Optimal Beam Partitioning and Scheduling for SWIPT Networks

Abstract

Simultaneous Wireless Information and Power Transfer (SWIPT) enables IoT devices to harvest energy and exchange data over the same wireless infrastructure. In directional SWIPT networks, beam design induces a trade-off between concentrating power on a few users versus broadening energy the harvesting opportunities to a wider set of devices, especially under harvest-then-transmit operation. This paper proposes a framework to optimize user-beam association and scheduling of transmission and energy harvesting. As a concrete instantiation, we consider a SWIPT network with directional beams and formulate the corresponding proportional-fair beam design problem under a constraint on the number of active beams. We develop an exact dynamic-programming algorithm that computes the globally optimal non-overlapping beam partition, and we derive an instance-wise upper bound on the loss induced by the non-overlapping restriction under beam-independent scheduling. Numerical results show that the proposed solution closely tracks this benchmark over a broad range of SWIPT regimes, while highlighting the role of heterogeneity in user distance, density, spatial concentration, and scheduling choices. These findings indicate that non-overlapping beam designs can provide near-optimal performance with a tractable optimization structure.