A 8-pole terahertz(THz) cavity bandpass filter with low insertion loss based on H ladder structure is presented. The filter is simulated and optimized by Ansoft HFSS, and it is finally processed by computer numerical control milling. The measured results show that the center frequency of the filter is located at 179.1 GHz with 1 dB fraction bandwidth of 8.4%. The insertion loss at 179.1 GHz is 0.34 dB, and the return loss is lower than 18.9 dB. The proposed THz cavity bandpass filter possesses the merits of low insertion loss, high selectivity and high stopband suppression, which can meet the requirements of THz communication systems.

It is difficult for the terahertz radiation to interact electromagnetically with most materials in nature due to its spectral specificity, which results in the terrible scarcity of terahertz functional devices. The terahertz metamaterial which is periodic arrangement of artificially designed units with subwavelength, can regulate the electromagnetic response of the terahertz band and become the research hotspot. In this paper, a kind of terahertz band-stop metamaterial with a single-layer metal square resonant ring on silicon dioxide substrate is designed, which has a narrow width and deep band-stop characteristics, and is insensitive to polarization. The band-stop resonance is attributed to the electric dipole resonance in the metal square resonant ring. Moreover, the proposed structure is easy for fabrication, and will be applied in the fields of terahertz modulation devices, THz communication and photoelectric detection.

A monolithic integrated mixer based on InP material Schottky diodes is introduced, and its operating frequency is 560 GHz. The mixer uses a new type of thin-film hybrid transmission structure, and the transmission loss of a passive structure based on a polymer material is reduced to 14.4-15.5 Np/m. Compared with the transmission line based on traditional quartz dielectric substrate and semiconductor dielectric substrate, the transmission loss is reduced by more than half. At the same time, in order to reduce high-frequency losses and improve circuit efficiency, the diode needs to use a sub-micron structure with a junction radius of 0.5 μm and a junction capacitance of 1.5 fF. The simultaneous optimization of the passive structure and the active structure makes the frequency multiplier operating in the 540-580 GHz band, with the conversion loss better than -8 dB, and the return loss better than 10 dB. And because the passive part adopts an innovative hybrid transmission structure, the overall physical size of the mixer circuit is increased under the condition of ensuring single-mode transmission, and the process difficulty of cavity processing is reduced, which makes the low loss plane transmission of THz signal above 1THz becomes possible in the future.

As the most important mid-infrared source at present, Quantum Cascade Lasers(QCL) has become one of the research hotspots due to its unique performance and frequency that can be extended to terahertz(THz). For QCL, there are many factors affecting their output power and operating temperature, and efficient heat dissipation is an important factor. In this paper, the thermal management research of two kinds of QCL including mid-infrared and terahertz is summarized. Secondly, the similarities and differences between the two are discussed and analyzed. Two methods of heat dissipation on the solid side of lasers are mainly discussed, including active areas design, processes improvement, and the optimization of device material systems. Finally, the future research trends of thermal management of QCL are analyzed and predicted. This result has certain reference significance for the performance improvement of QCL, especially the improvement of output power and working temperature.

Along with the development of terahertz(0.1-10 THz) spectroscopy, terahertz imaging has been applied in biomedical area, especially in skin burn detection. But how to transfer this technology from laboratory research to clinic application is still a great challenge. Terahertz technology has been studied in the area of skin burn severity assessment, including imaging system, ex-vivo experiments and in vivo experiments, and clear terahertz images can be obtained. In this paper, skin burn severity classification and diagnosis methods are reviewed, then focusing on the imaging system, detection results and burn severity assessment methods, the progress of terahertz detection of skin burn are introduced, and further explanation of its evaluation principle and shortages are analyzed. The status and problems of terahertz imaging, clinic application challenges are discussed in detail. Finally, the development tendency and problems to be solved of terahertz imaging on skin burn towards clinical application are proposed.

When the relative velocity between radar and targets is really high, the scale modulation effect of radar echoes is not negligible. The pulse compression of received Linear Frequency Modulated (LFM) echoes is degraded significantly, which causes defocused image of radar systems. Hyperbolic Frequency Modulated(HFM) waveform is of the characteristic of Doppler invariance, and it has a high application value in the imaging of high speed targets. First, the received echoes of high speed moving targets are modeled and the imaging results are compared between LFM and HFM by instantaneous frequency analysis. Then, the system implementation, imaging characteristics and application superiority of HFM echoes from high speed targets are analyzed in detail, and compared with that of LFM method. Finally, numerical results validate the effectiveness of the proposed method.

Aiming at the multi-objective pedestrian trajectory tracking problem in public space, a multi-objective pedestrian trajectory tracking algorithm based on reinforcement learning is proposed. Firstly, a high-precision target detector is utilized to detect pedestrian targets in space, and an independent single-target tracker is assigned for each tracking target trajectory. Each target is trained as an agent by means of deep reinforcement learning, and combined with the appearance and position characteristics between tracking trajectory and detecting target, the tracking target is constructed. Similarity cost matrix is built to realize data association through Hungarian algorithm. Experiments show that the tracking accuracy of this algorithm is 76.1% on common open data sets. Good results have been achieved in multi-objective pedestrian trajectory tracking in public space.

Data collection is the fundamental purpose of deploying Wireless Sensor Networks(WSNs). The adoption of mobile accommodation strategy can effectively alleviate the problem of energy consumption of WSNs. The mobile path is the key of the strategy. Therefore, Virtual Rendezvous points -based Data Collecting(VRDC) algorithm is proposed in this paper. The VRDC algorithm first plans lodging paths based on the host Points, and then selects Virtual Rendezvous Points(VRPs) based on the host Points. VRPs can transmit data directly to the mobile address via one hop, while other nodes can transmit data to the nearest VRPs, thereby reducing the number of transmission hops and energy consumption. Simulation results show that VRDC algorithm can effectively reduce energy consumption and balance energy consumption among nodes.

A high speed data acquisition and processing system based on clock network is developed aiming at the requirement of high speed data acquisition system in full waveform laser radars. The key techniques are studied in detail. Based on analyzing ChipSync technology and clock jitter, a high quality clock network management method based on PLL and clock buffer is proposed. By using the proposed method which is based on the active intervention of high speed ADC output on-line clock, the problem of multi-channel high speed data flip-latch is solved. Experiment results demonstrate that the realized system can reach the sampling rate of 1.2 GSPS and the Effective Number Of Bit(ENOB) above 8 bit.

Automatic modulation classifications would play an essential part in wireless spectrum anomaly detection and radio environment awareness. With the breakthrough in deep learning algorithms, this issue can be solved with unprecedented precision and effectiveness by using neural networks. Therefore, a novel neural network termed as Long short-term Convolutional Deep Neural Network(LCDNN) is proposed, which creatively combines the complimentary merits of Long Short-Term Memory(LSTM), Convolutional Neural Network(CNN) and deep network architectures. This model directly learns from raw time domain amplitude and phase samples in training dataset without requiring human engineered features. Simulation results show that the proposed model yields a classification accuracy of 93.5% at high SNRs. Further, the noise sensitivity of the proposed LCDNN model is examined and it is showed that LCDNN can outperform existing baseline models across a range of SNRs. Finally, in order to reduce the computational complexity of the LCDNN model, a ‘compact’ LCDNN model is proposed, which achieves the state-of-the-art classification performance with only 0.6% parameters of the original LCDNN model.

In millimeter-wave multiple-input-multiple-output(MIMO) array imaging, image reconstruction relies on point-by-point back propagat ion algorithm due to the irregular antenna arrangement. As a result, the reconstruction speed cannot meet the need of fast imaging. In order to solve this problem, a hierarchical “Coarse-To-Fine” reconstruction strategy is proposed. The fast coarse reconstruction based on phase compensation and the fine reconstruction based on back propagation are combined to process the MIMO holographic data. The millimeter wave imaging experiments by using orthogonal arrays and compact square arrays are simulated by Feldberechnung bei Korpern mit beliebiger Oberflache(FEKO), verifying the effectiveness of the proposed hierarchical reconstruction method.

A fast prediction method based on parabolic equation is presented for the timeliness of the radio wave propagation in large scale partly forested irregular terrain environment. The wide angle Parabolic Equation(PE) method is used, which is solved by Split-Step Fourier Transform(SSFT). In the SSFT step iteration process, the level step is determined dynamically, according to the equivalent dielectric permittivity of forest and irregular terrain. The feasibility and validity of the fast prediction method is verified by simulating the radio propagation characteristic in the partly forested irregular terrain. The results show that the method has higher accuracy compared with the uniform coarse step, and could improve the computational efficiency greatly compared with fine accuracy.

Based on the demand for antenna with wide-bandwidth and wide coverage angle, a miniaturized antipodal Vivaldi antenna is designed for 9 GHz to 11 GHz. A rectangle director is added to the back end of the antenna for better broadband matching effect. The HFSS software is used for structural optimization to obtain the specific parameters of the antenna. The antipodal Vivaldi antenna with the director is processed and tested. Through simulation and actual measurement, the return loss is less than -15 dB in the 9-12 GHz band, and the bandwidth below -10 dB is 6.9-14 GHz. At the same time, 3 dB beam width enables ±30° coverage range. The size of the antenna is only 15 mm×16 mm×0.762 mm. The proposed antenna is a miniaturized one with a wide bandwidth and wide coverage angle.

A novel Tightly Coupled Array(TCA) is proposed in this paper. The array element is made of bow-tie dipole, and the array realizes ultrawideband by increasing the coupling between the units. By loading Resistive Frequency Selective Surface(RFSS) between array and ground plane, the antenna short-circuit point can be suppressed, and the impedance matching bandwidth can be increased significantly. In addition, the parasitic layer is designed to replace traditional dielectric-slab wide-angle matching layer, which makes the height and weight of the antenna are greatly reduced and the scanning characteristics of the array are improved. The simulation results show that the antenna achieves 1.5-14 GHz bandwidth for Voltage Standing Wave Ratio(VSWR)<2, and realizes 45°scanning range at E-plane and H-plane.

In order to increase the bandwidth of the horizontal omnidirectional antenna, a new type of wide-band horizontal polarization omnidirectional antenna is designed. The antenna consists of an array of four arc-shaped dipole antennas forming a ring, and each dipole antenna is printed on the upper and lower surfaces of the Printed Circuit Board(PCB). The radiation fields of the four arc-shaped dipole antennas add together and compose a horizontal omnidirectional pattern in the far field. By slitting each dipole element and loading parasitic patches on the edge of the printed board, the bandwidth of the antenna is increased. An antenna sample is designed and manufactured. The measured results show that the antenna can achieve |S11|<10 dB in 2-3.71 GHz, the relative bandwidth is 60%, the mean gain of its horizontal omnidirectional is about 1.36 dBi, and the non-roundness of its horizontal omnidirectional pattern is less than ±1.5 dB. The test results are in good agreement with the simulation results.

A multi-beam Luneburg lens antenna with lobe equalization and operating in the X-band is designed and fabricated by 3D printing technology. The required gradient relative dielectric constant is implemented by changing the dielectric filling rate of each unit of the lens. A set of dual-mode conical horn antennas is used as the feed to generate multiple beams with lobe equalization characteristics. Experiments show that the antenna can generate multiple beams in three-dimensional space, covering the range of ±20° in altitude and azimuth, and the characteristics of each beam stay identical. The maximum gain of the antenna is 24 dBi. The 3 dB beam width is 8°. It has high gain and good focusing characteristics. Because of the antenna’s simple structure, short turn-around fabrication time and low fabrication cost, together with its good operational characteristics, it has good application prospects in the field of radar, detection, and sensing.

The purpose of this paper is to design a new type of decoupling structure for Ka band and to analyze the decoupling effect of this structure. The equivalent circuit model of the “d-shaped” Uniplanar Compact Electromagnetic Band Gap(UC-EBG) structure is established, and its band-stop filter characteristics are obtained through simplified analysis. The distribution parameters of the decoupling structure are calculated based on the conformal transformation and elliptic integral calculation. The simulation results of the equivalent circuit are consistent with the simulation of the decoupling structure. Applying this decoupling structure to the array antenna improves the isolation between array elements and reduces the mutual coupling effect of array elements.

Theoretically, the magnetic coupling resonance wireless power transmission system can be designed to achieve high efficiency within the critical coupling zone. However, the system efficiency will decrease sharply when the distance is in the weak coupling zone and the resonance frequency will split when the distance is in the strong coupling zone. Therefore, a planar spiral coupling coil with ferrite core is designed and fabricated. Then the frequency tuning system is designed to keep the high transmission efficiency even when the transmission distance changes in the strong coupling zone. Measurement results demonstrate, when the transmission distance changes between 5-20 cm, the designed wireless power transmission system can guarantee an efficiency above 80%.

The microwave dissipation mechanism of powder materials is different from that of bulk materials. The mixed powder of non-magnetic metal and oxide show strong absorption on microwave magnetic field. Researches which focus on the electrical conductivity and microwave magnetic absorption characteristics of the mixed powders are carried out. Samples of the mixed powders are loaded to the isolating bandpass filter which is achieved with magnetostatic surface wave, in order to study their effects on the isolation and insertion loss of non-reciprocal filter. The experimental results show that: the 30%Cu+70%CuO mixed powder sample with conductivity of 217.435 mS/m has strong absorption on the microwave magnetic field; after loading this mixed powder sample to the non-reciprocal filter, the insertion loss is reduced and the isolation is significantly increased to 21.56 dB.

Large pose face alignment is a prerequisite for many important visual tasks such as face recognition and 3D face reconstruction. However, most of the existing alignment methods use two-dimensional boundary markers to align, and the number of boundary markers used is limited, which greatly affects the accuracy of large pose face alignment. Therefore, an improved large pose face alignment method is proposed. Firstly, 3D deformable model is utilized to represent 2D face image. And the problem of face alignment with arbitrary pose is modeled as a fitting problem based on Three Dimensional Deformation Model(3DMM). And then a cascade regression method based on Convolutional Neural Network(CNN) is adopted to learn the mapping relationship between two-dimensional face image and its representation. Finally, two new pose invariant local features are proposed as the input layer of CNN learning, and CNN is applied for large pose face alignment through training. Simulation results on two classic face image data sets show that the proposed method is better than the latest face alignment method.

Recently, researches on resource-saving mixed-radix Fast Fourier Transform(FFT) technology with low complexity and high efficiency are of vital importance in digital signal processing. In this paper, a new mixed-radix FFT decomposition algorithm based on Field Programmable Gate Array(FPGA) is proposed and implemented. The proposed in-place algorithm adopts a hybrid decomposition mode combining prime factorization algorithm and Cooley-Tukey algorithm, which can save one-step multiplication operation of the rotation factor and also reduce the storage space and operation amount effectively, while it also uses the universal butterfly unit module to accommodate to the radix-2, radix-3, and radix-4 FFT operations. The simulation results indicate that the proposed algorithm can greatly improve the flexibility of FFT processing points and effectively save computing resources.

A Density Peaks Clustering algorithm based on optimized AdaBoost-DPC is proposed to deal with the class-imbalanced question of target and clutter in radar data sets. The method of density peaks clustering is introduced, and the clutter dots are under-sampled. The error function of AdaBoost algorithm is improved based on the asymmetric misclustering cost, which raises the weight of positive misclassification cost. Then the improved AdaBoost algorithm is combined with density peaks clustering method to cluster the imbalanced radar data sets consisting of the targets and clutter dots. The experimental results show that the optimized method can effectively improve the identification of target.

Fast retrieval on large scale human face data sets is a key problem in face recognition applications. The face Hashing method with short length can reduce the computational amount of face feature alignment, and is helpful to the application of large-scale face recognition. In this paper, a deep face Hashing method based on the loss function of ternary-group is presented. By optimizing the loss function of ternary-group, the deep convolution neural network is trained to extract the deep feature of images. The distance between the similar images can be as small as possible, and that between different kinds of images is as large as possible. The high dimension feature is mapped to the low dimension space by adding the random mapping layer following the deep network. And the low dimension space is further mapped to Hamming space by the threshold algorithm. Experimental results on multiple standard datasets show that the proposed method outperforms other state-of-the-art methods.

The inundation of recommendation information has seriously affected the user experience of mobile intelligent terminal, while the limited computing environment restricts the research on the filtering model and algorithm of mobile recommendation information. A mobile recommendation information filtering algorithm based on limited computing environment is proposed by combining user portrait technology and information filtering method. The algorithm constructs user profile through user trajectory information, filters all kinds of information recommended by the outside world according to user profile and demand characteristics, and finally recommends information the second time. The simulation results show that the algorithm has high filtering accuracy and strong personalized service ability. It needn’t occupy too much system computing and storage resources, and has high cost performance. It is suitable for the limited computing environment of mobile intelligent terminal, and has certain practicability.

For Permanent Magnet Synchronous Motors(PMSM) in a time-varying disturbance environment, the sliding mode control with better robustness is adopted, an improved reaching law is proposed based on the traditional exponential reaching law in sliding mode control. Based on the improved reaching law, a sliding mode speed controller of PMSM speed regulation system is designed to improve the running performance of the motor. Through comparative analysis of simulation results, the sliding-mode speed controller based on the improved reaching law can not only improve the dynamic performance of the system, but also weaken the chattering phenomenon of the system, and it can make the motor still have high running performance under time-varying disturbance.

In order to ensure the security of sensor access to the information intranet and the confidentiality, integrity and availability of data in the process of transmission, storage and use, this paper proposes a multi parameter collaborative authentication scheme with joint consideration of the equipment performance and digital signature. Then a new identity based key establishment protocol is proposed by using elliptic curve cryptosystem. Through security performance analysis, compared with the existing classical schemes, the proposed scheme is more secure with less computing time consumption while facing the imitation attack, MITM(Man-In-The-Middle) attack and desynchronization attack.

Vacuum arc ion source is suitable for miniaturized pulse neutron generator because of its simple structure and high deuterium ion density. A small vacuum arc ion source is developed with the diameter of 20 mm and the length of 25 mm. TiD works both as the cathode and as deuterium source, which avoids using the complicated gas pipes. The light spots caused by the ion source are measured by a Charge Coupled Device(CCD) camera. It is found that the larger the arc current, the brighter the spots. The saturated ion current is measured by a biased plate. When the arc current is larger than 100 A, the saturated ion current could be above 1 A. The deuterium ion ratio in the plasma is measured by a simple magnetic analysis. The ratios grow with the arc current, and the maximum is about 27%. When the arc current is 100 A and the pulse duration is 5 μs, the neutron yield is about 1×105 n for the TiD target under 120 kV voltage. The source can be used in small pulsed neutron generators suitably.

A new input interface circuit with single event effect hardened is proposed, which adopts the scheme of combinational logic operating after delay-processing. This circuit is designed and implemented in 0.8 μm 600 V Bipolar-CMOS-DMOS(BCD) process of Communication Mode Selector Control(CMSC). The single event radiation experiment has been completed in National Space Science Center. The results of simulation and test show that the circuit proposed in this paper can effectively immune to Single Event Upset(SEU) whose Linear Energy Transfer(LET) is under 80 MeV·cm2/mg. Especially for the case that multiple nodes occurring upset at the same time, the proposed circuit has high capacity of SEU–tolerant.

The diagnostic unit of pulsed xenon lamp transient discharge current is an important part of the energy system in an Inertial Confinement Fusion(ICF) laser device. Based on the analysis of problems such as strong electromagnetic interference, noise coupling, and topological constraints of multiple parallel acquisition circuits, a design method of multi-level isolation and Field Programmable Gate Array(FPGA)-based multi-channel synchronous parallel architecture is proposed and the prototype is developed. Experiments show that the current measurement range of the diagnostic unit is 0-25 kA, the sampling rate is 200 kHz, and the measurement accuracy is better than ±0.5%. This design meets the requirements of synchronous diagnosis of the discharge current of 20 pulsed xenon lamps under 31 kV of working point voltage. Compared with traditional design, this method has stronger electromagnetic compatibility and higher integration of parallel acquisition design.

In order to meet the requirements of Single Event Effect(SEE) evolution for a wide variety of integrated circuits with complex functions and overcome the limitations such as time constraints and limited physical space in current domestic ground single event irradiation experimental environment, an efficient and universal single event effect test system is designed and implemented for integrated circuits, which innovatively adopts rotating solid verticalstructure, including a multi-Field Programmable Gate Array(FPGA) electrical test platform, a motion control subsystem and some loading boards of the Devices Under Test(DUT). The entire test system is packaged as a portable box, which only needs three DB9 interfaces to complete all connections with the outside world. Therein the PC interface is achieved based on LabVIEW, friendly to users. The lower computer test program is implemented based on the multi-FPGA platform, which is flexible, extensible and versatile. It can realize one-time installation, automatic switching and angle radiation test of 8 types of integrated circuits within 300 PINs, meanwhile monitor in real-time and background record the detail information such as flip data, flip time, circuit status and so on, whose test frequency achieves 100 MHz. The reliability, efficiency, stability, high integration, and convenient installation and debugging have been verified by multiple tests on Application Specific Integrated Circuit(ASIC), Static Random-Access Memory(SRAM), Controller Area Network(CAN) interface circuit and other integrated circuits.

The laser simulation method is adopted to study the photocurrent characteristics of N-Metal-Oxide-Semiconductor(NMOS) devices under gamma ray irradiation. The laser experiments are carried out, and the relationship between photocurrent and laser incident energy is obtained for Partial-Depleted Silicon On Insulator(PDSOI) NMOS devices with different sizes and different structures. The photocurrent characteristics of devices are simulated by using TCAD simulation tool. By comparing TCAD simulation and laser experiment data, the results of the two groups of data are basically consistent, which verifies the feasibility and accuracy of the laser simulation technology. The relationship of photocurrent between theoretical calculations and laser experiments is obtained, based on which, the magnification factor of the parasitic bipolar transistor is calculated under the laser condition.