Training Design for Channel Estimation in Uplink Cloud Radio Access Networks

Abstract

In this paper, a training design and channel estimation scheme is considered for uplink cloud radio access networks (C-RANs) consisting of multiple user equipments (UEs), remote radio heads (RRHs), and a centralized baseband unit (BBU) pool. Since most signal processing functions in C-RANs are moved from RRHs to the BBU pool, the individual channels over the links between UEs and RRHs and the links between RRHs and the BBU pool cannot be estimated directly. To address this issue, segment training based individual channel estimation for C-RANs is proposed in this paper, in which channel state information acquisition is performed through two consecutive segments. By using the Kalman filter, the sequential minimum mean-square-error (SMMSE) estimator is developed to efficiently estimate the individual channel states through prior knowledge of long-term channel correlation statistics and the latest radio channel state. A training structure design subject to a power constraint is obtained by minimizing the mean-square-error (MSE) of the SMMSE estimator. Since the MSE is insufficient to fully evaluate the overall performance of C-RANs, the uplink ergodic capacity is derived to exploit the impact of channel estimation on the data transmission by taking the estimation errors into consideration, and the tradeoff between the lengths of two segment training sequences is optimized by maximizing the corresponding spectral efficiency. Furthermore, the Cramér-Rao bound is used to evaluate the proposed SMMSE estimator's performance. Simulation results show that the SMMSE estimator and the corresponding training design can effectively decrease MSE and significantly increase the quality and efficiency of data transmission in C-RANs.