Abstract This research investigates the experimental implementation of a feed-forward (FF) optical neural network using optical injection (OI). Two follower laser diodes (FLD1, FLD2) are subjected to chaotic modulation, with their respective signal weights adjusted via optical filtration. The study analyzes the behavior of these FLDs during FF operation, focusing on frequency spectra derived from time series data. The research examines the impact of both signal weights and control parameters, including the bias voltage of the influencer laser diodes (ILD1, ILD2), on the FLDs. A maximum FWHM of 1.7 GHz is observed for FLD2 when both the bias voltage is set to 4V, and the modulated signal is attenuated to -20 dB. To assess the synchronization state, the correlation between the ILDs and FLDs is calculated. Results show fluctuations between positive and negative values, with a maximum correlation of 0.42 observed for FLD1 when influenced by FLD2's voltage. These findings demonstrate an anti-synchronized relationship between the ILDs and FLDs, a crucial requirement for ensuring privacy in a chaotic optical communication system, simulating an optical neural network. (ILD1,ILD2) bias voltage in additional tow FLDs. Maximum FLD2 measured FWHM 1.7
Read full abstract