Cooperative-Memory Photonic Reservoir using Modulation Nonlinearity: Circumventing the Speed Constraints of Nonlinear Silicon Microring Resonators
Preprint
Abstract
Complex dynamics of silicon microring resonators loaded by delayed feedback elements enable high-speed photonic reservoir computing. Implementing feedback is especially challenging when the required delay should match the time scales of silicon's nonlinearities. To increase the computation speed and preclude any need for very long delay lines, we avoid relying on silicon's nonlinearity and merely employ either amplitude or phase modulation along with direct detection. By supplementing its memory with that of the electronic output layer, the proposed photonic reservoir composed of a Mach-Zehnder interferometer and a microring resonator is predicted to perform computations one order of magnitude faster than those based on silicon's nonlinearity with its speed only limited by the modulation/detection bandwidth. This reservoir performs accurately in NARMA-10, Mackey-Glass, and Santa-Fe prediction tasks, and enables signal equalization in optical communication systems.Keywords
All Related Versions
This publication has 31 references indexed in Scilit:
- Demonstration of high-frequency self-pulsing oscillations in an active silicon micro-ring cavityScientific Reports, 2024
- Next generation reservoir computingNature Communications, 2021
- Reservoir computing based on a silicon microring and time multiplexing for binary and analog operationsScientific Reports, 2021
- Thermo-Optic Multistability and Relaxation in Silicon Microring Resonators with Lateral DiodesPhysical Review Applied, 2020
- PAM-4 Transmission at 1550 nm Using Photonic Reservoir Computing Post-ProcessingIEEE Access, 2019
- Multibistability and self-pulsation in nonlinear high-silicon microring resonators considering thermo-optical effectPhysical Review A, 2013
- Information processing using a single dynamical node as complex systemNature Communications, 2011
- New results on recurrent network training: unifying the algorithms and accelerating convergenceIEEE Transactions on Neural Networks, 2000
- Experimental realization of any discrete unitary operatorPhysical Review Letters, 1994
- High-dimensional chaotic behavior in systems with time-delayed feedbackPhysica D: Nonlinear Phenomena, 1987