Intelligent Reflecting Surface Enabled Random Rotations Scheme for the MISO Broadcast Channel

TL;DR

This paper proposes a random rotations scheme for IRS-assisted MISO broadcast channels that requires no CSI at the IRS, leveraging multi-user diversity to improve sum-rate and energy efficiency, approaching coherent beamforming performance as user numbers grow.

Contribution

It introduces a novel IRS scheme with random phase rotations that reduces CSI requirements and derives sum-rate scaling laws for large user regimes.

Findings

Random rotations increase sum-rate via multi-user diversity.

Performance approaches coherent beamforming with many users.

Energy efficiency can be optimized using derived scaling laws.

Abstract

The current literature on intelligent reflecting surface (IRS) focuses on optimizing the IRS phase shifts to yield coherent beamforming gains, under the assumption of perfect channel state information (CSI) of individual IRS-assisted links, which is highly impractical. This work, instead, considers the random rotations scheme at the IRS in which the reflecting elements only employ random phase rotations without requiring any CSI. The only CSI then needed is at the base station (BS) of the overall channel to implement the beamforming transmission scheme. Under this framework, we derive the sum-rate scaling laws in the large number of users regime for the IRS-assisted multiple-input single-output (MISO) broadcast channel, with optimal dirty paper coding (DPC) scheme and the lower-complexity random beamforming (RBF) and deterministic beamforming (DBF) schemes at the BS. The random rotations scheme increases the sum-rate by exploiting multi-user diversity, but also compromises the gain to some extent due to correlation. Finally, energy efficiency maximization problems in terms of the number of BS antennas, IRS elements and transmit power are solved using the derived scaling laws. Simulation results show the proposed scheme to improve the sum-rate, with performance becoming close to that under coherent beamforming for a large number of users.