Channel Estimation In MIMO-OFDM Wireless Communication System

Multiple-input multiple-output (MIMO) techniques can be implemented to obtain acapacity gain in rich scattering environment without increasing the bandwidth or trans-mitted power. Orthogonal frequency division multiplexing(OFDM) converts a frequencyselective channel into a parallel number of frequency ?at subchannels. The subcarri-ers have the minimum frequency separation required to maintain orthogonality of theircorresponding time domain waveforms, yet the signal spectra corresponding to the dif-ferent subcarriers overlap in frequency. Hence, the available bandwidth is used verye?ciently.OFDM as a multi-carrier modulation technique, is e?ective for supporting highspeed transmission as well as combating multipath fading in broadband wireless commu-nications.MIMO can be used in OFDM systems to not only o?er the potential of highspectral e?ciency and the promising to exploit maximum spatial resources,but also lessenthe severe e?ects of frequency selective fading.Therefore,MIMO-OFDM is the one of keytechniques of the next generation mobile communications system.This dissertation has mainly made further research on the related theories about threemain aspects:channel capacity,High Peak to Average Power Ratio(PAPR) and channelestimation for MIMO-OFDM systems , and a few new conclusions were derived in thedissertation.Moreover,a number of new useful experimental results were obtained by largenumbers of MATLAB simulations.The main contents of the dissertation are expressed as follows:1.The capacity of MIMO and MIMO-OFDM systems is a key issue for 4G,it is veryimportant for optimal wireless resource management of MIMO and MIMO-OFDM sys- tems.At present, a variety of analysis methods of the capacity have existed for MIMO sys-tems,but,the research of the capacity is not enough for MIMO-OFDM systems.In this dis-sertation,the capacity is analyzed for MIMO systems in ?at-fading channels,first.BecauseOFDM divides a broadband channel into a number of narrow-band channels,the frequencyselective fading channels can be seen as ?at fading channels on every subcarriers of OFDMsystems,the capacity of MIMO-OFDM systems on the frequency selective fading channelscan be studied by the capacity analysis method on ?at fading channels for MIMO sys-tems.in addition,the capacity is compared between MIMO and MIMO-OFDM systems.2.High PAPR of the transmitted signal is a major drawback in OFDM and MIMO-OFDM systems.Selected mapping (SLM) is very attractive because of without distortionof OFDM and MIMO-OFDM signal in a variety of methods to reduce PAPR. In thisdissertation, a named MSLM improved SLM technique for OFDM signal is proposed. Inthis technique, we divide the transmitted signal into multi-groups which the data length issame for every group at transmitted end of OFDM systems,then perform SLM.Comparingwith conventional SLM-OFDM systems,the MSLM method has two advantages,one isthat PAPR is reduced further,other is that computational complexity is decreased,too. Inaddition,we applied MSLM method to MIMO-OFDM systems, simulation results showthat good PAPR performance can achieve in MIMO-OFDM systems,and symbol errorratio performance is better,too.3.Considering intersymbol interference(ISI) in wireless fading multipath channel, anew time domain least square(LS) channel estimation using special training sequencesfor MIMO-OFDM systems is proposed. In this algorithm,using a constant envelope ofexponential sequence by inserting zeros as training symbols, the training symbols hascharacteristics of desirable PAPR,eliminating the in?uence of ISI,and assuring orthog-onality for every antenna training symbols. simulation results show that the proposedalgorithm(inserting zeros) outperforms conventional algorithm(no inserting zeros) withrespect to the MSE by 10dB if the SNR is 25dB when using constant envelope as trainingsequences.4.The channel estimation method based on DFT for OFDM is very e?ective inRayleigh ?at fading channel.This method is fit for sample spaced channel.When the chan- nel is non-sample spaced channel,transformed by the DFT, the energy of channel impulseresponse will leak to all taps. A improved channel estimation method based on DFT forOFDM systems is presented in this dissertation.The energy leakage of channel impulse re-sponse is enough calculated in this method, using a real compute of energy increasing rateinstead of complex matrix calculation, the compromise is performed between estimationperformance and complexity on non-sample spaced channel.5.In the blind channel estimation, however, do not send the training sequence, savechannel resources.The blind channel estimation based on frequency domain for MIMOsystem has been received extensive attention because they do not rely on the lengthof time-domain channel impulse response .In the dissertation ,first,a channel estimationmethod base on HOS in the frequency domain is introduced ,then,the e?ect of estimationperformance by window function is analyzed. Theoretically proved that the bispectrumestimate can be made to obtain the smallest MSE by Optimal window,and Optimal win-dow is applied to blind channel estimation base on frequency domain for MIMO systems.Our works need further studying and perfecting, such as followings.(1) same Randomphase sequence is used in every group for MSLM method,the sideband information will bereduced.(2)MSLM method will be applied to the channel estimation base on the pilot,nosideband information will be achieved by the channel estimation base on special trainingsequences.(3)the performance of blind estimation base on frequency domain with respectto the MSE will be theoretically analyzed for MIMO systems.The innovation works consist of the following main components in the paper.1.Considering high PAPR for OFDM and MIMO-OFDM systems,PAPR reductionmethod based on MSLM is presented for OFDM and MIMO-OFDM systems;2.Considering ISI in wireless fading multipath channel, a new time domain leastsquare channel estimation using special training sequences for MIMO-OFDM systems isproposed.3.A improved channel estimation method that is fit for non-sample spaced channelbased on DFT for OFDM systems is presented.4.Theoretically proved that the bispectrum estimates can be made to obtain thesmallest MSE by Optimal window,and Optimal window will be applied to blind channel estimation base on frequency domain for MIMO systems.