Channel Estimation And Phase Noise Suppression For SC-FDE Based60GHz Communication Systems

This paper studies the channel estimation and phase noise suppression algorithmfor the single-carrier with frequency domain equalization (SC-FDE) based60GHzcommunication systems. Our research work focuses on the least square (LS) andrecursive least square (RLS) channel estimation methods, the phase noise (PN)autoregressive simulation model, the impact of phase noise on SC-FDE systems andthe phase noise suppression algorithm employing low complexity channel estimation.The system model is the basis of channel estimation and phase noise suppression.Therefore, this paper introduces the SC-FDE system model impacted by PN first.To reduce the complexity of traditional channel estimation based on correlation ofGolay sequence and improve the performance of traditional unstructured channelestimation (UCE), this paper proposes the low complexity structured channelestimation (SCE) based on LS or RLS criteria. First, channel frequency response (CFR)at few frequencies are estimated by LS criteria or recursive updated by RLS criteria,then interpolated by the low complexity discrete Fourier transform (DFT) filter toobtain the whole CFR which is required by frequency domain equalization (FDE).Since the low-order diagonal matrix inversion in the frequency domain and the lowcomplexity DFT interpolation, the LS-SCE or RLS-SCE of this paper has lowcomplexity. The complexity analysis and simulation results show that our LS-SCEowns lower complexity than that of traditional channel estimation based on correlationof Golay sequence in the case of the almost same performance while betterperformance than that of traditional LS-UCE in the case of the almost same or evenlower complexity. The mean square error (MSE) of our RLS-SCE is the minimumamong above channel estimation algorithms. Compared with traditional RLS channelestimation algorithms, similar conclusions to LS-SCE can be drawn for our RLS-SCE.In the study of PN, the autoregressive (AR) model of60GHz PN meeting the“one-pole/one-zero” model of the60GHz communication standard is given at first. Aninfinite impulse response (IIR) digital filter is designed by using the bilinear transformmethod in this paper. The phase noises in the simulation can be obtained at the output of this IIR filter by filtering the Gaussian white noise. Then, the performance ofSC-FDE systems impacted by PN is analyzed and the level range of PN which causeperformance impairments too seriously to need be suppressed is summarized.For the PN causing serious performance impairments, this paper proposes a phasenoise suppression algorithm employing our low complexity channel estimation. Whenthe PN exist, the estimated CFR is impacted by common phase error (CPE) of PN. Inthis paper, we take the CPE and CFR as a whole. Firstly, the initial compensation ofPN is implemented by FDE. The CPE ratio (CPE of the data block to the CPE ofestimated channel ratio) is obtained by linearly interpolating the partial CPE ratio (CPEof cyclic prefix (CP) to the CPE of estimated channel ratio) which is estimated byminimum mean-square error (MMSE) criteria. Then, the whole channel is tracked bythe product of CPE ratio and estimated channel for the MMSE based FDE to suppressthe PN interference. Secondly, the equalized signal which is impacted by remain CPEis compensated in time domain based on decision-directed signal. Simulation resultsshow that our phase noise suppression algorithm has some advantages compared to thetraditional algorithm and the system performance impacted by serious PN can beimproved up to4dB by implementing our phase noise suppression.