As deep neural networks increase their demand for computing power, traditional electronic architectures are experiencing bottlenecks in terms of computational speed and energy consumption. Optical computing, with its high scalability, high parallelism, fast operation, and low power consumption, is increasingly showing its advantages. To this end, we constructed an optical computing architecture based on an arrayed waveguide grating router (AWGR) optical convolution unit (OCU) to accelerate convolutional neural networks. The wavelength routing characteristics of the AWGR enable convolution operations to be mapped to the optical domain, achieving a linear relationship between input size and device number. We used the ResNet-50 neural network for demonstration experiments, trained and fine-tuned using the CIFAR-10 dataset. In this hybrid optical-electrical architecture, only the first-layer convolution operations were accelerated, while the remaining operations were performed in the electronic domain. Results show that, driven by a 2GHz FPGA, a single-frame convolution takes only approximately 0.3ms, resulting in a system processing capacity of over 3320 frames per second for color images and a theoretical peak computing power of 65.536 TOPS. The research results verify the potential of AWGR-OCU in efficient parallel computing and provide a feasible idea for the design of future large-scale optical acceleration AI computing chips. © 2025 SPIE.

We propose a calibration-free MZI and demonstrate its large-scale scalability with a 64 ×64 MZI. This robust methodology for suppressing random phase imbalance can be generalized for any essential phase-sensitive photonic integrated devices. © Optica Publishing Group 2025.

We propose and simulate a frequency-angle encoded frequency-modulated continuous wave (FMCW) lidar system that achieves high-fidelity 4D (x, y, z, and velocity) point cloud imaging via synchronized spatio-spectral modulation. By incorporating a programmable frequency-shifted FMCW transmitter with a focal plane switch array (FPSA), the system enables deterministic angular coordinate encoding within the optical frequency domain. At the receiver end, coherent detection enables direct extraction of scanning angle information from echo signal frequencies, along with target distance and velocity measurements. Compared with conventional FMCW systems, this innovation supports bistatic transmitter-receiver configurations and asynchronous operation, significantly enhancing system flexibility, signal-to-noise ratio, and detection speed. The methodology offers novel sensing solutions for autonomous driving and intelligent robotics applications. © 2025 SPIE.

Ultra-Compact 90 degrees Hybrid Design: The optical 90 degrees hybrid uses a 4 x 4 MMI coupler with an ultra-compact footprint of similar to 9 mu m x 107 mu m, which is 3 times shorter than previous work.Novel CMRR Calculation Method: A new method to assess the CMRR using a single unequal Mach-Zehnder interferometer (MZI), reducing fabrication complexity and measurement errors.Excellent Performance: The device features low excess loss (0.55 to 1.4 dB), low phase errors (< +/- 4.5 degrees), and high CMRR (< -17.8 dBe) over a wide wavelength range (1530 to 1605 nm)

In this paper, a 53 Gbps widely tunable transmitter is experimentally demonstrated for the first time, to our knowledge. An InGaAlAs/InP multiple-quantum-well (MQW) wafer is used with an identical layer structure for both the V-coupled cavity laser (VCL) and the electro-absorption modulator (EAM). The VCL uses a shallow-etched waveguide to reduce loss, while the EAM uses a deep-etched waveguide to increase the 3-dB modulation bandwidth. With the temperature varying from 19.5 to 30°C, the transmitter achieves wavelength tuning of 42 channels with a spacing of 100 GHz, corresponding to a tuning range of 32.6 nm from 1538.94 to 1571.54 nm. The static extinction ratio (ER) for all channels is higher than 14 dB. The measured 3-dB electro-optic (E0) bandwidth of the transmitter is over 40 GHz, which fits well with the calculated 3-dB bandwidth. At a fixed peak-to-peak driving voltage of 2.4 V, all channels exhibit clearly an open eye diagram with a 53 Gbps non-return-to-zero (NRZ) signal, while the dynamic ER is higher than 4.5 dB. © 2024 Optica Publishing Group.

In this Letter, we propose and demonstrate an integrated polarizer on thin film lithium niobate (TFLN). The polarizer consists of a width-tapered 180° Euler bending waveguide featuring thin thickness and bilevel mode convertors with silica cladding. Notably, the TE0 mode is efficiently confined in the waveguide, while the TM0 mode confronts significant bending losses. The measurements reveal that the excess loss remains below 1.5 dB, and the extinction ratio surpasses 19 dB within a working bandwidth spanning from 1480 to 1578 nm. The proposed polarizer holds considerable promise for enhancing polarization handling within TFLN photonic circuits. © 2024 Optica Publishing Group (formerly OSA). All rights reserved.

Optical convolution computing is gaining traction owing to its inherent parallelism, multi-dimensional processing, and energy efficiency. To handle input dimensions of N, conventional implementations necessitate N2 optical elements, such as Mach–Zehnder interferometers or micro-ring resonators, to process multiply-accumulate (MAC) operations, limiting scalability and resulting in elevated power consumption. Here, a direct convolution computing method based on wavelength routing, utilizing the unique sliding property of an arrayed waveguide grating router (AWGR) to perform the sliding window operation of the convolution in the wavelength–space domains is proposed. With two input vectors directly loaded onto two modulator arrays, the convolution result is instantaneously produced at a photodetector array. The entire convolution computation is executed within a single clock cycle without the need for preprocessing or decomposition into elementary MAC operations. The number of active elements is minimal, only needed for input/output. The proposed optical convolution unit has striking advantages of high scalability, high speed, and processing simplicity compared to those based on optical matrix-vector multipliers. In the first experimental demonstration, a remarkable classification accuracy of up to 98.2% in handwritten digit recognition tasks using a LeNet-5 neural network is achieved. © 2024 Wiley-VCH GmbH.

A high performance optical phased array (OPA) combined with frequency-modulated continuous-wave (FMCW) technology is essential for coherent all-solid-state light detection and ranging (LiDAR). In this work, we propose and experimentally demonstrate a coaxial transceiver based on a single OPA for a LiDAR system, which releases the off-chip circulator and collimator. The proposed scheme is demonstrated on the commonly used silicon-on-insulator (SOI) platform. For realizing the long optical grating antenna with only one-step etching, the bound state in the continuum is harnessed to simplify the fabrication process and ease the fabrication precision. Experimental results indicate that the OPA is with 0.076° vertical beam divergence under a 1.5 mm-long grating antenna. The measured field of view (FOV) is 40° × 8° without grating lobes under a wavelength band of 60 nm. The coaxial transceiver of the single OPA is also demonstrated with the FMCW method for ranging measurement at different angles. © 2024 Optica Publishing Group.

An optical phased array (OPA) with 2-D circular sparse array aperture has been proposed and demonstrated in the silicon integrated photonic platform. The sparse distribution of the antenna array can realize no grating lobes in 2-D full field of view (FOV). To achieve fast and accurate phase calibration for OPA, an improved rotating element electric field vector algorithm based on golden section search method (GSS-REV) has also been proposed and verified. The 32-element antenna sparse distribution of the proposed OPA is designed and fabricated. A far-field beam steering measurement across 20° × 20° range features the side lobe suppression ratio (SLSR) of larger than 4.81dB and a full width at half-maximum (FWHM) of approximately 0.63° × 0.59°. The resolvable points are derived to be ∼1076. The OPA chip has also been demonstrated on range measurement with frequency-modulated continuous-wave (FMCW) system. © 2023 the author(s), published by De Gruyter, Berlin/Boston 2023.

Large-scale photonic switches are essential components for photonic circuits for classical and quantum applications. To develop high-performance large-scale photonic switches, we propose and demonstrate a low-loss, C+L+S-band (1460–1625 nm), and low-random-phase-imbalance 2 × 2 Mach-Zehnder switch (MZS) designed with fast quasi-adiabatic couplers (FACs). The FAC is designed to accelerate slow adiabatic evolutions by a tailored geometry tilt, and further optimized by the particle swarm optimization (PSO) algorithm. It features a simulated uniform splitting ratio that varies between 52.5%:47.5% and 47.9%:52.1% with a low excess loss of 0.020–0.054 dB across the C+L+S band. To achieve a large operation bandwidth of the MZS, 1-μm width phase shifters are implemented to reduce the wavelength dependence of the half-wave heating power Qπ of the thermo-optic (TO) phase shifter. They also suppress random phase imbalance caused by the size variations in fabrication between the MZS arms. The simulated excess loss and extinction ratio of the present MZS are 22.6 dB across C+L+S band. Such an MZS has been fabricated with standard 180-nm silicon photonic foundry processes, featuring a low excess loss of ∼0.25 dB and an extinction ratio of >17 dB across the C+L+S band. Meanwhile, an extinction ratio of >20 dB is achieved across a slightly narrower bandwidth, 1470–1617 nm, with an excess loss of ∼0.2 dB. The MZS is an attractive option for scaling up N × N MZSs, which would be beneficial for large-scale and ultrahigh-capacity WDM and BDM telecom and interconnect applications, as well as on-chip near-infrared spectrometers. IEEE