Joint Energy Beamforming and Transceiver Optimization for Wireless Powered Over-the-Air Computation Systems

Abstract

This paper considers a basic wireless powered over-the-air computation (AirComp) system, which includes one separately located energy transmitter (ET) and wireless data access point (AP), and multiple low-power sensors equipped with energy harvesting (EH) circuits. The multi-antenna ET is deployed to wirelessly charge multiple low-power sensors simultaneously via energy beamforming, and the AP aims to accomplish the EH based sensors' data functional computation. Under a harvest-then-sense-and-transmit protocol, we investigate a joint wireless powered AirComp design to minimize the ET's transmit energy, subject to the sensors' EH constraints and the AirComp distortion constraints. In particular, we jointly optimize the ET's energy beamformers, the AP's aggregation beamformers, and the sensors' transmit coefficients. To address the non-convexity, we decompose the joint design problem into a series of convex subproblems of alternately optimizing the ET's energy beamforming design and the AirComp transceiver design. Furthermore, we proposed an alternating-optimization-based algorithm to obtain a low-complexity solution. Numerical results show the fast convergence performance and the effectiveness of the proposed design solution over the existing schemes.