Experimental Demonstration on Enhanced Separation of Mixed High-Dimension Optical-Chaotic Signals Using Double Photonic Reservoir Computing Based on Optically Pumped VCSELs

We present an experimental methodology designed to separate two groups of mixed optical chaotic signals, whether the mixing fractions are known or unknown. This separation is achieved using a VCSEL-based reservoir computing (RC) system. In the experiment, one group of mixed optical chaotic signals is linearly combined with two beams of chaotic X-polarization components (X-PCs) or Y-polarization components (Y-PCs) emitted by optically pumped spin-VCSELs with optical feedback. Double reservoirs are formed using the chaotic X-PC and Y-PC outputs from the optically pumped spin-VCSEL, which is subjected to both optical feedback and optical injection. Moreover, we experimentally demonstrate the performance of separating each group of linearly mixed chaotic signals into their individual components. The results show that two groups of mixed optical chaotic signals can be effectively separated using two reservoirs in a single RC system, with separation errors, characterized by normalized mean square error, being no more than 0.1 through system parameter optimization when the mixing fractions are known in advance. If the mixing fractions are unknown, we utilize two cascaded RC systems to separate each group of mixed optical signals. The mixing fractions can be accurately estimated using double reservoirs in the first RC system. Based on these estimated mixing fractions, the two groups of mixed optical chaotic signals can be effectively separated using double reservoirs in the second RC system, with separation errors also being no more than 0.1 through further optimization of the system parameters. The photonic reservoir computing hardware proposed in our experiment for separating complex optical chaotic signals has the potential to significantly impact the development of novel principles and hardware implementations for multiple access and demultiplexing in multi-channel chaotic cryptographic communication.