Abstract: Terahertz communications are one of the 6G technologies, offering very high data rates with an extremely large path loss and very short range, making it extremely challenging for energy-limited IoT devices. Simultaneous Wireless Information and Power Transfer (SWIPT) has emerged, but new designs employ an optimal linear energy-harvesting model that does not account for nonlinear rectifier operations. By contrast, we introduce a novel hybrid beamforming solution for wideband THz massive MIMO communications that incorporates a realistic nonlinear energy-harvesting model. We formulate an optimization problem to maximize a weighted sum of the rate and the total harvested energy, subject to transmit power constraints, constant-modulus constraints, and quality-of-service requirements. The derived nonconvex optimization problem can be efficiently addressed employing a two-stage solution. The first stage focuses on optimizing analog beamformers via shaping signals on the main Line-of-Sight components and addressing frequency-dependent beam squint effects. Digital precoders and coordinate-wise power splitting ratios will be jointly optimized with an efficient Successive Convex Approximation (SCA) method using Block Coordinate Descent (BCD). Simulation experiments demonstrate superior gains for the proposed solution compared with traditional designs that exploit optimal linear energy models, and these benefits will be more pronounced for near- and mid-field THz communications. Our work demonstrates an intrinsic trade-off in realistic THz communications and energy-limited IoT networking. These findings and challenges underscore the urgent need to model THz communications that consider RF circuit modules and to enable efficient, energy-saving 6G wireless networking for simultaneous wireless power and high-throughput communications.

Keywords: THz communication, Wireless Information and Power Transfer (SWIPT), energy harvesting, massive MIMO, Beamforming, Joint Optimization Problem, next-generation wireless communications, Block Coordinate Descent.

Pages: 456-466