Harnessing Random Media for Advanced Optical Signal Processing and Computing

Recent studies have highlighted the potential of utilizing light transport in disordered media for optical signal processing and optical computing. One approach involved employing random projections through scattering samples in imaging applications. Integrated photonic random projectors can play a crucial role as a reprocessing component in photonic integrated circuits for analog signal processing. In this study, we present and analyze basic multiport configurations capable of executing random projections within a photonic chip. We explore the behavior of light propagation within an on-chip randomly pixelated dielectric waveguide slab, equipped with multi-input and multi-output ports, serving as a photonic random projector. They enable essential functionalities like dimensionality reduction and compressive sensing, making them valuable tools in the realm of optical information processing. We use Spins-B to generate the device, run finite-difference frequency-domain (FDFD) Maxwell simulations and study statistical properties of the device transmission matrix. COMSOL is also utilized to verify our conclusions. Notably, our research has extended this concept by demonstrating the feasibility of random projections using optical waveguide arrays with transverse disorders.