Theoretical investigation of ultrafast binary data patterns recognition based on VCSEL-SA

Abstract

Based on a vertical-cavity surface-emitting laser with saturated absorber (VCSEL-SA) subject to optical injection, we proposed an ultrafast pattern recognition scheme of four-bit binary data and theoretically investigated the recognition performances. The results show that, patterns recognition of different four-bit binary data at Gb/s rate can be realized by adjusting the injection weight of each bit number and optimal weight values can be determined. Although noise has some influences on the patterns recognition speed and accuracy, this proposed system has a certain robustness to noise on the whole. These results provide a promising application prospect for VCSEL-SA based ultrafast photonic neuromorphic system in pattern recognition field.