Movable-RIS Assisted Symbiotic Radio Systems

In symbiotic radio (SR) systems, the direct link is typically dedicated to transmitting primary signals, and its absence may result in the failure of joint decoding of primary and secondary signals. To address this challenge, this paper proposes a movable-reconfigurable intelligent surface (RIS) based flexible RIS partitioning (FRP) strategy, which exploits the mobility of RIS elements to adaptively partition them into two discrete sub-surfaces for transmitting primary signals and supporting secondary transmission, respectively. Based on this design, we formulate a mixed- integer nonlinear programming (MINLP) problem to maximize the primary transmission performance under the secondary transmission constraint. To efficiently solve this problem, we develop a hybrid optimization-learning framework, where the inner-layer employs an alternating optimization algorithm to jointly optimize base station beamforming and RIS phase shifts, while the outer-layer leverages a deep reinforcement learning method based on unit action simulation proximal policy optimization (UAS-PPO) to update the RIS elements' positions and partitioning strategy. Numerical results demonstrate that the proposed FRP scheme achieves significant primary transmission performance gains over benchmark methods.