Experimental Demonstration of Modulation Format Recognition Using Reservoir Computing Based on VCSEL With Gradient-Like Boosting

Abstract

We present a simple experimental approach based on a photonic time delay reservoir computing (RC) system for modulation format recognition. Here an optically injected vertical cavity surface emitting laser with single feedback is trained with the cross sequence of instantaneous characteristics of modulation signals. Three widely used modulation formats including on–off keying, binary phase shift keying, and binary frequency shift keying, where the optical signal-to-noise ratio varies from 4 dB to 36 dB are considered. Besides, we propose a post-processing method of using multiple gradient-like boosting operations for the time delay RC system, in which the residual value between the target value and the actual value is regarded as the target value for multiple training. Experimental and numerical results demonstrate that using multiple gradient-like boosting operations for post-processing can efficiently identify all three modulation formats with the highest accuracy of 100% after optimizing the control parameters such as the feedback strength, injection strength, and number of virtual nodes of the time delay RC system. The simple experimental approach with the proposed multiple gradient-like boosting operations offers a resource-efficient alternative approach to modulation format recognition of future communication.