Photonic Spiking Reservoir Computing System Based on a DFB-SA Laser for Pattern Recognition

In the field of information processing, traditional computing methods encounter limitations in handling increasingly complex tasks and meeting growing performance requirements. Reservoir computing, as a new computing paradigm, has demonstrated excellent performance in time series prediction tasks. However, traditional photonic reservoir computing still needs improvement in certain aspects, such as high computational complexity and relatively high-power consumption for information processing. In our work, a photonic spiking reservoir computing system based on a single distributed feedback laser with saturable absorber (DFB-SA laser) is demonstrated numerically and experimentally. Owing to the advantages of DFB-SA, the system is easy to manufacture, has faster response time, and better control over gain current and reverse voltage. Compared to traditional continuous signals, pulse emission requires less energy, helping to reduce system power consumption. The reported system effectively implements DFB-SA based spiking reservoir computing and shows its successful application in complex nonlinear classification tasks. The system achieves favorable performance with lower input dimensionality, opening new avenues for future processing systems based on spiking reservoir computing machines integrated with laser saturable absorption regions and providing a new approach for lightweight system architectures.