Microwave photonic radio-frequency (RF) frontends, exhibiting the advantages of large frequency coverage, reconfigurable operation frequency bands and instantaneous bandwidths, and immunity to electromagnetic interference, have shown great potential in wireless communication, software-defined radio, radar systems, and electronic warfare. To further balance the performance superiority with the requirements of miniaturized application scenarios on size, weight, and power consumption (SWaP), the implementation of integrated microwave photonic RF frontends has been considered as an upgrade path. In this article, the challenges of the demonstration of integrated microwave photonic RF frontends are briefly analyzed from the device design to the microsystem integration, and then the research of our group on the hybrid integrated reconfigurable microwave photonic RF frontend are presented from such three aspects: the design of reconfigurable narrowband integrated photonic filters, the architecture of hybrid integrated microsystems, and the method for suppressing the frequency drift.

Due to the advantages of rich dynamics, small size and easy integration, semiconductor laser-based signal generation technology has become a potential candidate for high-performance microwave photonic signal generation. Under proper optical injection, semiconductor lasers can operate at the period-one (P1) oscillation state and break through the limitation of the intrinsic relaxation oscillation frequency to realize the generation of the widely tunable microwave signal. In addition, assisted by dynamical control of injection parameters, wideband reconfigurable microwave frequency-modulated (LFM) signals can be generated, which has important applications in radar systems. In this paper, the research progresses of the wideband radar signal generation and application based on an optically injected semiconductor laser were reviewed. Firstly, microwave LFM waveforms with large time-bandwidth products were successfully generated by exploring the controlled P1 dynamics of an optically injected semiconductor laser. The main operating parameters are adjustable, including the central frequency, bandwidth, frequency band, and temporal period. Secondly, a delay-matched optoelectronic feedback structure was introduced to greatly improve the spectral purity and phase coherence of the generated wideband radar waveform. Finally, by applying the generated wideband waveforms, a microwave photonic radar system was experimentally demonstrated, and its high-resolution target detection and imaging ability were verified.

Intelligent photonic processing system (IPS) combines artificial intelligence (AI) technology and photonic technology to achieve intelligent, high-speed, large-bandwidth and high-performance signal processing. IPS mainly includes AI-powered multifunctional photonic processing system, intelligence signal processing with photonics-facilitated AI and neuromorphic photonic processing system. In this article, the concept of IPS is briefly introduced first,then the research progresses of our resaerch group in the AI-powered multifunctional photonic processing system are described, and further the development trends and necessity of explainable AI are discussed. And the research of our group on the explainable AI-powered multifunctional photonic processing system is introduced.

High-performance photonic analog processing chip is the core component of microwave photonics processing system. In this paper, an ultra-wideband reconfigurable photonic analog computing chip was realized by optimizing the optical waveguide network structure, which can realize arbitrary switching and tuning of various analog signal processing functions by configuring topological networks. The optical matrix computing chip with self-configuration capability and the on-chip photonic convolution accelerator for image processing were also studied. Finally, prospects of cross research between microwave photonics and artificial intelligence were provided.

A microwave photonic compressive sensing radar is proposed for distance measurement based on cascaded Mach-Zehnder modulators (MZMs). In the transmitter, a dual-parallel Mach-Zehnder modulator (DPMZM) is used to generate the quadrupled linearly frequency-modulated (LFM) signals. In the receiver, the echo signal can be de-chirped and the de-chirped signal can be mixed with a pseudo-random bit sequence (PRBS) in the optical field by using cascaded MZMs. Then the mixed signal can be acquired by a low pass filter (LPF) and sampled at a sub-Nyquist sampling rate by using a low-speed analog-to-digital converter (ADC). Finally, orthogonal matching pursuit (OMP) algorithm can be used to recover the de-chirped signal. The results of the simulation and proof-of-concept experiment demonstrate that when the data compression ratio is 8, the system can reconstruct the frequency of the de-chirp signal with less observed value and the distance measurement error is less than 2cm. In addition, it reduces the pressure of data sampling, storage and information processing, and it breaks Nyquist sampling theory.

Based on analyzing the latest development trends of digital array radar, in this paper, it is summarized the problems that need to be solved in the aspects of modulation freedom, signal characteristic processing, parameter control and algorithm acceleration in the development of digital array technology to “wideband, discrete, intelligent and flexible”. Meanwhile, the suitability, advantages and disadvantages of microwave photonic technology are discussed. For pushing forward the progress of practical implementation, the prospects of several microwave photonic technologies in digital phased array radar system are analyzed, including photonic analog-to-digital conversion, photonic quadrature demodulation, photonic arbitrary waveform generation, RF bus, etc.

With the development of electronic technology, constant beamwidth beam forming techniques are widely applied in electronic reconnaissance system in complex electromagnetism environments. In this article, the principles of wideband constant beamwidth optical multi-beam forming are introduced, the method is introduced for obtaining the sub-frequency band weighting matrix for shaping frequency invariant beam pattern by means of window functions, and the system design scheme of wideband constant beamwidth optical multi-beam forming is proposed. Simulation results demonstrate that the proposed method can form the pattern with constant beamwidth and better direction.

Ultra-broadband optoelectronic devices are key components for optical fiber transmission and processing signals in next-generation wireless communications and advanced electronics systems. The interactions between photons, electrons, and electromagnetic fields are the fundamental problems to determine performance of broadband optoelectronic devices. In this article, our recent progresses in ultra-wideband photo-detectors and electro-optic modulators are introduced to share key technical solutions of such problems to improve the device performance.

Compared with the traditional electronic analog-to-digital converter, the photonic analog-to-digital converter (PADC) has the advantage of high sampling rate, large bandwidth, low timing jitter and low comparator ambiguity. PADC provides an effective solution for high-speed, large-bandwidth, and high-resolution processing in ultra-wideband radar, ultra-high speed oscilloscope, high-capacity optical communication and so on. In this paper, different architectures of PADC are classified and compared. Then, the recent progresses of our group in channel-interleaved PADC are introduced. Besides, the status of integrated PADC are analyzed briefly and the main challenges of future integrated PADC are discussed.

Conventional microwave photonic link is realized by using intensity modulation method to complete the electro-optical conversion. Since the Mach-Zender interference structure (MZI) is used for the intensity modulator, the system performance is not only restricted by its own sine-like response characteristics, but also additional bias controller is needed to maintain the bias point. Hence the dynamic range is limited, and also 3dB inherent loss due to quadrature point will decrease the transmission efficiency. Phase modulation can avoid these problems. In this paper, two methods to convert phase modulation to intensity modulation by using optical filter through sideband suppression and sideband passing are proposed, and the relationship between link RF performance and device parameters is also analyzed. Finally, the performance between the phase modulated link and the conventional intensity modulated link are measured and compared. The results show that the phase modulated link has higher transmission efficiency under the condition of same link configuration, and the 3dB bandwidth of the phase modulated link is twice larger than that of the intensity modulated link due to the equalization effect of optical filtering.

In order to meet the requirements of advanced electronic systems such as microwave communication, radar, electronic warfare and navigation, a turnkey coupled optoelectronic microwave oscillator with low phase noise based on Sigma-shaped fiber ring structure is proposed. In this scheme, the Sigma-shaped fiber ring structure is used to eliminate the harmful influence of the polarization-induced fading, and the quality factor of the oscillator is greatly enhanced by the regenerative gain characteristic of the active cavity. As a result, a stable 10GHz signal with low phase noise is generated as soon as the oscillator is powered on. The phase noise at 10kHz offset frequency is about -139.26dBc/Hz.

A broadband microwave photonic front-end that suits for electronic warfare is proposed in this paper. Significant improvement in noise figure and spur-free dynamic range is accomplished by using double sideband suppressed carrier modulation and balanced coherent heterodyne detection. In the frequency bandwidth of 6~18GHz, a spur-free dynamic range up to 110dB·Hz2/3, a noise figure of 8dB, an amplitude consistency better than ±1dB and a phase consistency better than ±10° are implemented by using coherent detection, which makes it especially applicable for microwave photonic beam-forming, in addition to the RF transmission.

The development of optically controlled phased array antenna is reviewed briefly, the working principles of optically controlled phased array antenna are introduced, and the advantages of optically controlled phased array antenna in satellite payload are discussed. It is pointed out that the application of high throughput satellite ultra-wideband and multi-function load multi-band is one of the development trends of satellite communication, and the optically controlled phased-array antenna has obvious advantages in the application of large aperture, ultra-wideband and wide angle scanning.

High-speed electro-optic modulator is one of the key components in broadband optical communication networks and microwave photonic systems. Compared with the legacy lithium niobate, the thin-film lithium niobate material has unique advantages in constructing high-performance electro-optic modulation chips with small size, broadband and low half-wave voltage due to its strong optical field confinement ability. In this paper, an electro-optical modulation chip with a 3dB bandwidth not less than 50GHz was developed based on thin-film lithium niobate materials, and a fully packaged thin-film niobate electro-optic modulators was realized by using an optical packaging scheme of horizontal edge coupling between optical fiber and waveguide and a radio frequency packaging scheme based on 1.85mm coaxial connectors. Measurement results indicate that the optical insertion loss of the packaged device is less than or equal to 5dB, the 3dB bandwidth is greater than or equal to 40GHz, and the RF half-wave voltage is less than or equal to 3V@1GHz.

Zinc oxide (ZnO) is a natural semiconductor with a theoretical bandgap of 3.37eV. It has become one of the most popular materials in the application of ultraviolet photodetectors in recent years. However, due to the intrinsic defects of ZnO, the directly prepared ZnO ultraviolet photodetectors always display low responsivity, high dark current and slow response speed. To obtain the better performance of ultraviolet photodetectors, various feasible methods for performance improvement and modification are proposed. In this paper, the typical methods to improve the performance of ZnO ultraviolet photodetectors are reviewed from the three aspects of element doping, surface modification and hetero-structure construction, the existing problems of these methods are pointed out and the potential development of high-performance ultraviolet photodetectors in the future is prospected.

A terahertz photoconductive antenna with adjustable polarization is proposed, which consists of four arc structure metal electrodes, a low-temperature gallium arsenide substrate and an anti-reflection layer of silicon nitride. Through the analysis of the working principles of photoconductive antenna, it can be found that the arc metal electrode structure determines the radiation polarization of the antenna, and the carrier mobility, carrier life and laser absorption of the substrate material can directly affect the radiation characteristics of the antenna. The photoconductive antenna is modeled and simulated by COMSOL software, and the results show that the antenna can multiply the outgoing polarized terahertz wave in the 45° direction, and the radiation intensity is 30% higher than that of the conventional photoconductive antenna and the radiation bandwidth is up to 10THz. The designed terahertz photoconductive antenna has the characteristics of adjustable polarization, simple structure and easy processing, owning broad application prospects in the field of terahertz spectrum detection.

For the imaging requirements such as high frame rate, global shutter and flat spectral response characteristics, a kind of high-frame rate CMOS image sensor for hyperspectral imaging was designed. The circuit employed 5T (5-transistor) photodiode pixel for global shutter operation. And high-speed readout and signal processing circuits were designed to achieve low fixed pattern noise (FPN) and non-uniformity. The sensor was fabricated in ASMC 0.35μm standard CMOS process. In order to suppress the spectral interference of the photodiode, back-end spectral flatness process was implemented, and the photoelectric response of the device was tested and evaluated. The test results indicate that the circuit has met the design expectation, resulting in pixel array size of 512×256, pixel size of 30μm×30μm, high full well capacity of 400ke-, wide dynamic range of 72dB and frame rate of 450f/s. The quantum efficiency difference between neighboring 10nm spectral band is less than 10%, being able to meet the general requirements of a hyperspectral imaging system.

A high-speed back illuminating InAlAs avalanche photodiode(APD) was developed to meet the requirements of single-channel 25Gbit/s long-distance transmission. The APD adopts vertical back-illumination separate-absorption-grading-charging-multiplication (SAGCM) structure. Triple-layer mesa was formed by etching process, and the electric field was limited to the center of the maximum mesa multiplier layer, effectively reducing the mesa edge breakdown. The chip adopts reverse welding structure with integrated microlens on the back to improve optical coupling aperture. The responsivity of the developed APD chip is 0.84A/W (λ=1310nm) at the gain of M=1. When M=10, 3dB bandwidth reaches 19GHz. The gain-bandwidth product is 180GHz. The best sensitivity at 5×10-5 bit error rate is -22.3dBm, which can support 100GBASE-ER4 communication standard.

The existing cantilever FBG acceleration sensor can be subjected to uneven force due to the optical fiber surface adhesion, and can not work in the complex environment of temperature change and vibration. In order to solve such problems, a dual fiber-cantilever structure FBG acceleration sensor is proposed. The influence of structural parameters on the sensitivity and natural frequency of the sensor was analyzed theoretically, and the static stress and modal simulation analysis was carried out by ANSYS finite element analysis software. Finally, a test system was built to test the performance of the sensor. The results show that the natural frequency of the acceleration sensor is 84.86Hz, and it has a flat sensitivity response in the low frequency range of 15~60Hz. The dual optical fiber can effectively eliminate the influence of temperature changes while increasing the sensitivity of the sensor, obtaining the acceleration sensitivity of 156.70pm/g and the linearity of 99.38%. The rigid beam effectively increases the stability of the structure, and the lateral crosstalk within the working frequency band is -26.97dB.

SnO2 buffer layers with different thicknesses were prepared on silicon substrate by electron beam evaporation, and their influence on the microstructure, phase composition and phase transformation properties of the upper vanadium oxide films deposited by magnetron sputtering were investigated. Experimental results show that the lattice of tetragonal-rutile structure SnO2 matches well with that of vanadium oxide films, leading to the thin films contains more V4+. When the thickness of SnO2 buffer layer increases, accompanied by the appearance of more homogeneous sizes of SnO2 crystallite, larger grain sizes, narrower thermal hysteresis loops and sharper phase transition can be observed for the deposited VO2 films. Accordingly, this work indicates that the addition of SnO2 buffer layer with optimal thickness is beneficial to the growth of high-quality VO2 films with excellent phase transition performance.

A two-step hydrothermal method was applied to create WO3/NiWO4 composite films on conductive glass (FTO). The structure and morphology of WO3/NiWO4 composite films were characterized by XRD and SEM. The photoelectric properties of WO3/NiWO4 composite films were analyzed by UV-VIS, photocurrent test, photoelectrocatalytic test and AC impedance test. The analysis results indicate that the WO3/NiWO4 composite film has better light absorption characteristics, photocurrent density and photoelectrocatalysis activity than WO3 film, and the WO3/NiWO4 composite film with 3 hours hydrothermal reaction has the best photoelectrochemical properties. The photocurrent density of WO3/NiWO4-3h was 1.94mA/cm2 at 1.4V(vs. Ag/AgCl), and the photoelectrocatalytic efficiency of WO3/NiWO4-3h for methylene blue solution was 57.1% for 210min. The AC impedance diagram reveals that the charge transfer resistance of WO3/NiWO4 film is less than that of WO3 film, which corresponds to improved photoelectrochemical properties.

Based on constructal theory, a heat dissipation model of two-dimensional double-sided square column discrete heat sources on a heat conduction base in a jet channel is established. The total longitudinal section area and the height of the discrete heat sources are set as the constraints, the optimization objectives are as follows: minimizing the maximum temperature, the temperature uniformity factor and the temperature gradient uniformity factor of the system, and the length ratio between each heat source is set as the optimization variable, then the effects of the jet speed and the heat source spacing on heat dissipation are analyzed. When the heat source spacing and jet speed are given, the maximum temperature, the temperature uniformity factor and the temperature gradient uniformity factor of the system can be reduced by changing the length ratio between the heat sources. When the heat source spacing and the length ratio between heat sources are given, with the increase of the jet velocity, the maximum temperature, the temperature uniformity factor and the temperature gradient uniformity factor of the system all decrease at different degrees.

Based on Newton law of viscosity and computational fluid dynamics (CFD), on the basis of the actual project, it explores the cooling performance of finned radiators with five different vertical sections under the condition of forced convection. According to their respective heat dissipation performance and economy, the optimum vertical cross section shape of the fin type radiator is selected to be the triangular section. Compared with the traditional rectangular fin radiator, its heat dissipation performance is improved by 4.23% and the material cost is reduced by 17.95%.

In order to improve the accuracy of homography estimation and solve the problem that it is difficult to obtain real labels, an unsupervised homography estimation algorithm with correction function is proposed. The algorithm adopts a cascade structure, and its idea is similar to iteration, in which each level network maintains the same number of layers and parameters, and the homography of the output of the next level network is estimated as the residual of the sum of the real matrix and the previous output homography matrix. Considering the requirements of model complexity and real-time, a two-level network cascade is adopted in this paper. Through the verification on 5000 images in the COCO dataset, it shows that the cascade unsupervised algorithm designed in this paper has more accurate estimation ability than traditional methods and other methods based on deep learning. Its average pixel error in the test set is 0.54, which is 95.38% lower than that of traditional methods, and the running speed reaches 95f/s.

Most current face recognition algorithms are difficult to deploy to mobile terminals and embedded devices because of their many parameters and large amount of calculation, thus a face recognition method based on improved MobileFaceNet is proposed. By adjusting the structure of the MobileFaceNet model, the bottleneck module is optimized to the sandglass module, the relative position of depthwise convolution and pointwise convolution are improved, and the number of output channels of the sandglass module is appropriately increased, thereby reducing information loss during feature compression and enhancing extraction of the facial spatial features. The experimental results show that the improved method achieves an accuracy of 99.15% on the LFW test data set, and the model size and calculation amount are only 61% and 45% of that of the original algorithm, verifying the effectiveness of the proposed method.

In the task of camera pose estimation, the measurement of the position of the reference point is easily affected by the measuring instrument or the environment, leading a certain degree of measurement error. And the larger error point has a great influence on the final solution accuracy, which is called the outliers. Aiming at the problem that the existing algorithms have weak ability to resist outliers, a fast camera pose estimation algorithm based on measurement error uncertainty weighting is proposed. Based on the classic orthogonal iterative algorithm, the algorithm introduces M estimation method, assigns the corresponding weight to the objective function according to the measurement error of the reference point, eliminates the influence of the outlier point to the greatest extent, and introduces the Kronecker product to optimize the calculation process. Thus the redundant calculations in the iterative process are reduced and the calculation speed is increased. The simulation results show that the improved algorithm not only improves the anti-outliers ability of the original algorithm, but also realizes faster calculation speed.

The effect of atmospheric refraction on the background light bit error rate of the satellite-to-ground quantum key distribution (QKD) system was studied. Three different background light conditions of sunny day, full moon night, and moonless night were considered. First, a stratified satellite-to-ground QKD channel transmission model was established to simulate the influence of atmospheric refraction effect on the transmission distance. Then based on the model, the bit error rate of the background light varied with the zenith angle was simulated and analyzed for the cases in the presence and absence of the atmospheric refraction effects. The results show that, the atmospheric refraction has a certain impact which should not be neglected on the background light bit error rate. In addition, the relatively lowest bit error rate can be obtained under the moonless night background light condition.

In order to meet the monitoring requirements of three-way load of a certain aircraft landing gear, a landing gear load test method was proposed based on fiber Bragg grating (FBG) sensing and multiple linear regression. The strain distribution of the landing gear under different loads was obtained with finite element analysis software. According to the characteristics of strain distribution, the position of the FBG strain sensor was determined, and the strain sensing network was formed. On this basis, the ground static load calibration experiment of the main landing gear was carried out, and the strain-load calculation equations were formed by multiple linear regression method. Finally, the load calculation equations were verified by experiments. The research results show that the method based on fiber Bragg grating can effectively back-calculate the three-way load of landing gear, and the relative error between the back-calculated load and the applied load is within 5%, which meets the actual engineering requirements.

In order to improve the real-time performance in situation awareness of intelligent optical fiber composite overhead line, the artificial neural network (ANN) method is introduced into the strain demodulation along the optical fiber. The structure of the ANN for strain demodulation is determined. The programs of the least-squares spectrum fitting method and ANN method based on Lorentzian model are written. The ANN is trained by the Brillouin spectra with different signal-to-noise ratios (SNRs) and Brillouin frequency shifts. The trained ANN is applied to the optical fiber strain measurement for an optical fiber composite overhead line. The results of the two methods are compared from different aspects. The results indicate that the ANN method can effectively obtain the Brillouin frequency shift along the optical fiber and then obtain the strain, which has the similar accuracy as the spectral fitting method. However, its computation time is much less than that of the spectrum fitting method. The work provides a reference value for improving the real-time performance of situation awareness of intelligent optical fiber composite overhead lines.

Aiming at the practical requirement of accurately calibrating external parameters of multiple solid-state lidars, a multi-solid-state lidar automatic calibration algorithm based on point cloud registration is proposed. The calibration algorithm consists of three stages: calibration object segmentation, initial registration and precise registration. In the target segmentation stage, a calibration point cloud is made by superimposing multiple frames of non-repetitive scan data, and then radius filtering and voxel down-sampling filtering are used to segment the target point cloud with obvious texture characteristics. In the initial registration stage, key points are extracted by 3D-HARRIS algorithm, and then key points are described by the signature of histogram of orientation (SHOT) feature descriptor, then the sampling consensus algorithm is performed to calculate the corresponding points and complete the initial registration. In the fine registration stage, the iterative closest point (ICP) algorithm performs precise registration to obtain precise external parameter calibration effects. Experiments are conducted on the Bunny rabbit dataset and the data obtained on site. The results show that the performance of the proposed algorithm is better than many existing algorithms under the premise that the average registration error is less than 1mm.