Making Intelligent Reflecting Surfaces More Intelligent: A Roadmap Through Reservoir Computing

This article introduces a neural network-based signal processing framework for intelligent Reflecting surface-aided (IRS) wireless communications systems. By modeling radio frequency (RF) impairments inside the meta-atoms of IRS - including nonlinearity and memory effects) we present an approach that generalizes entire IRS-aided system as reservoir computing (RC) system, efficient recurrent network (RNN) operating in state near “edge chaos.” enables us to take advantage this “fabricated” environment overcome link degradation due model mismatch. Accordingly, randomness channel RF imperfections are naturally embedded into RC framework, enabling internal dynamics lying on edge chaos. Furthermore, several practical issues, such information acquisition, passive beamforming design, physical layer reference discussed.