[Paper Review] Weighted Sum-Rate Optimization for Intelligent Reflecting Surface Enhanced Wireless Networks

Huayan Guo, Ying‐Chang Liang+2 more|arXiv (Cornell University)|May 20, 2019

Advanced Wireless Communication Technologies43 references149 citations

TL;DR

The paper studies joint active BS beamforming and passive IRS beamforming for IRS-aided multiuser MISO downlink to maximize weighted sum-rate, incorporating practical discrete/continuous phase-shift models and proposing low-complexity closed-form algorithms.

ABSTRACT

Intelligent reflecting surface (IRS) is a promising solution to build a programmable wireless environment for future communication systems. In practice, an IRS consists of massive low-cost elements, which can steer the incident signal in fully customizable ways by passive beamforming. In this paper, we consider an IRS-aided multiuser multiple-input single-output (MISO) downlink communication system. In particular, the weighted sum-rate of all users is maximized by joint optimizing the active beamforming at the base-station (BS) and the passive beamforming at the IRS. In addition, we consider a practical IRS assumption, in which the passive elements can only shift the incident signal to discrete phase levels. This non-convex problem is firstly decoupled via Lagrangian dual transform, and then the active and passive beamforming can be optimized alternatingly. The active beamforming at BS is optimized based on the fractional programming method. Then, three efficient algorithms with closed-form expressions are proposed for the passive beamforming at IRS. Simulation results have verified the effectiveness of the proposed algorithms as compared to different benchmark schemes.

Motivation & Objective

• Motivate and address the challenge of maximizing weighted sum-rate in IRS-aided multiuser MISO downlink systems.
• Formulate a non-convex joint optimization problem for BS transmit and IRS reflection coefficients under practical RC constraints.
• Develop low-complexity, closed-form algorithms to solve the active and passive beamforming subproblems.
• Provide an iterative framework with convergence guarantees and assess performance via simulations.

Proposed method

• Decouple the joint problem using Lagrangian dual transform to handle the log objective.
• Apply fractional programming and quadratic transforms to optimize active beamforming at the BS.
• Develop three low-complexity algorithms for passive beamforming under ideal, continuous, and discrete RC assumptions.
• Reformulate the passive-beamforming subproblem as a convex QCQP under the ideal RC, and extend to non-convex RC with projection-based approaches.
• Propose an alternating optimization framework that updates auxiliary variables, active beamformers, and IRS phase shifts iteratively until convergence.

Experimental results

Research questions

• RQ1 How can the weighted sum-rate of an IRS-aided multiuser MISO downlink be maximized through joint BS and IRS beamforming?
• RQ2 What low-complexity algorithms can efficiently compute passive beamforming under ideal, continuous, and discrete phase-shift constraints?
• RQ3 Can an alternating optimization framework with closed-form updates converge to a high-performance solution for WSR in this setting?
• RQ4 What is the performance gap between ideal RC assumptions and practical RC implementations?

Key findings

• An iterative algorithm with closed-form updates effectively alternates between optimizing active beamforming and passive RCs to maximize weighted sum-rate.
• Three low-complexity passive beamforming algorithms are proposed and applicable to ideal RC, continuous phase shifts, and discrete phase shifts.
• The continuous phase shifter setup can approach the performance of the ideal RC case, with modest degradation for low-resolution (e.g., 2-bit) phase shifters.
• Simulation results demonstrate significant capacity gains over benchmark schemes across RC models.
• A unified passive-beamforming design under ideal RC provides insight into performance limits and serves as a good initialization for non-convex RC cases.