Model-Free Configuration of Intelligent Reflecting Surfaces: Toward Pervasive Adaptability and Enhanced Robustness

Abstract

Intelligent reflecting surface (IRS), with the capability of configuring radio environment, is an enabling technology for 6G communications. However, owing to its low-cost and passive nature, the IRS suffers from imprecise reflection control and channel uncertainty, rendering the accurate modelling of the exact relationship between the system performance and IRS's configurable reflection coefficients a formidable challenge. To address this challenge, we propose model-free designs regardless of the exact relationship to configure IRS. In this article, we first review the framework of model-based IRS configuration and analyze the key factors that result in its performance loss in practice. Then, we introduce the concept of model-free IRS configuration and explain its robustness in engineering applications. Moreover, we provide an overview of the current progress in model-free IRS configuration and make a comprehensive comparison among the current model-free IRS configuration technologies. Finally, we discuss the challenges and opportunities in the model-free IRS configuration.