Channel Estimation,Performance Analysis And Optimal Design For One-bit Massive MIMO Systems

In order to largely improve the capacity of wireless communication system and meet the requirement of the huge amount of future wireless communication data services,the massive MIMO technique has attracted considerable research interest both in academia and industry.As one of the key techniques of the fifth generation wireless communica-tion,massive MIMO is able to significantly improve the spectral efficiency and average out the effect of the intra-cell user interference and the thermal noise by equipping hun-dreds of antennas at the base station(BS).However,when deploying the practical massive MIMO systems,the economic cost and hardware energy consumption will be prohibitive if each BS antenna is equipped with a quantizer with high-resolution and high sample rate.Therefore,in order to reduce the economic cost and hardware energy consumption caused by the power hunger quantizer,recent research interest has shifted by combing the low-resolution quantization,especially the one-bit quantization technique with massive MIMO technique.This thesis investigates the channel estimation,performance analysis and perfor-mance optimization for the one-bit massive MIMO systems,and aims at providing theo-retical basis for future research.Major innovative results are summarized as follows.1.The best performance that can be achieved by unbiased channel estimator is first examined and the Cramer-Rao lower bound of the channel estimate for one-bit MIMO narrowband systems and one-bit MIMO-OFDM systems are derived,respectively.Owing to the one-bit ADCs,it is found that any unbiased channel estimator will fail to estimate the channel in the high SNR region.By employing the Bussgang decomposition,two linear channel estimators that are applicable for both narrowband MIMO systems and MIMO-OFDM systems are proposed.Theoretical analysis shows that the accuracy of the channel estimate depends on which orthogonal pilot sequences are used.Moreover,in contrast to the conventional systems,it is shown that the perfect channel state informa-tion cannot be acquired by unlimitedly increasing the SNR in one-bit systems.Numerical results show that the spatial correlation of the quantizer noise has an impact on perfor-mance of channel estimate and should be taken into account to improve the performance of channel estimate.2.The uplink-downlink SINR duality for one-bit massive MIMO narrowband sys-tems and one-bit massive MIMO-OFDM systems is proved by using Bussgang decompo-sition,respectively.A new precoding scheme and downlink power allocation strategy that results in uplink-downlink SINR duality for one-bit systems are also proposed.Impor-tantly,it is shown that the OFDM technique can still convert the frequency-selective fad-ing channels into multiple parallel flat fading channels in one-bit systems.By employing the Bussgang decomposition,the closed-form expressions for the uplink achievable rate for the maximal ratio combing receiver and zero-forcing receiver with perfect/imperfect channel state information known at the BS are derived,respectively.Then the power efficiency of one-bit massive MIMO systems are investigated based on the closed-form expressions.It is shown that the transmit power for one-bit massive MIMO systems can still be scaled down with the increase of the BS antennas while maintaining the desirable system spectral efficiency.Moreover,it is shown that,if the system only has imperfect channel state information,then the one-bit massive MIMO can achieve the same spec-tral efficiency as the conventional massive MIMO systems by deploying 2.5 times more antennas at the BS.3.Based on the closed-form expressions for the achievable rate,the optimal resource allocation scheme that maximizes the sum spectral efficiency is investigated.It is proved that the optimal training length in one-bit massive MIMO systems is no longer equal to the number of users and depends on various parameters such as the coherence interval and the average transmit power.Also,unlike conventional systems,it is observed that in terms of sum spectral efficiency,there is relatively little benefit to separately optimizing the training and data power.By using the multiple objectives optimization theorem,the Pareto boundary,which shows the tradeoff between the sum spectral efficiency and ener-gy efficiency for one-bit massive MIMO systems is obtained.Furthermore,the optimal values of spectral efficiency and energy efficiency by using the weighted sum method and weighted product method are obtained,respectively.Numerical results show that if the hardware energy consumption is not taken into account,then the losses in both spectral efficiency and energy efficiency for one-bit massive MIMO systems can be compensated for by deploying 2-2.5 times more antennas at the BS.