Integrated sensing and communication (ISAC) systems increasingly rely on movable antennas (MAs) and reconfigurable intelligent surfaces (RISs) to shape radio environments with fine spatial control. These components can steer signals for both communication and target illumination, but their flexibility comes at a cost: the optimal MA positions and RIS phases change with each channel realization.
A new arXiv paper tackles this by introducing a GPU-accelerated robust pre-optimization approach. Instead of recomputing settings from scratch for every channel, the method pre-optimizes configurations in a way that is robust across realizations, and uses GPU parallelism to speed up the computation.
The work highlights a practical bottleneck in deploying MA-RIS systems for ISAC: the gap between theoretical spatial control and real-time adaptability. By offloading the optimization to GPUs and building robustness into the pre-optimization stage, the authors aim to make these systems more feasible in dynamic environments. The abstract does not provide quantitative results, so the reported gains remain unspecified. The source is a single arXiv preprint, and no independent verification is available.