Multiantenna beamforming has potential to improve the efficiency of simultaneous wireless information and power transfer (SWIPT). Existing designs are focused on the downlink of multiple-input-single-output under the assumption of single-antenna users and negligible energy consumption in users' circuitry, despite the fact that using multiple antennas on the user side can further improve system efficiency. In this paper, novel multiuser collaborative multiple-input multiple-output SWIPT systems are studied under the assumption of nonnegligible circuit energy consumption. Particularly, we convexify and maximize the uplink sum rate of active users, while maintaining the quality of service of their downlink data. The beamformers and durations of both links, and the power splitting factors of individual users are jointly optimized, using semidefinite programming and golden search. Further, the selection of active users is optimized, where all users are assumed to be active in the beginning and those detrimental to the sum-rate maximization are continually deactivated. Evident from simulations, the proposed approaches can eliminate the need for computationally prohibitive combinatorial integer programming at a marginal cost of the sum rate.
- Simultaneous wireless information and power transfer (SWIPT)
- power splitting
- semidefinite programming