Key Transmission Techniques For Cooperative Communications With Frequency-selective Channels

By sharing antennas of neighboring users in a cooperative manner to construct a vir-tual multiple-antenna environment, cooperative communication systems can combat thesignal fading in wireless propagation effectively. Without increasing the hardware com-plexity, the same spatial diversity gains of multiple-antenna systems can be achieved in asingle-antenna based cooperative system. Since the cooperative technique improves thereliability and bandwidth efficiency of communication systems, it becomes an importantresearch area for the wireless communications. The low complexity multiple parame-ters synchronization algorithms at destination node, multiple Carrier Frequency Offsets(CFOs) robust data detection algorithms, pilot design for optimizing channel estimation,and power allocation algorithms of relay nodes with distributed property are the key andchallenging techniques for cooperative communication research. Thus, this dissertationfocuses on these practically key techniques in the presence of frequency-selective chan-nels.Consideringthe characteristicsofCyclicPrefix(CP)block transmissionbased coop-erative communication systems with frequency-selective channels, this dissertation pro-poses a Frequency Division Multiplexing (FDM) based tile-structure training sequenceand low complexity estimation algorithms correspondingly, which have addressed themultiple synchronization parameters estimation problem at the destination node. Due tothe distributed property of relay nodes, there are multiple Timing Offsets (TOs) and Car-rier Frequency Offsets (CFOs) presenting in the received signals at destination node. Ex-ploiting the time domain redundancy of CP, the CP block transmission based cooperativesystems can combat the multiple TOs among relay nodes in the presence of frequency-selective channels. However, for large transmission delay cooperative systems, the extralong CP reduces the bandwidth efficiency. Moreover, uncompensated multiple CFOs de-crease the systems'performance significantly. For this case, the estimates of the relaynodes'TO and CFO are needed for the compensation procedure. Thus, multiple TOs andCFOs estimation is an important issue which needs to be solved firstly in the deploymentof cooperative systems. To achieve a good tradeoff between computational complex-ity and bandwidth efficiency, a tile-structure based training sequence is proposed in thisdissertation. Employing the frequency-domain orthogonality and time-domain repetitiveproperty of proposed tile-structure training sequences, the destination node can estimate multiple parameters in one training sequence period via using a cross-correlation typetiming estimation algorithm and signal subspace decomposition based CFOs estimationalgorithm. Proposed synchronization methods have overcame the low bandwidth effi-ciency of Time Division Multiplexing (TDM) based synchronization schemes and highcomputationally complexity of the conventional FDM based synchronization algorithms.By judiciously designing the size of the tile, proposed algorithms are shown to have bettersynchronization and Bit Error Rate (BER) performance than the conventional synchro-nization methods.For the cooperative space-time and space-frequency coding systems, where feed-back synchronization mechanism is useless, multiple CFOs interference cancellation datadetection algorithms are proposed in this dissertation to mitigate the multiple CFOs in-terference with low computationally complexity. Employing the feedback synchroniza-tion mechanism, relay nodes can adjust their own transmitting windows and carrier fre-quencies so that the relay nodes'signals arrive at destination node with synchronous TOsand CFOs. However, for the cooperative Orthogonal Frequency-Division Multiplexing(OFDM) systems without feedback synchronization mechanism, uncompensated multi-pleCFOswillcauseInter-SymbolInterference(ISI)andInter-CarrierInterference(ICI)atdestination node, and result in performance degradation, even destroy the diversity gainsof cooperative systems. For cooperative space-time block coded (STBC) systems in thepresence of multiple CFOs, Enhanced Iterative Maximum Likelihood (EIML) and Ad-vance Enhanced Iterative Maximum Likelihood (AEIML) data detection algorithms areinvestigatedinthisdissertation. BothproposedEIMLandAEIMLdetectorscangetridofinter-carrier-interference (ICI) caused by multiple CFOs in STBC systems perfectly viaa simple matrix multiplication. The residual inter-symbol-interference (ISI) is removedby performing iterative symbol detection. Simulation shows that proposed EIML andAEIML detectors not only outperform the conventional Iterative Maximum Likelihood(IML) and Minimum Mean-Square Error (MMSE) detectors significantly, but also avoidthe large matrix inversion required by MMSE detector. Moreover, to mitigate the inter-ference caused by multiple CFOs in cooperative Space-Frequency Block Coding (SFBC)systems, Ordered-Successive Parallel Interference Cancellation (OSPIC) and complexityreducedOSPIC(CR-OSPIC) algorithmsareproposed. By performing orderedsuccessiveand parallel interference cancellation detection successively, proposed OSPIC and CR-OSPIC detectors can reduce the inter-block-interference (IBI) caused by multiple CFOseffectively. Without large scale matrix inversion and iterative operation, proposed algo- rithms have both detection performance and complexity gains compared to conventionaldata detection methods in cooperative SFBC systems.For OFDM based cooperative systems in the presence of virtual subcarriers, pilotschemes with optimal channel estimation performance are proposed in this dissertation.Proposed pilot schemes have overcame the problem that channel estimation with conven-tional pilot schemes suffers large performance degradation under the virtual subcarriersscenario. Employing CP based OFDM transmission scheme, cooperative communica-tion systems not only can combat multiple transmission delays, but also can overcomethe equalization problem under frequency-selective channels. However, to avoid data be-ing distorted by non-ideal filters, some carriers at the spectrum edges of each symbol aredeactivated in practical OFDM based cooperative systems. These deactivated virtual sub-carriers break the equal-powered equal-spaced pilot structure for the channel estimationof fully loaded OFDM systems and will cause severe performance degradation on pilot-based channel estimation. Existing pilot schemes designed for the virtual subcarriers casehave a large channel estimation performance degradation compared to that of the optimalpilot sequences. To solve this problem and consider the frequency domain equalizationproperty of OFDM systems, a new criterion, which minimizes the channel estimationMean Square Error (MSE) on data subcarriers, is proposed in this dissertation. Unlikeconventional pilot design using minimizing channel estimates MSE as design criterion,proposed criterion only optimizes the frequency domain channel estimates on the datacarriers, which is more proper for the virtual subcarriers scenario. Employing proposedcriterion and disjoint pilot sequence, a closed-form expression for the optimal power dis-tribution and an effective suboptimal placement solution are provided. The simulationresults show that the channel estimation performance of proposed disjoint pilot designhas 4-20dB gains over that of conventional partially equal-spaced pilot schemes underfrequency-selective channels.Considering the characteristics that the relay nodes in cooperative systems are dis-tributed in space independently and limited in energy, the Cooperative Transmit Beam-forming (CTB) algorithms under Per-antenna Power Constraint (PPC) are studied in thisdissertation. Proposed PPC CTB schemes avoid the practical drawbacks of Total PowerConstraint (TPC) CTB schemes, i.e. high-power amplifier design perspective and lownetwork lifetime, since the powers allocated to different antennas may considerably varyover time under TPC. For the flat fading case, PPC beamforming design is equivalent tothephaseoptimizationproblemsincetheoptimizationobjectonlycontainsoneparameter. On the contrary, PPC beamforming design under frequency-selective channels has highcomputationally complexity and cannot find a closed-form solution. In this dissertation,the PPC transmit beamforming coefficients are obtained by employing convex optimiza-tiontoolsfirstlyfortheCPbasedOFDMandSC-FDECTBsystem. Toovercomethehighcomputationallycomplexityandlimitedaccuracyproblemsofconvexoptimizationmeth-ods, three suboptimal PPC beamforming schemes are proposed under different scenarios.Moreover, proposedsuboptimalPPCschemesareextendedtotheCTBsystemswherethedestination node is equipped with multiple receiving antennas. With closed-form expres-sions, proposed suboptimal PPC solutions have much lower complexity than the convexoptimization algorithms. It is shown in the simulation that suboptimal PPC solutions canhave a close performance to the TPC under different power allocation scenarios.To reduce the large feedback overhead required by feeding back the Channel StateInformation (CSI) to the transmitters in CTB systems under frequency-selective chan-nels, a low rate time domain CSI feedback method is proposed. With bandwidth limitedfeedback channels, the performance of CTB systems is dominated by the accuracy ofbeamforming vectors used at transmitters. Existing research results show that the VectorQuantization (VQ) algorithms can solve the limited feedback problem effectively withflat fading channels. However, extending existing VQ methods to frequency-selectivechannels will require a large amount of feedback. Noting that the length of time domainCSIismuchsmallerthanthelengthofdatablock, timedomainCSIVQfeedbackschemesare studied in this dissertation. By analyzing the affection of phase and amplitude parts ofCSI on the performance of CTB systems, a non-uniform feedback scheme is proposed tofurther improve the feedback performance. Under the total feedback rate constraint, pro-posed non-uniform feedback scheme can achieve the best feedback performance by judi-ciously allocating the feedback rate on phase and amplitude parts of CSI nonuniformly.Simulation results show that proposed time domain feedback method outperforms bothcluster based and interpolation based feedback schemes significantly. Combining withthe PPC coefficients design algorithms studied previously, the CTB schemes proposed inthis dissertation can achieve a good tradeoff between performance and feedback rate.