Signal processing is always the heart of the overall information technology and industry, providing enabling solutions for the processing, understanding, learning, retrieval, mining, and extraction of information from different signals. Regarding the all-optical signal processing, it dates back to the 1980s when the electronic bottleneck was considered as a big obstacle to the bandwidth and capacity of optical communications. Fortunately, all-optical signal processing provides intrinsic wide instantaneous bandwidth and large time-bandwidth product (TBP), powered by optics or photonics technology. Although we are happy to see various breakthrough achievements in electronic signal processing over past decades, all-optical signal processing are still highly required today along with the exponential growth of data capacity, the ubiquitous access, and the realtime interactions. There are some increasing motivations listed as follows, all-optical storage and switching to for large-capacity and energy-efficient networks, broadband and complex microwave signal processing for 5G and beyond systems and defense scenarios, and innovative materials and devices for quantum information processing. In particular, all-optical signal processing is of great significance in both optical and radio frequency (RF) domains, providing intrinsic advantages on large bandwidth to circumvent the electronic bottleneck. Here, latest and state-of-art advances on all-optical processing to both optical and RF signals are comprehensively reviewed, with warped transform and compression, integrator and differentiator, optical memory, fully photonics-based radar, and nonlinear photonic RF signal processing being highlighted in details. Moreover, advances on photonic integrated circuits (PICs) for all-optical processing are demonstrated, especially the CMOS-compatible PICs and high-density nanophotonic integration. Optical signal processing refers to a wide variety of techniques and applications, including compression, transform, analog/logic operations, arbitrary waveform processing, compensation and equalization, wavelength and format conversion, all-optical storage/switching/ memory [1–10]. Here, we will concentrate on the latest advances on warped transforms and compression, differentiation and integration, and optical memory. First of all, a concept of the photonic hardware accelerator is proposed to process wideband waveforms [1]. In principle, nonlinear group delay dispersions (sub-linear or super-linear ones) are introduced to perform warped transform or anamorphic stretch transform, such that wideband waveforms are transformed and processed according to the non-uniform entropy of their spectra. Potential applications include non-uniform Fourier domain sampling, data compression, Big Data predicament in biological cell screening, signal-to-noise ratio (SNR) enhancement, spectrotemporal encoding, and so on. For example, a compression in the TBP of a signal is achieved to circumvent the Big Data problem in high-throughput measurements and ultrafast imaging [1, 2], without sacrificing the spectral resolution or bandwidth. The authors demonstrated a modulation bandwidth compressed by 500 times, while the input optical signal had a 1,000-GHz modulation bandwidth and a 50-ps duration. Consequently, the TBP and then the digital data size were reduced by 2.73 times. These requirements are already available in optics X. Zou (&) W. Pan L. Yan Center for Information Photonics and Communications, Southwest Jiaotong University, Chengdu 610031, China e-mail: zouxihua@swjtu.edu.cn
Read full abstract