A Study On Iterative Receive Technology For OFDM And MIMO Systems

Multiple-Input Multiple-Output (MIMO) technology can increase the channelcapacity of wireless communication systems in flat fading channels; moreover, it canenhance the transmissionreliability with diversity gain. Orthogonal FrequencyDivisionMultiplexing (OFDM) technology is robust to the frequency selective fading in thebroadband wireless multipath channel. The combination of MIMO technology andOFDM technology can greatly expand the application area of MIMO and make thesystem work well in frequency selective fading channels. Iterative receiver with Turbosignal processing technology can obtain near optimal performance with manageablecomplexity in wireless communication systems. This thesis investigates the key issuesin iterative receiver for OFDM,MIMO and MIMO–OFDM systems based on thesoft-information processing technology, including the signal detection, channelestimationandcarrierfrequencyoffset(CFO)estimation.In the first part of this thesis, we discuss mutual information transfer properties ofthe Soft-In Soft-Out (SISO) components in the Turbo receiver with the ExtrinsicInformation Transfer (EXIT) chart. Two soft detection algorithms are proposed forGroup-wise Space-Time Block Coding (G-STBC) systems. The convergence behaviorof the corresponding Turbo receiver is studied via the EXIT chart analysis when theproposed detectors areused with Turbo, LDPC and convolutional code. Based ontheseanalyses,thesoftdetectoranddecoder withmatchmutualinformationproperties canbeselectedtoconstructtheeffectiveiterativereceiver.The second part of the thesis is devoted to the SISOdetection for MIMO systemsin flat fading channel. An improved list sphere decoder with Maximum a posterioriprobability search (MAP-LSD) is proposed.The new pre-processing stage and symbolsorting method are designed to control the increase in complexity due to the a prioriinformation with large value range. The clipping threshold for the extrinsiclog-likelihood ratios (LLRs) is chosen based on the EXIT chart analysis. A scalingfactor is used to correct the magnification effect due to the approximation calculationand positive feedback to compensate the performance loss.Theoretical analysis andsimulation results show that the proposed algorithm can reduce the complexity of thetranditionalMAP-LSD andimprovethedetectionperformance.The third part of the thesis studies the channel estimation for OFDM systems. Weproposeanewadaptivedual-modechannelestimatorbasedoniterativesoftinformationprocessing.Toavoid the performance degradation due to the soft information with lowreliability in the soft-decision directed (SDD) channel estimation, a new reliabilitymeasure function is constructed to calculate the reliability factor for the reconstructedsignal. An option is made adaptively between the decision-directed mode and EM (Expectation-Maximization) estimation mode.The combination of the two estimationmodes can reduce the pilot overhead and track the time-varying channel effectively.Moreover, the existing APP (A Posteriori Probability) channel estimation is extendedbased on the adaptive SISO algorithm. A bi-directional channel estimator in timedomain is used instead of the traditional forward-only wiener predictor based on thetime-reversedprocess analysis.Theoretical analysisandsimulationresultsshowthattheproposed bi-directional estimation algorithm has better theoretical completeness thantheforward-onlyestimation anditis morerobusttothetime-varyingchannel.The fourth part of the thesis studies the channel estimation for MIMO-OFDMsystems. A new Bayesian EM (BEM) channel estimator is proposed. Using thecandidates list from the LSD and the a priori information from the decoder, theapproximation of the soft information in the conventional EM channel estimator ismodified, the precision of the statistic information of the soft symbol decisions areimproved and a new maximum a posteriori (MAP) estimation of the channel impulseresponse is proposed. Since the new BEM algorithm makes full use of statisticalinformation, it has better performance than the traditional EM algorithms.A new jointsymboldetectionandchannelestimationalgorithmisproposedbasedon the VariationalBayesian Expectation-Maximization (VBEM) algorithm. Compared with the EMalgorithm whichproduces point estimate,the VBEM algorithm can providedistributionestimates,thus the VBEM-based algorithm considers more statistical information forthe channel and soft symbols. Simulation results demonstrate that the proposedalgorithmhasbetterbit-errorrateperformanceovertheconventionalEM techniques.In the end, we consider the estimation of the time-varying channel and CFOparameters in mobile communication. A new channel estimationand trackingalgorithmis proposed for MIMO-OFDM systems over the time-varying multipath fading channel.The recursive EM algorithm and Kalman predictor are combined to track thetime-varying channel. The channel frequency response of each subcarrier is estimatedby the recursive EM algorithm which uses sequential processing over all the symbolcombinations in the candidate list from the LSD. Moreover, with the second-orderstatistics of the fading channel, the channel impulse responses are tracked by a Kalmanpredictor.The new algorithm executes recursive processing in both space-frequencydomain and time domain, and it combines the channel statistical information with thesoftsymbol information to provide goodestimation performance with comparative lesspilot symbols. A new algorithm is proposed to estimate the time-varying channel andCFO parameters in MIMO-OFDM systems.The processing steps are given.In thestart-up stage, soft symbols reconstructed from the SISO decoder and the MaclaurinexpansionapproximationareusedtoestimatetheCFO,thenMAPchannelestimationisfollowed. In the later stage, CFO estimation is refined withiteration; Taylor expansion approximation based on the previous estimated CFO is used.Simulation results showthat the proposed algorithm caneffectively estimate the time-varying channel and CFOparameterstoachievegoodsystemperformance.