Digital Back Propagation

Digital backpropagation (DBP) and its learned variant (LDBP), using deep neural networks, are being explored to improve signal processing in various applications, particularly in optical communication systems and hybrid optical-digital neural networks. Current research focuses on optimizing DBP algorithms for enhanced performance in mitigating nonlinear effects in high-speed fiber optic transmission and improving the efficiency of hybrid optical-digital systems by co-optimizing optical and digital components. These advancements offer potential for significant improvements in data transmission speed and energy efficiency, impacting fields ranging from telecommunications to artificial intelligence hardware.