Sensing-Then-Serve: A Novel Framework From ISAC Toward Sensing-Enhanced SWIPT

We study a two-phase, sensing-enhanced simultaneous wireless information and power transfer (SWIPT) framework for a multi-user system under position uncertainty. Phase-I dedicates a fraction of the frame to cooperative sensing that refines user positions; Phase-II performs SWIPT using the refined statistics. We define the information-energy (I-E) region to capture the fundamental tradeoff between sensing-induced enhancement and SWIPT duration, and then optimize it via bisection-based phase scheduling and robust beamforming. The sensing-induced enhancement is quantified by the derived closed-form Cramér-Rao Lower Bound (CRLB), and then optimized via a direction alignment water-filling procedure. For the SWIPT phase, we derive the ergodic capacity upper bound (ECUB) and the ergodic power havest (EPH) to characterize, respectively, the communication and power transfer efficiencies that harness the sensing enhancement. Accordingly, we devise a closed-form eigenbeamforming for power transfer; a quadratically constrained quadratic program (QCQP)-based beamforming for communication; and a QCQP-based beamforming with successive convex approximation (SCA) for SWIPT. Numerical results demonstrate substantial robustness and efficiency gains over conventional beamforming across a wide range of uncertainty levels.