Delay-based photonic reservoir computing: phase encoding in long vs. short cavity reservoirs

Time-delay photonic reservoir computing (TDRC) is a photonic system that is simple to implement and has shown great efficiency in handling a variety of complex tasks. Several implementations have been proposed, including reservoirs that consist of a semiconductor laser with external optical feedback and optical injection. Such systems were successful in solving signal equalization tasks, including data recovery in fiber transmission systems, demonstrating their potential for real-world applications. TDRC has been used effectively to recover signals that are distorted by optical fiber transmission, particularly when the information is encoded in the amplitude domain. The optical-to-electrical signal conversion is performed using only the amplitude information. Here, I will present that TDRC efficiency is not limited to just incoherent signals. It is possible to process coherent signals, introducing also the phase information of the signals to be processed to the reservoir. I'll show the advantages of this approach, verified numerically and experimentally. I will also address the challenges of such implementations in terms of robustness and stability, particularly when using a long fiber-optic cavity. To resolve such issues, I will introduce strategies based on short-cavity designs that favor stability along with distinct dynamical behaviors of the photonic reservoir.

https://zoom.us/j/98286706234?pwd=bm1JUFVYcTJkaVl1VU55L0FiWDRIUT09