Neural Network-Based Optimization for SLIPT-Enabled Indoor VLC Systems With Energy Constraints

In this paper, we consider a simultaneous lightwave and power transfer-enabled indoor visible light communication system and aim to investigate how to manage the transmission power from multiple transmitters to provide both information and energy harvesting. We formulate three different optimization problems, all aiming to minimize the total average transmit power at the luminaries, assuming different performance constraints, such as data rate, energy harvest, and illumination requirements. The first problem aims to find the optimal beamforming matrix and the transmit powers at light emitting diodes (LEDs), while the second problem aims to use zero-forcing beamforming and finds the optimal transmit powers. Finally, the third problem aims to find the minimum number of LEDs required to satisfy the given constraints. Relying on a Machine Learning approach, our solution is capable of predicting the user mobility patterns, and receiver orientation angles and accordingly optimizing parameters leading to a near-optimal result under different blockage conditions with low computational complexity. Moreover, a comparison with other approaches shows the effectiveness of the proposed solution in terms of significantly reducing the transmit power in a wide range of orientation errors. Specifically, up to 50% of the average transmit power can be minimized using the presented approach.