Enabling ISAC with Simultaneous Transmission and Reflection-Stacked Intelligent Metasurfaces

Multi-layer stacked intelligent metasurfaces (SIM) have shown considerable potential for wave-domain beamforming in integrated sensing and communication (ISAC) systems. However, conventional SIM is confined to half-space coverage imposed by transmission-only components. While simultaneous transmission and reflection reconfigurable intelligent surfaces (STAR-RIS) overcome this via full-space beamforming, the single-layer architecture inherently limits the sensing and communication capabilities. Therefore, we propose a novel STAR-SIM architecture that embeds a SIM with a STAR surface for full-space beamforming, and formulate the joint optimization problem to maximize the sum rate subject to the Cramér-Rao bound (CRB) constraint. To address the challenges of intractable sensing constraints and unique asymmetric multi-layer structure of STAR-SIM, we establish an intuitive link between the CRB and beampattern gain and facilitate an alternating optimization framework, where the STAR-SIM phase shifts are optimized by penalty-based gradient ascent (GA) and projected GA (PGA) algorithms with high efficiency and low complexity compared with conventional semidefinite relaxation (SDR) methods. Simulation results validate the effectiveness of the proposed STAR-SIM structure with full-space beamforming capability and showcase its transformative potential for next-generation ISAC networks.