Research Of Phase Noise Estimation And Compensation In Millimeter-Wave Communication Systems

With the development of communication technology,people’s demand for data rate and quality of information transmission is increasing.60 GHz communication system has become a hot spot for research with the advantages of rich continuous bandwidth resources,strong security and anti-interference.But because it is in high frequency band,compared with low frequency communication(2.4GHz and 5GHz etc.),it brings convenience to people and also brings corresponding challenges,such as the problems caused by nonlinear distortion of RF devices.This paper focuses on the phase noise of the RF devices’ nonlinear distortion in60 GHz millimeter wave communication systems,and suppresses the phase noise of the received signal by studying the phase noise estimation and compensation algorithms.Firstly,the research background and significance of the phase noise estimation and compensation algorithm in 60 GHz communication system are described,the research status of phase noise model and phase noise estimation and compensation algorithm is reviewed.Secondly,two system architectures for 60 GHz millimeter wave communication systems are introduced,the fundamentals of phase noise in nonlinear distortion are reviewed,the sources of phase noise are analyzed,and two phase noise models(a one-pole one-zero model and a Wiener model)and the effects of phase noise on the signal,constellation map,and bit error rate are investigated.Thirdly,the principles of three traditional phase noise suppression algorithms and their corresponding advantages and disadvantages are described.The averaging and interpolation algorithm based on a pilot sequence and a unique word(UW)uses the pilot sequence and the UW to estimate the average phase noise,and then interpolates the average phase noise to obtain the estimated phase noise for compensation.This algorithm is simple and practical.But the frame structure is poorly utilized and the algorithm is only suitable for slowly changing phase noise models.The self-correcting decision feedback algorithm first uses the UW to estimate the phase noise,then finds the wrong estimated phase noise using the phase noise threshold value estimated by the pilot sequence,and corrects the wrong estimated phase noise by using the average of the estimated phase noise before and after the wrong estimated phase noise.This algorithm has low frame structure utilization and is more suitable for phase noise models with short-time sudden deterioration.The phase noise estimation and compensation algorithm based on LMS/ML first estimates the phase noise using the UW,and then estimates and compensates the residual phase noise using the LMS/ML algorithm.This algorithm does not require a pilot sequence and the frame structure utilization is improved,but it is not suitable for systems with a small gap of length between channel and protection interval.Fourthly,in view of the characteristics of a one-pole one-zero model and the relative sizes of channel length and protection interval length,an improved algorithm of phase noise suppression based on multiple data blocks and decision feedback is proposed.It chooses to perform phase noise estimation after equalization to avoid the problem of fewer accurate sample points in the calculation.Then multiple data blocks are used to estimate the average phase noise to reduce the effect of additive noise.After that,it uses decision feedback algorithm based on a sliding window to further suppress the residual phase noise.In order to further improve the performance of the phase noise suppression algorithm,the estimation method of SNR is improved.Simulation results show that the phase noise suppression algorithm outperforms the traditional phase noise suppression algorithm,is applicable to two phase noise models(a one-pole one-zero model and a Wiener model),and the optimized SNR estimation method can improve the performance of the phase noise suppression algorithm.Finally,a joint IB-DFE equalization and phase noise suppression algorithm is investigated in this paper for the improved Wiener phase noise model.The algorithm considers the reliability of each data block decision,replaces MMSE equalization with IB-DFE equalization to improve the algorithm performance,and reduces the algorithm complexity by combining equalization and phase noise suppression algorithms at each iteration.The algorithm derives the estimation algorithm of phase noise at high SNR according to the characteristics of the improved Wiener phase noise model,and uses a decision feedback algorithm based on sliding window to estimate the phase noise at low SNR.To further improve the algorithm performance,the algorithm selects an improved SNR estimation method.Simulation results show that this joint algorithm outperforms the phase noise suppression algorithm based on LMS and the decision feedback algorithm based on sliding window in terms of bit error rate.