Cooperative Channel Estimation in Distributed Antenna Systems

Abstract

Mobile data traffic has exploded around the world due to increased usage on mobile devices. Consequently, the future communication systems are required to increase the capacity so that it can support such growth in the mobile data traffic volume. Massive multi-input multi-output (MIMO) is under consideration as one of suitable communication technology to increase capacity and conserve energy with an unlimited number of antennas. For ease of implementation, massive MIMO schemes are often employed in a distributed antenna system (DAS) configuration, where a digital unit (DU) controls a number of radio units (RUs) in a centralized manner. Channel sounding reference signal (SRS) is transmitted to estimate channel state information (CSI) in the uplink of a MIMO-based time division duplex wireless system. The transmission of SRS is indifferent from the number of transmit antennas, making it quite feasible in MIMO transmission environments. It may be desirable to make users use SRS signals orthogonal to each other. However, conventional systems like LTE-Advanced can provide a limited number of orthogonal SRS signal, suffering from so-called pilot contamination when the number of active users is large. In this thesis, we consider the generation of SRS by imposing phase rotation, yielding an increased number of SRSs which are slightly correlated with each other. Thus, we can simultaneously allocate more users SRS for the channel sounding. We consider channel estimation with the use of correlated SRS in a cooperative manner. Since the DU can have information of all RUs, it can exploit interference from other RUs as additional information. We design a cooperative channel estimation scheme that exploits SRSs from other RUs, which are treated as interference signal to be removed in conventional systems. Finally, the performance of the proposed scheme is verified by computer simulation, showing that the proposed scheme can make the system simultaneously serve a significantly increased number of users without performance degradation.