MMSE Equalization And Channel Estimation For CP-FREE OFDM System

OFDM system has been widely applied in LTE system due to its great performance in high speed transmission.The cyclic prefix(CP)plays a crucial role in OFDM,which solves the problem of Inter-symbol interference(ISI)and Inter-carrier interference(ICI)and leads to the great performance.However,sometimes the CP occupies about 20% bandwidth which is a huge cost for the system.And with the need for a higher speed in the future,adding CP might not be a very pleasant method.Many researchers have made research on CP-FREE OFDM system,and there is one system named SCSE-OFDM system,which was proposed by Xiqing Liu.In SCSE-OFDM system,Decision Feedback Equalization is utilized to combat with ISI.Besides,a CP-restoration process is used to deal with ICI.Nevertheless,in this system the author only considered Zero Forcing for equalization which might cause the amplification of noise,and channel estimation was also omitted.To improve the performance and make the system complete,in this paper I will work on several aspects.To avoid the noise amplification by Zero Forcing,in this paper the MMSE equalization which focuses merely the channel noise is first considered.Then based on Ghosh's assumption that the decision feedback error can be considered as Gaussian distributed,MMSE equalization taking feedback error is also considered.To make certain the feature of the feedback error,a simulation on it is made and I found that the error more likely follows Alpha stable distribution,which is impulsive and heavy tailed.And to attenuate effect of feedback error,a threshold is set in the feedback loop,which reduce the possibility of huge feedback error,thus the performance is improved.Since in SCSE-OFDM system,hard decision was applied in the receiver,to further improve the performance,Turbo equalization is considered in CP-FREE OFDM system.In the Turbo equalization at the receiver part,the equalization process is combined with Turbo decoder.External information is transmitted in the loop between the MMSE equalization and the decoder.Taking advantage of the superior performance of Turbo coding,the performance can be improved greatly.To make the system more complete,in this thesis I studied on channel estimation for the CP-FREE OFDM system.First,under the assumption that pilots are used,two channel estimation methods based on LS and LMMSE are designed.Then,channel estimation method requiring no pilots is also studied,which is called subspace based channel estimation.Using techniques such as oversampling and repetition index,channel estimation is successful without the aid of pilots.Finally,considering a scenario where no much prior information on the channel can be attained,an adaptive filter based channel estimation method is studied.Specifically,in the thesis signal is considered to be surrounded by Alpha stable distributed noise.With help of RLP and LMP,the accuracy of channel estimation is guaranteed.