The rise of the big data era presents major challenges for information processing, particularly in terms of handling large volumes of data and managing energy consumption. These issues are further compounded by the fact that over 90% of data is transmitted using light waves, while the actual processing still predominantly occurs in the electrical domain. To address this mismatch, two main approaches have emerged. One involves converting signals from optical to electrical and back again (optical-electrical-optical, or O-E-O conversion). The other focuses on processing the data entirely within the optical domain, a method known as all-optical information processing (AOSP).
While O-E-O conversion encounters significant limitations, including constraints related to transparency and challenges with achieving parallelism using optoelectronic components, AOSP offers a more scalable alternative. With the right nonlinear processes, AOSP can achieve improved system performance in terms of complexity, cost, and energy efficiency. Interest in AOSP dates back to the 1980s, when it was initially explored using bulk nonlinear devices. However, recent breakthroughs in photonic integration have significantly accelerated its development.
Among the various platforms for integration, silicon-based photonics has emerged as one of the most promising for advancing AOSP technologies. Silicon photonics supports a wide range of functionalities that are closely tied to the architecture of modern optical networks. To meet future demands, optical networks must demonstrate the capabilities of 3T (format transparency, wavelength transparency, bandwidth transparency), 3M (multi-function, multi-channel, multi-network), and 3S (self-perceiving, self-learning, self-adopting). Therefore, achieving a high degree of reconfigurability and adaptability is essential for both future optical networks and the broader application of AOSP in systems requiring ultra-large capacity.
A collaborative team of researchers, including Prof. Xinliang Zhang (Huazhong University of Science and Technology), Prof. Yikai Su (Shanghai Jiao Tong University), Prof. Kun Qiu (University of Electronic Science and Technology of China), and Academician Ninghua Zhu (Nankai University), has successfully developed a monolithically integrated, programmable all-optical signal processing (AOSP) chip. This chip supports key functions such as optical filtering, signal regeneration, and logic operations. The project stems from a national initiative aimed at creating silicon-based, reconfigurable AOSP technologies.
To read more, click here.