High-speed photonic reservoir computer based on a delayed Fano laser under electrical modulation

Abstract

We propose and numerically demonstrate a high-speed photonic reservoir computing (RC) system using a compact Fano laser (FL) with optical feedback under electrical modulation. Benefiting from its insensitivity to external feedback, an FL has a wider dynamic steady-state region compared with a conventional Fabry–Perot laser, which significantly extends the ranges of desirable RC implementation. Interestingly, we observe two separate regions of good RC performances corresponding to two scenarios of the dynamic steady state of the FL, respectively. Moreover, the robust RC performance versus the feedback phase can be achieved in one of the steady-state regions, where the laser is not destabilized for lower external reflectivity. Owing to the ultra-short photon lifetime in the FL, the information processing rate of our proposed RC system may reach 10 Gbps. More importantly, as a specific type of microscopic laser, the FL offers potential applications to RC-based integrated neuromorphic photonic systems.

Enhanced Prediction Performance of Reservoir Computing Based on Mutually Delay-Coupled Semiconductor Lasers via Parameter Mismatch

“…L is the total number of the datasets, and n is the time index of the input data. Usually, the performance of the RC system is acceptable when NMSE < 0.1 [8,13,21,22,31,32,38,39].…”

“…This simplified structure accelerates the hardware process of the RC [16]. Various delay-based RC structures, such as all-optical-feedback-based RC [7,[13][14][15][16][17][18][19][20][21][22] and optoelectronic-feedback-based RC [23][24][25][26], have been extensively studied in the past decade. The all-optical-feedback-based RC carries information by photons, which may allow for low power consumption and high-speed computing [27,28].…”

High-Speed Reservoir Computing Based on Circular-Side Hexagonal Resonator Microlaser with Optical Feedback

“…Brunner et al realized the RC experimentally using semiconductor laser diodes as nonlinear nodes, reaching a 1 Gbps R inf with 10% error in chaotic time series prediction tasks [12]. Guo et al and Yu et al used nanolaser and Fano lasers with an ultra-short photon lifetime to increase the R inf to 10 Gbps, respectively [17,18]. In addition, Estébanez et al proved the positive effect of bandwidth enhancement on R inf in the RC system [19].…”