Frequency-multiplexed optical reservoir computing using a microcomb

Jonathan Cuevas ¹, Yue Hu ², Baoqi Shi ²

⁰Graduate School of Sciences and Technology for Innovation, Tokushima, Japan.

Nanophotonics (berlin, Germany) +

|

September 19, 2025

View abstract on PubMed

Summary

This study introduces a novel frequency-multiplexed optical reservoir computing (ORC) system using microcomb modes. This approach achieves high-speed temporal inference, overcoming limitations of previous ORC designs for faster, integrated photonic processors.

Area Of Science

Photonics
Optical Computing
Nonlinear Dynamics

Background

Optical reservoir computing (ORC) offers potential for fast, energy-efficient temporal inference.
Existing ORC systems face limitations in clock rate and monolithic integration due to reliance on fiber delay lines or spatial multiplexing.

Purpose Of The Study

To develop a frequency-multiplexed ORC architecture for enhanced speed and scalability.
To leverage the dynamics of a microcomb in a silicon nitride microresonator for high-dimensional nonlinear mapping and memory.

Main Methods

Utilized a dissipative Kerr-soliton microcomb generated in a high-Q Si3N4 microresonator as the computing nodes.
Encoded input signals via rapid pump laser detuning modulation.
Implemented optical output weighting using microring resonator arrays.

Main Results

Numerical modeling demonstrated a normalized mean-square error (NMSE) of 0.015 on the Santa Fe chaotic time-series task at 50 MSa/s.
Achieved over a tenfold reduction in symbol-error rate for nonlinear equalization (NLEQ) at 100 MSa/s.
Experimental validation using 37 microcomb modes confirmed performance on benchmark tasks.

Conclusions

The frequency-multiplexed ORC architecture effectively addresses scalability and speed challenges in nanophotonic computing.
CMOS-compatible fabrication offers a pathway to compact, energy-efficient photonic processors operating beyond 1 GSa/s.

Keywords:

microresonator optical frequency comb reservoir computing