Joint User, Channel, Modulation-Coding Selection, and RIS Configuration for Jamming Resistance in Multiuser OFDMA Systems

Reconfigurable intelligent surfaces (RISs) can potentially combat jamming. It is non-trivial to perform holistic selections of users, data streams, and modulation-coding modes for all subchannels, RIS configuration in a downlink multiuser OFDMA system under jamming attacks, because mixed-integer program nature difficulties acquiring the channel state information (CSI) channels from an uncooperative jammer. We propose new deep reinforcement learning (DRL)-based approach that learns through changes rates users reject maximize sum rate. The key idea decouple continuous discrete modes. Another critical aspect we show optimal almost surely follow winner-takes-all strategy. Accordingly, DRL framework with twin-delayed deterministic policy gradient takes strategy evaluate reward, thereby reducing action space accelerating learning. Simulations converges fast fulfills benefit RIS. With no need CSI jammer, offers practical value.