On the Optimality of STARS-Embedded Sensors for NOMA-ISAC: A Sensing Centric Design

This paper investigates a simultaneously transmitting and reflecting surface (STAR-RIS)-assisted integrated sensing and communication (ISAC) system employing non-orthogonal multiple access (NOMA), where sensors are physically embedded within the STAR-RIS. We consider that the target is located on the transmission side and represents the farthest entity from the base station, capturing a practical sensing scenario. Adopting a sensing-centric phase-shift design, we characterize the sensing performance using the average maximum achievable sensing rate (AMASR), which represents the best attainable sensing capability under given constraints. Closed-form expressions for AMASR are derived for both small and large numbers of STAR-RIS elements, with the latter obtained via a central limit theorem (CLT)-based approximation. A fractional programming framework is further developed to optimize the number of sensing elements, revealing that excessively increasing sensors beyond certain limit does not enhances the sensing performance. In addition, closed-form power allocation strategies are obtained to guarantee minimum communication rates (at both individual and sum rate level) for NOMA users while maximizing sensing efficiency. Simulation results corroborate the analysis and highlight the effectiveness of the proposed STAR-RIS-assisted NOMA-ISAC design against the STAR-RIS OMA-ISAC counterpart.