Transmit-Power Efficient Linear Precoding Utilizing Known Interference for the Multiantenna Downlink

• Published in: IEEE transactions on vehicular technology Vol. 63; no. 9; pp. 4383 - 4394
• Main Authors: Razavi, S. Morteza, Ratnarajah, Tharmalingam, Masouros, Christos
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.11.2014; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Antennas; Channels; Covariance matrices; Downlink; Interference; Inversions; Linear precoding; Mathematical analysis; multiantenna downlink transmission; Performance enhancement; phase alignment (PA); power-efficient communications; Receiving antennas; Reduction; Signal to noise ratio; Symbols; Transmitting antennas; Vectors; power-efficient communications; Linear precoding; multiantenna downlink transmission; phase alignment (PA)
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2014.2317716

It has been shown that the knowledge of both channel and data information at the base station prior to downlink transmission can help increase the received signal-to-noise ratio (SNR) of each user without the need to increase the transmitted power. Achievability is based on the idea of phase alignment (PA) precoding, where instead of nulling out the destructive interference, it judiciously rotates the phases of the transmitted symbols. In this way, they add up coherently at the intended user and yield higher received SNRs. In addition, it is well known that regularized channel inversion (RCI) precoding improves the performance of channel inversion (CI) in multiantenna downlink communications. In line with this and similar to the RCI precoding, in this paper, we propose the idea of regularized PA (RPA), which is shown to improve the performance of original PA precoding. To do this, we first rectify the original PA precoding, deriving a closed-form expression to evaluate the amount of transmit-power reduction achieved for the same average output SNR compared with CI precoding. We then use this new analysis to select the appropriate regularization factor for our proposed RPA scheme. It is shown by means of theoretical analysis and simulations that the proposed RPA precoding outperforms CI, RCI, and PA precoders from both symbol error rate (SER) and throughput perspectives and provides a more power-efficient alternative. This is particularly pronounced as the number of transmit antennas becomes larger, where up to a 50-times reduction in the transmit power is achieved by RPA (PA) compared with RCI (CI) precoding for a given performance.