Learning-Based Resource Management for SWIPT

In this article, we consider the joint optimization of transmit power and power splitting ratio to maximize the energy efficiency in a simultaneous wireless information and power transfer based interference channel, in which receivers use a power splitting policy to harvest energy from a wireless signal. We propose an optimization-based iterative algorithm (O-IA) from well-known optimization techniques as a comparative scheme, and also devise a neural network based learning algorithm (NN-LA) to deal with nonconvexity caused by cochannel interference among multiple nodes. Through simulations, we provide a comparative study of the two approaches in terms of energy efficiency and time complexity. In particular, we find that NN-LA achieves a near-optimal energy efficiency, whereas its time complexity is significantly reduced, in comparison with O-IA.