Cross-gain modulation-based photonic reservoir computing using low-power-consumption membrane SOA on Si.

Area of Science:

• Photonics
• Nonlinear Optics
• Integrated Photonics

Background:

• Reservoir computing (RC) requires efficient nonlinear processing units.
• Semiconductor optical amplifiers (SOAs) offer nonlinear optical effects.
• Integrating photonic devices on silicon platforms is crucial for scalability.

Purpose of the Study:

• To demonstrate photonic reservoir computing using cross-gain modulation (XGM) in a membrane SOA.
• To achieve low-power nonlinear operation for RC applications.
• To evaluate the performance of the XGM-based reservoir.

Main Methods:

• Utilizing cross-gain modulation (XGM) in a membrane semiconductor optical amplifier (SOA) on a silicon (Si) platform.
• Configuring the XGM-based reservoir with delayed feedback in an anti-parallel injection.
• Evaluating reservoir computing performance using information processing capacity (IPC) and a nonlinear benchmark task.

Main Results:

• Achieved low-power operation with 10^1-mW-scale power consumption and 10^2-µW-scale optical input power.
• Demonstrated robust feedback circuit operation due to phase insensitivity and elimination of loop oscillation risk.
• Showcased strong nonlinear transformation of input time-series signals by the XGM-based reservoir.

Conclusions:

• The membrane SOA enables RC-applicable nonlinear operations through XGM at a low power scale.
• The proposed photonic reservoir computing system offers an efficient and robust solution for signal processing.
• This work paves the way for compact and low-power integrated photonic reservoir computing systems.