability and average sumrates.In the MPTWRNs, three problems are studied. As the existing optimizationschemes all only focus on how to cancel the inter pair interference (IPI), and the noiseitems are neglected, we first investigate the general form of the relay beamforming (RaB) matrix which is designed based on the MSMSE criterion. A block diagonalizaton(BD) based beamforming method, in which the beamforming parameters are designedto minimize intra-pair sum of mean square errors, is carried out. This design finds asolution to the problem where the IPI cancelation and the noise equilibrium cannot besolved together. Simulations indicate that this method can cancel the IPI thoroughly andit can also balance the useful signal and the noise. It has a better system performancecompared with the MMSE-based design and the BD-based design. Compared with theMSMSE method, this method has a lower complexity and faster convergence speed.Secondly, a user beamforming-based IPI cancelation method is proposed, and theconditions which this method should satisfy before applying are also deduced. Thismethod can be applied to the system where the relay node only has limited computationcapabilities when only simple amplify-and-forward (AF) protocol is adopted.Simulation results show that, when the conditions are satisfied this method canefficiently cancel the IPIs, it is very useful when the relay nodes are simply constructedand temporarily deposed. At last, a cooperative relying scheme in the cognitive radio(CR) enviroment is investigated. In this system, the second users (SUs) are employed bythe primary users (PUs) as their communication relays, and the SUs and PUscooperatively share a certain spectrum. This forms a special kind of MPTWRNs. Whenone SU is equipped with multiple antennas, a beamforming-based spectrum sharemethod, in which the beamforming parameters are designed by minimizing the sum ofweighted mean square errors (MSWMSE), is proposed. By this method, the PUs andSUs can transmit their signals via only two time slots, and the channel resources aresaved while the system efficiency is enhanced. Due to the MSWMSE criterion, the PUscan control their performance by changing the weight factor, and the whole system isvery angile.