Channel Estimation in Massive MIMO Systems

We introduce novel blind and semi-blind channel estimation methods for cellular time-division duplexing systems with a large number of antennas at each base station. The methods are based on the maximum a-posteriori principle given a prior for the distribution of the channel vectors and the received signals from the uplink training and data phases. Contrary to the state-of-the-art massive MIMO channel estimators which either perform linear estimation based on the pilot symbols or rely on a blind principle, the proposed semi-blind method efficiently supp resses most of the interference caused by pilot-contamination. The simulative analysis illustrate s that the semi-blind estimator outperforms stateof-the-art linear and non-linear approaches to the massive MIMO channel estimation problem. I. I NTRODUCTION Deployment of large numbers of antennas at the base stations in a cellular network can lead to huge gains in both, spectral and energy efficiency. The increase i n energy efficiency is mostly due to the array gain and the high spectral efficiency is achieved by serving s everal users simultaneously through spatial multiplexing. Under favorable propagation conditions [2], namely, i.i.d. channel coefficients for the links from the different antennas at a base station to a user terminal, the s patial channels to distinct users are close to orthogonal in such a massive MIMO system due to the law of large numbers. Consequently, the interference caused by spatial multiplexing can be suppressed with simple signal processing methods, e.g., matched filters [2]‐[4] or even constant envelope prec oding [5].