Machine learning technologies have been extensively applied in high-performance information-processing fields. However, the computation rate of existing hardware is severely circumscribed by conventional Von Neumann architecture. Photonic approaches have demonstrated extraordinary potential for executing deep learning processes that involve complex calculations. In this work, an on-chip diffractive optical neural network (DONN) based on a silicon-on-insulator platform is proposed to perform machine learning tasks with high integration and low power consumption characteristics. To validate the proposed DONN, we fabricated 1-hidden-layer and 3-hidden-layer on-chip DONNs with footprints of 0.15 mm2 and 0.3 mm2 and experimentally verified their performance on the classification task of the Iris plants dataset, yielding accuracies of 86.7% and 90%, respectively. Furthermore, a 3-hidden-layer on-chip DONN is fabricated to classify the Modified National Institute of Standards and Technology handwritten digit images. The proposed passive on-chip DONN provides a potential solution for accelerating future artificial intelligence hardware with enhanced performance.
A novel symmetric WDM-PON scheme with colorless ONU is proposed. The baseband 4-ASK Fast-OFDM signal is upconverted by an intermediate frequency carrier, reserving a frequency gap between the FOFDM signal and the optical carrier. After distributing different wavelengths to corresponding ONU by AWG, periodic BPFs are employed to extract the optical carriers for upstream transmission, achieving colorless ONUs. A WDM-PON system with 16 colorless ONUs is established, and 10-Gb/s symmetric transmission for each ONU is also realized. Experiment shows that the system tolerance to the intrachannel crosstalk is greatly improved when the crosstalk signal locates at relatively higher frequency band.
A passively mode-locked Er3+-doped ZBLAN fiber laser around 3 μm with a wide wavelength tuning range is proposed and demonstrated. The laser cavity was comprised of a semiconductor saturable absorber mirror and a blazed grating to provide a wavelength tunable feedback. The central wavelength of the mode-locked fiber laser can be continuously tuned from 2710 to 2820 nm. The pulse train had a maximum average power of higher than 203 mW, a repetition rate of 28.9 MHz and a pulse duration of 6.4 ps, yielding a peak power of exceeding 1.1 kW. To the best of our knowledge, this is the first demonstration of a wavelength-tunable passively mode-locked mid-infrared fiber laser at 3 μm.
A diode-cladding pumped mid-infrared passively Q-switched Er3+-doped ZBLAN fiber laser with an average output power of watt-level based on a semiconductor saturable absorber mirror (SESAM) is demonstrated. Stable pulse train was produced at a slope efficiency of 17.8% with respect to launched pump power. The maximum average power of 1.01 W at a repetition rate of 146.3 kHz was achieved with a corresponding pulse energy of 6.9 μJ, from which the maximum peak power was calculated to be 21.9 W. To the best of our knowledge, the average power and the peak power are the highest in 3 μm region passively Q-switched fiber lasers. The influence of gain fiber length on the operation regime of the fiber laser has been investigated in detail.
A silicon-on-insulator (SOI) narrow-passband filter based on cascaded Mach-Zehnder interferometers (MZIs) is theoretically simulated and experimentally demonstrated, indicating that the free spectral range (FSR) of the proposed filter can be significantly enlarged by increasing the number of the MZI stages. A filter using three-stage cascaded MZIs structure is successfully realized in the experiment and a 3-dB bandwidth of about 1.536 GHz and FSR about 13.5 GHz have been achieved. The performance of a downconverting analog photonic link (APL) employing the designed filter for microwave signal processing is also measured and a spurious free dynamic range (SFDR) as high as 104.1dB-Hz(2/3) is observed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.