Efficient Modulation Format Identification Using Transfer Learning

Abstract

An efficient modulation format identification (MFI) at an optical signal-to-noises ratios (OSNRs) spanning from 20 to 30 dB is proposed using the transfer learning (TL) technique. Transmission setup are created to demonstrate the technique for 8QAM, 16QAM, 64QAM, and 128QAM systems. TL can process constellation diagrams from an image processing approach owing to its self-learning capabilities. The obtained research shows that even at low OSNR, the suggested techniques may accurately be utilized to detect the modulation format with classification rates up to 100%The suggested method can intelligently analyse the basic hardware to allow MFI, and the analysis results are used to identify additional modulation schemes at various transmission rates for better management of optical systems.