[Paper Review] Intelligent Omni-Surfaces (IOSs) for the MIMO Broadcast Channel

Abdelhamed Mohamed, Nemanja Stefan Perović+1 more|arXiv (Cornell University)|Jun 27, 2022

Advanced Wireless Communication Technologies4 citations

TL;DR

This paper proposes a joint optimization framework for Intelligent Omni-Surfaces (IOSs) in MIMO broadcast channels, simultaneously optimizing base station precoding, IOS reflection/transmission coefficients, and power splitting ratios. By leveraging duality and an alternating optimization algorithm that accounts for the coupled nature of reflection and transmission, the method achieves fast convergence and significant sum-rate gains, especially under discrete phase shifts, demonstrating the superiority of IOSs over conventional RISs.

ABSTRACT

In this paper, we consider intelligent omni-surfaces (IOSs), which are capable of simultaneously reflecting and refracting electromagnetic waves. We focus our attention on the multiple-input multiple-output (MIMO) broadcast channel, and we introduce an algorithm for jointly optimizing the covariance matrix at the base station, the matrix of reflection and transmission coefficients at the IOS, and the amount of power that is reflected and refracted from the IOS. The distinguishable feature of this work lies in taking into account that the reflection and transmission coefficients of an IOS are tightly coupled. Simulation results are illustrated to show the convergence of the proposed algorithm and the benefits of using surfaces with simultaneous reflection and refraction capabilities.

Motivation & Objective

• To address the limited 360° coverage of conventional reconfigurable intelligent surfaces (RISs) that only reflect signals.
• To design a joint optimization framework for base station precoding, IOS reflection/transmission coefficients, and power ratio in MIMO broadcast channels.
• To explicitly model the coupling between reflection and transmission coefficients, based on a real IOS prototype with non-independent response elements.
• To maximize the achievable sum-rate under practical constraints, including continuous and discrete phase shifts.
• To evaluate the performance gain of IOSs over traditional RISs in terms of sum-rate and robustness to phase quantization.

Proposed method

• Formulates a sum-rate maximization problem using duality between the MIMO broadcast and multiple access channels.
• Applies block coordinate maximization and dual decomposition to optimize users’ covariance matrices.
• Derives closed-form expressions for optimal reflection and transmission phase shifts under coupling constraints.
• Uses projected gradient ascent to iteratively optimize the power ratio between reflected and refracted signals.
• Employs alternating optimization to jointly update precoding matrices, phase shifts, and power splitting ratio.
• Implements a projected gradient method with backtracking line search for power ratio update, ensuring convergence within bounds.

Experimental results

Research questions

• RQ1 How does the coupling between reflection and transmission coefficients in IOSs affect system sum-rate performance in MIMO broadcast channels?
• RQ2 What is the achievable sum-rate gain of IOSs over conventional RISs that only reflect signals?
• RQ3 How does phase quantization (discrete vs. continuous) impact the performance of IOS-aided MIMO systems?
• RQ4 Can joint optimization of precoding, beamforming, and power splitting yield faster convergence and better spectral efficiency?
• RQ5 What is the impact of the power ratio between reflected and refracted signals on overall system performance?

Key findings

• The proposed algorithm converges rapidly, achieving a local optimum in just a few iterations.
• The sum-rate performance is significantly enhanced when using IOSs with simultaneous reflection and refraction, compared to conventional RISs.
• The coupling between reflection and transmission coefficients leads to a performance loss when using discrete phase shifts, quantified by a measurable sum-rate degradation.
• The discrete-phase IOS testbed with 15×15 elements and 225 elements shows a performance gap compared to the continuous case, but still outperforms traditional RISs.
• The optimization of the power ratio via projected gradient ascent ensures convergence and stability, with the second derivative of the objective function confirming concavity.
• The system achieves a high sum-rate under realistic parameters: 8 transmit antennas, 2 receive antennas per user, 1 W transmit power, and 2 GHz carrier frequency.