Research Of Channel Estimation Scheme Based On Clustering And Decision Feedback In OFDM System

The orthogonal frequency division multiplexing (OFDM) technology, which is the coretechnology of4G, has high spectrum efficiency and ability of anti-disturbance, originated in themulti carrier modulation (MCM) technology. OFDM is such a kind of highly efficient transmissiontechnique that has been widely used in various communication systems, becoming a kind of basicmodulation schemes. But it also has several disadvantages that can not be ignored: the large peak toaverage ratio, large Doppler frequency shift, the highly strict requirements on the synchronizationof time and frequency, all these will exert error on the transmission of signal in system, so thechannel estimation based on OFDM is essential.This thesis firstly introduces the basic principle of OFDM system and channel characteristics,and introduces the domestic and international mainstream channel estimation technique. This thesisdeeply researches the least squares (LS) estimation algorithm, minimum mean square error (MMSE)algorithm, LMMSE algorithm and SVD algorithm, together with the simulation comparison. Theexisting channel estimation algorithm based on DFT is relatively easy to put into practice, noisewithin the protection interval is not processed, bringing certain error on channel estimation. Thethreshold de-noising algorithm based on DFT, is in a certain degree of difficulty to determine athreshold, how to determine the optimal threshold remains to be unsolved. This thesis researcheschannel estimation scheme based on clustering and proposes an improved algorithm DFT-B,effectively solve the problem of threshold selection in channel estimation based on DFT bysimulation. The traditional channel estimation based on DFT is not ideal in the non-sample spacedchannel estimation. This thesis proposes an improved scheme based on clustering of channelestimation, Its performance is close channel estimation algorithm based on LMMSE but greatlyreduces the computation complexity by simulation. Then, this thesis also researches channelestimation scheme based on decision feedback, the merits of feedback in decision-feedback systemdeeply depends on the accuracy of feedback results.For this undesirable phenomenon of denoisingnoise, an improved time-domain filtering scheme DFT-B is proposed to effectively improve thesystem performance and its SNR can upgrade about2dB, as BER of0.01, compared with traditionalchannel estimation based on DFT.