Efficient Channel State Information Acquisition in Massive MIMO Systems using Non-Orthogonal Pilots

Paul Ferrand1, Alexis Decurninge2, Maxime Guillaud3, Luis Ordóñez4

  • 1Huawei Technologies Co. Ltd., France
  • 2Huawei Technologies
  • 3Huawei Technologies, Mathematical and Algorithmic Sciences Lab
  • 4Huawei Technologies Co. Ltd.

Details

11:50 - 12:10 | Thu 16 Mar | Main Room | S3.5

Session: Signal Processing for Massive MIMO

Abstract

The objective of this article is to review and benchmark strategies to acquire channel state information (CSI) in time-division duplex (TDD) Massive MIMO systems. In particular, we consider the use of statistical CSI at the base stations (BSs), together with non-orthogonal pilot sequences.
Such techniques can theoretically reduce the amount of spectral resources dedicated to channel sounding, thanks to the increased pilot reuse that they allow. We review their application, and present a holistic evaluation of their practical implementation and limitations, including factors that are often neglected in theoretical works, including covariance estimation and tracking, non-orthogonal pilot sequences, signaling associated with dynamically allocated pilot sequences, realistic channel model (spatial covariance, stationarity), and imperfect statistical CSI. We present a possible design addressing these various problems with reasonable complexity, and benchmark the achieved performance. Our results show that when the cost of CSI acquisition is properly accounted for, the tracking of statistical CSI together with the use of non-orthogonal pilots enables to reach a higher effective spectral efficiency than what can be achieved with orthogonal pilot sequences.