Full-Integrated Photonic Pre-Synaptic Spiking Neuron Based on DML-DFB Laser Chip

Kaifei Tang, Yitong Liu, Shijian Guan, Jiahui Liu, Wentao Sun, Ling Wang, Jingjing Li, Zhenxing Sun, Pan Dai, Rulei Xiao, Zhenzhen Xu, Yunshan Zhang, Xiangfei Chen · Journal of Lightwave Technology · 2024

Photonic neuromorphic computing based on spiking dynamics has been a promising approach for constructing low-latency and energy-efficient non-von Neumann architecture. However, the system is severely limited by the discrete optical injection and modulation devices. We proposed and demonstrated a true full-integrated spiking neuron based on a monolithic integrated directly modulated laser-distributed feedback laser (DML-DFB) chip. While the directly modulated master laser (DML section) inject the optical stimulus into the integrated slave laser (DFB section), rich neuron-like spiking nonlinear responses are observed, including typical spiking threshold, temporal integration, and refractory period. Huge optical discrete system is integrated to be an InP-based chip with area only 0.25 mm2. To the best of our knowledge, this is the highest level of integration for current photonic spiking system, which paves the way for realizing large-scale photonic spiking neural network (PSNN) that can handle complex tasks.

Read the paper · More papers on PaperTik