Iterative channel estimation methods in large antenna systems

Abstract

Large antenna system, also known as massive multiple-input multiple-output (MIMO) system, has been identified as a promising technique for future 5G cellular systems. Through the use of a large number of antennas, huge gain can be achieved through coherent reception (matched filtering) or coherent transmission (beamforming). To achieve such gain, channel state information (CSI) is the key. This fact motivates the research activities on the channel estimation problem in large antenna systems. Iterative (turbo) detection/estimation principles have been successfully applied to many coding, communication and signal processing problems. In this thesis, we will study iterative channel estimation methods in the large antenna context. After introducing the background work in Chapter 1, we consider in Chapter 2 the channel estimation problem for an uplink multi-cell system where each cell contains only one user. We analyze the performance of a data-aided channel estimation technique where partially decoded data is used to estimate the channel. We show that there are two types of interference components in this scheme that do not vanish even when the number of antennas grows to infinity. The first type, referred to as cross-contamination, is due to the correlation among the data signals from different users. The second type, referred to as self-contamination, is due to the dependency between the channel estimate and the estimation error. For efficient use of the channel, the data part in a signaling frame is typically much longer than the pilot part for a practical system. Consequently, compared with pilot signals, data signals naturally have lower cross correlation. This fact reduces the cross-contamination effect in the data-aided scheme. Furthermore, self-contamination can be effectively suppressed by iterative processing. These results are confirmed by both analyses and simulations. Compared to other channel estimation schemes for large antenna systems, the data-aided channel estimation scheme is non-cooperative and does not rely on specific channel conditions. Therefore it is simpler and more flexible in practice. In Chapter 3, we extend the analysis of the data-aided channel estimation scheme to the multiple-user scenario. We consider joint linear minimum meansquared error (LMMSE) channel estimation and matched filtering (MF)-based data detection with soft-interference cancellation. In Chapter 4, we study a scheme where pilots are superimposed on data. This scheme avoids the rate loss incurred by dedicate pilot positions. We investigate the power allocation problem between pilot and data for the superimposed scheme. Intuitively, there is an optimal pilot power ratio that achieves the best tradeoff between pilot power overhead and channel estimation quality. Interestingly, through the use of a recently developed relationship between minimum mean-squared error (MMSE) and mutual information (referred to as the MMSE-I relationship), and based on several assumptions, we show that the optimal portion of the power allocated to pilot approaches zero for Gaussian signaling. We also consider the practical realization of the theoretical prediction and show that considerable performance improvement can be achieved by matching the transfer functions of the forward error correction (FEC) code and the channel estimator/data detector. In Chapter 5, we propose a compressed sensing channel estimation algorithm for millimeter-wave large antenna systems. Millimeter-wave communication channel is sparse due to the directional transmission nature at high frequencies. Compressed sensing techniques can be leveraged to exploit the spatial sparsity. We first propose a generic turbo signal recovery (TSR) algorithm for partial unitary sensing matrix-based compressed sensing problems. We analyze the proposed TSR algorithm using state evolution. We show that the stationary point of the state evolution of the proposed algorithm is consistent with that of the Bayesian optimal MMSE performance derived using the replica method. This indicates the potentially excellent performance of the proposed algorithm, as verified by numerical results. For a millimeter-wave system with a large uniform linear array, the array manifold can be approximated by a discrete Fourier transform (DFT) matrix. Motivated by this fact, we apply the proposed turbo signal recovery algorithm to the millimeter-wave sparse channel estimation problem. We follow the hybrid analog-digital processing architecture where the entries of the analog processing matrix are constrained to have constant magnitudes. A technique to design the analog processing matrix is proposed. We provide numerical results to demonstrate the effectiveness of the proposed algorithm. Finally, Chapter 6 concludes the thesis and outlines possible future research directions. In summary, in this thesis we have studied the applications of iterative channel estimation methods in large antenna systems. Our results show that by appropriately exploiting available information, such as soft feedback of a decoder and a priori sparsity information, we can reduce the training overhead and improve the channel estimation accuracy. The contributions contained in this thesis show that iterative channel estimation methods provide a powerful means to enhance the performance of massive MIMO systems

Date of Award	2 Oct 2015
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Ping LI (Supervisor)