Reconfigurable Microwave Photonic Filter Based on Space-Division Multiplexing Powered by Artificial Neural Networks

Abstract

Space-division multiplexing (SDM) techniques bring new approaches for various applications in microwave photonic signal processing. In this work, an all-fiber reconfigurable microwave photonic filter (MPF) is realized based on SDM techniques using multicore fiber (MCF) and multimode fiber (MMF). The multimode interference (MMI) generated from the proposed MCF-MMF-MCF structure can be manipulated in dimensions of space and wavelength simultaneously, thus configuring the power distribution in all the taps of the MPF. The trained artificial neural network can accurately predict the operating wavelength and spatial input core for a target power distribution. In a low-cost, light-weight, and compact way, SDM fibers combined with MMI allow for greater flexibility and diversity in the response of the reconfigurable MPF.