Terahertz(THz) waves have a wide range of applications in the field of sensing because of their good coherence, high signal-noise ratio, and low radiation energy. In addition, THz sensing also has the advantages of real-time, non-contact, label-free and non-ionization, so it has important applications in biochemical sensing, especially the sensing of biologically active substances. Nevertheless, THz sensing also has disadvantages such as low sensitivity, strong water absorption, limited detection information, and poor applicability. This review introduces the series of work of our research group in THz time-domain polarization spectroscopy sensing technology, which takes microstructure devices as sensors and uses transmission or reflection sensing methods to detect several biochemical samples of cells, amino acids and Deoxyribo Nucleic Acid(DNA), respectively. The experimental results show that: compared with the traditional resonance sensing method, the Q-factor and sensing sensitivity of the polarization sensing method are significantly improved; the reflective sensing effectively avoids the absorption of THz by water, and realizes the sensing of active biochemical samples in liquid environment; taking chiral microstructure devices as sensors, or taking chiral THz waves as excitation fields, the polarization response of the sample can be enhanced as well as the sensing sensitivity, and the sensing of chiral molecular is realized.

With the rapid development of THz technology, the waveguide attenuator used in THz system has also become a hotspot. The THz waveguide attenuator can accurately attenuate and control the power transmission, which shows special significance and plays an irreplaceable role in solving the problems of loss, radiation and interference. The current waveguide attenuator would destroy the wave guide transmission line and result in the RF leakage because it places the attenuation plate paralleling to the electric field. Based on the principle of the absorbent waveguide attenuator, a waveguide attenuator with the attenuation plate perpendicular to the electric field is proposed. The attenuation plate is attached to the inner wall of the waveguide. The waveguide attenuator is realized and optimized by using the HFSS software and changing the shape, position and other parameters of the attenuation plate. The return loss is less than -27.5 dB in 110-170 GHz, and the attenuator accuracy is less than ±2 dB in the attenuation of 20 dB.

Photoconductive antenna has the advantages of room temperature operation, compact design and wide-band radiation, but its wide application is limited by low radiation power caused primarily by low optical absorptivity. The traditional antenna electrode has no tip structure and the edge electric field is weak, which leads to low light absorption. In order to improve the radiation power of photoconductive antenna, a triangular array antenna electrode structure is designed. The electrode structure is composed of five triangular tips. Finite Difference Time Domain(FDTD) method is adopted to study the electric field enhancement of the electrode and the light absorption of the substrate under 800 nm femtosecond laser irradiation. The results show that this structure increases the area of laser incidence on the substrate and reduces the optical carrier transmission distance. In the case of no electric field, the optical absorption reaches 30.57%, which is 161% higher than that of the traditional antenna. Triangular array electrode structure provides a new idea for the design of traditional electrode structure, which is expected to be combined with nanostructures to further improve the radiation power.

A tunable broadband terahertz absorber based on graphene metamaterial is proposed and numerically demonstrated. The absorber consists of three layers: the upper is the graphene metamaterial layer, the middle is the SiO2 layer, and the bottom is the metallic layer. Simulation results demonstrate that the proposed absorber achieves over 90% absorption in 1.11- 2.61?THz with a relative bandwidth of 80.6% when Fermi level ?c=0.7?eV. The peak absorption rate of the proposed absorber can be tuned from 20.32% to 98.56% by changing the Fermi energy of graphene from 0?eV to 0.7?eV. Additionally, the proposed absorber is insensitive to polarization and has high absorbance to wide incidence angles. Such design may have some potential applications in thermal imaging, thermal detecting, and stealth technique.

A -16 dB waveguide coupler for the Local Oscillator(LO) signal in 400-500 GHz band is developed in order to meet the application requirements of terahertz sideband-separating receivers, mainly considering the weak coupling of LO signal, the high directivity of Radio Frequency(RF) signal and current milling process accuracy. The characteristic of the coupling degree of the branch coupler is analyzed. The design of a -16 dB waveguide coupler is conducted. The device is processed based on Computer Numerical Control(CNC) technology and the result is discussed. The measured results of both samples show that this coupler can work in 400-500 GHz band (factional bandwidth of 22.5%), obtaining the coupling coefficient for the LO signal in -16~-17 dB, the directivity for RF signal of -1.2 dB, the isolation degree better than -20 dB, and the return loss of each port better than -15 dB. The performance is highly consistent with that of the simulations, which indicates that current CNC technology can meet the high precision requirements for the fabrication of such high frequency waveguide coupler.

The Likelihood Ascend Search(LAS) algorithm is a heuristic neighborhood search algorithm that detects the received signals of large–scale Multiple–Input–Multiple–Output(MIMO) systems with space–division multiplexing. A Constellation Constraint–LAS(CC–LAS) is proposed for reducing the computational complexity of the traditional LAS algorithm. The algorithm first introduces a novel CC structure to determine the reliability of each candidate solution. Then, according to the reliability determination result, the neighborhood space of the candidate solution is narrowed. Finally, the unreliable candidate solution is detected by using the LAS algorithm. The proposed CC–LAS algorithm eliminates the inefficient processing of low–reliability signals by ignoring a large number of unnecessary neighbor vectors in the LAS neighborhood space. Hence, CC–LAS algorithm is capable of greatly reducing the computational complexity of the traditional LAS algorithm. The simulation results show that the BER performance of the proposed CC–LAS algorithm is very close to that of the traditional LAS algorithm; nevertheless, the computational complexity can be greatly reduced under the same Signal–to–Noise Ratio(SNR) compared to traditional LAS algorithm.

A joint estimation algorithm based on grid search is proposed in order to realize the joint estimation of synchronous damage and channel in Multiple Input Multiple Output(MIMO)-Orthogonal Frequency Division Multiplexing(OFDM) system. Firstly, by constructing a system model to reflect the influence of synchronous damage and channel response, the multi-dimensional optimization problem of each damage parameter estimation is then simplified into two-dimensional grid and one-dimensional grid search, so as to realize the joint estimation of carrier frequency offset, sampling frequency offset and symbol timing error. The numerical simulation results show that the proposed joint estimation algorithm has better estimation performance than the non-joint estimation algorithm.

The phase difference measurement accuracy for the channels receiving signals in the interferometer is discussed. After the limitations of the traditional formula are pointed out, a theoretical computation formula of the standard deviation for the phase difference measurement error of the channels receiving signals in the interferometer is proposed by using the signal energy SNR (Signal-to-Noise Ratio) instead of the in-band power SNR. In theory, the universality of the formula is proved from the point of view of demodulation and match filtering. And the correctness of the theoretical analysis is verified by the simulation for various modulated signals and pulse series signals. It reflects that the standard deviation of the phase difference measurement error for the channels receiving signals in the interferometer is inversely proportional to the square root of the signal energy SNR. This theoretical result provides a general guidance for the computation, demonstration and analysis of direction finding accuracy index in interferometer engineering application.

A robust beamforming algorithm based on covariance matrix reconstruction is proposed to solve the sensitivity problem of Capon beamforming under error condition. The algorithm divides the spatial domain of the signal set into interference region and signal region, and then divides the two regions into several independent and non-overlapping parts. The interference covariance matrix is constructed by integrating the interference region, and then the noise covariance matrix is reconstructed by using the minimum eigenvalue of the sample covariance matrix. Finally, the expected signal guidance vector error is modeled by the ring uncertainty set, and the Capon spectrum integration is performed on the ring uncertainty set to estimate the expected signal covariance matrix, and the expected signal guidance vector is obtained according to its main eigenvector. Simulation results show that compared with the traditional robust beamforming algorithm, the performance of this method is more excellent and stable under the conditions of different snapshot numbers and input Signal-to-Noise Ratio(SNR). At the same time, it bears the advantage of low computation.

Non-intrusive eye tracking plays an important role in many vision-based human computer interaction applications. How to ensure the robustness of external interference and tracking precision during eye tracking is the key problem to its applications owing to the strong nonlinearity of eye motion. To improve the robustness and precision of eye tracking, the Strong Tracking fifth-degree Cubature Kalman Filter(ST-5thCKF) algorithm is proposed. The algorithm introduces the suboptimal fading factor of Strong Tracking Filter(STF) into fifth-degree Cubature Kalman Filter(5thCKF) which almost has the least cubature sampling points while maintaining the fifth-degree filtering accuracy. The proposed algorithm bears the high filtering precision to strong nonlinearity of 5thCKF, as well as the robustness to external interference of STF. The experimental results under practical conditions show the validity of the proposed algorithm in eye tracking.

In order to enable secondary or unlicensed users to obtain idle subbands on a given broadband for use, the broadband spectrum sensing technology in cognitive radio is discussed and an effective broadband spectrum sensing algorithm is proposed. The algorithm first uses Hidden Markov Model(HMM) to model the dynamic behavior of primary users to overcome the limitations of current broadband sensing technologies. Secondly, the proposed algorithm uses the existing narrowband sensing technology to divide the sensing spectrum band into smaller channels and model it as a balanced binary tree; then the spectrum holes are recursively searched. If any holes are detected to be adjacent in frequency, they are merged into a single spectrum hole for maximizing the capacity of cognitive secondary users over the entire frequency band. The simulation results show that compared with the existing broadband spectrum sensing methods, the proposed broadband spectrum sensing algorithm has better sensing performance gain and stronger robustness.

Aiming at the issue of localization for mobile nodes in hybrid Line-Of-Sight/Non-Line-Of-Sight(LOS/NLOS) environments, a large-scale network node localization scheme based on Time-of-Arrival(ToA) measurements and error mitigation is proposed. Firstly, nodes equipped with Ultra-Wide Band(UWB) radio move randomly for collecting ToA measurements, and the shortest path selection algorithm is conducted to obtain the multi-hop node distance including the one-hop node distance with NLOS error mitigation. Then, the initial node positions are determined by using Multi-Dimensional Scaling(MDS). Finally, unlocalized node positions are obtained by the combination of iterative trilateration and error accumulation management. The simulation results show that the location accuracy of the proposed location scheme is superior to that of other common schemes.

Near–field imaging is an important application area of interference passive millimeter-wave imaging technology, and the array configuration is an important factor affecting the near–field phase error. In this paper, a 2–D synthetic aperture near–field imaging simulation system is built to implement functions such as target scene generation, near/far field forward simulation, image reconstruction, and near–field phase error correction. This system is utilized to quantitatively evaluate and analyze the near–field errors of different 2–D antenna arrays with the same spatial resolution. For the two–unit near–field scanning imaging test system, a self–calibration method is proposed to correct the receiver channel error and near–field phase error respectively. Compared with the near–field imaging method based on a reference point source, this method requires only a priori distance information and no need to image the reference point source, which bears the advantages of simple operation and fast imaging speed.

Traditional multi-band low-profile parabolic antennas bear the disadvantages of complex structure and the inability to multi-band work simultaneously. A novel design method is introduced in this paper. The proposed quad-band antenna works on the frequency of 14-14.5 GHz, 11.45-12.75 GHz, 19.6- 21.2 GHz, 29.4-31 GHz. The whole antenna adopts a two-slot-depths horn feed, shaped sub-reflector and main reflector. The reflectors are designed by the aperture field distribution function and polynomial transition function. The overall antenna structure is optimized with full-wave electromagnetic software. The test result shows the efficiency in the entire frequency band is improved by at least 12%, and the first side lobe is controlled below -14 dB, which meets the performance requirements. This design has been applied to actual equipment successfully.

A miniaturized tri-band Long Term Evolution(LTE) microstrip antenna based on the Composite Right/Left-Handed Transmission Line(CRLH) theory is studied. The microstrip antenna is miniaturized by using the zero-order resonance characteristic of CRLH. The simulation results show that the three operating frequency bands of the microstrip antenna are: 1?840-1?860?MHz, 2?130-2?160?MHz, 2?620-2?660?MHz. The Band3, Band4, and Band7 of the Frequency Division Duplexing(FDD) LTE plan are accurately covered. The electrical length is 0.17λ. Finally, the microstrip antenna is processed and measured to verify the accuracy and effectiveness of the simulation. The multi-frequency miniaturized antenna can not only save the antenna feed resources of the base station greatly, but also cut the cost of base station construction effectively.

The shipborne missile is considered as an on-board device when it is on duty and an aircraft when it is in flight. GJB151B-2013 specifies that the shipborne platform equipment needs to be tested by the CS106 project. Nevertheless, the missile is not directly connected to the ship's power grid. If the assessment is still based on the spike signal index specified in GJB151B-2013, it is obvious that there exists a case of over testing. In view of the tailoring requirements of the shipborne missiles in the electromagnetic compatibility CS106 index, the evaluation requirements are fully analyzed and studied, and experimental cutting recommendations are derived.

A method of rapidly building radio environment map based on three-dimensional parabolic equation is studied, and the accuracy and applicability of several spatial interpolation methods are discussed, which provides an effective tool for network optimization and management. Firstly, the electric field distribution in large area and complex environment is calculated by using parabolic equation. Then, four typical research areas are selected with reference to the best statistical unit in geography, and Kriging, Inverse Distance Weighting(IDW), biharmonic spline interpolation and several interpolation methods based on triangulation are adopted to analyze the research area. It shows that several kinds of triangulation interpolation and biharmonic spline interpolation have the smallest RMSE and the shortest time in each region, but the degree of spatial fitting is the lowest; Kriging has the highest spatial fitting degree in each region, and the RMSE and time vary with the semi-variance model; IDW has a good spatial fitting and RMSE in each region, but it takes the longest time. Among several spatial interpolation methods, the Kriging of the exponential model can construct the radio environment map based on the three-dimensional parabolic equation with the highest accuracy and applicability.

With the increasing density of 5G base station in recent years, the influence of the electromagnetic radiation level of 5G base station has become one of the most important factors restricting the development of 5G communication technology. In order to solve the problem of electromagnetic radiation limits around 5G base station, an evaluation method is proposed to predict the electromagnetic radiation level of 5G base station. Assuming that the user capacity of the cell covered by the base station is known, a statistical model is adopted to calculate the maximum power density around the base station. On ANSYS Savant, the accuracy and feasibility of the predictive method are verified by simulating the maximum power density around the 5G base station. This method has certain guiding significance for future location of base station.

In Risk Priority Number(RPN) analysis based on Multi-Criteria Decision Making(MCDM), it is critically important to determine the weight of each expert in Failure Mode, Effects and Criticality Analysis(FMECA) of a product. For a widely used type of experts’ opinions, interval number information, subjective weights are determined by the index system of employment years, working experience and familiarity degree. Then based on the similarities of each two experts’ opinions, the method to calculate the objective weight is conducted. The multi-sources experts’ opinions are fused by an integrated weighting on the basis of above subjective and objective weights. At last, the effectiveness of proposed method is validated by a criticality analysis of the intrinsically safe power supply applied in mine.

Manual screening of trigeminal nerves requires high professional quality and is time consuming for clinicians. Using deep learning to automatically detect trigeminal nerve regions in cranial Magnetic Resonance Imaging (MRI) can provide a reliable input image for subsequent trigeminal nerve segmentation. YOLO(You Only Look Once) network is utilized to automatically detect the trigeminal nerve region of the cranial magnetic resonance image to improve the inference speed, and to systematically evaluate the inference performance of the NVIDIA TensorRT framework under different computing platforms. The experimental results show that the YOLO target detection network can accurately detect the area where the trigeminal nerve is located. Simultaneously, under the NVIDIA TensorRT framework, when the input brain MRI resolution is (204×204), the YOLOv2 network detects the optimized trigeminal nerve through the CPU platform, embedded GPU platform, desktop GPU platform and professional GPU computing card platform, the frame rates per second can reach 0.1 FPS, 23.4 FPS, and 793.7 FPS . This provides important reference for the subsequent development of portable trigeminal neural segmentation equipment.

The severe overlapping of particle tracks seriously affects the measurement accuracy of particle track. A new method is proposed based on track overlapping classification to separate and measure the severe overlapping of particle tracks. Firstly, the overlapped tracks are classified according to the inclusion of particle track centers to identify the severity of particle overlapping. Then, the non-overlapping particles are scaled to half to form structural units. The finite erosion operations are applied to the severely overlapped particle tracks based on these structural elements. These particle tracks are separated, and form four kinds of track classification images. Experimental results show that these severely overlapped particle tracks can be separated by combining track overlapping classification and finite erosion operations, which can eliminate the influence of track overlapping on the image measurement accuracy.

Effective compression of hyperspectral images is of great significance for real-time transmission. In this paper, spectral linear decomposition is introduced into efficient compression of hyperspectral image. According to the Linear Mixed Model(LMM), the hyperspectral data is decomposed into the product of endmember and abundance. At the encoder, the necessary data processing is performed on the endmembers and abundance, followed by JPEG-LS lossless compression. At the decoder, the original hyperspectral image is reconstructed by multiplying the final decoded endmembers and abundance, moreover, the effect of the quantization step on the rate-distortion performance is discussed. Experimental results show that the proposed method can achieve certain compression performance.

Aiming at the complicated iterative operations of traditional Compressed Sensing(CS), long reconstruction time and poor quality, a compressed sensing reconstruction algorithm for Non-linear Measurement Convolutional Neural Network(NMECNN) is proposed by combining the deep learning method. This algorithm compresses the overall width and height of the image as a measurement network to replace the traditional random measurement matrix for image reconstruction. At the same time, it uses multiple expanded convolutional layers and upsampling PixelShuffle methods to obtain detailed information of different scales of the image. Through experimental comparison with other documents, the average Peak Signal to Noise Ratio(PSNR) values of this algorithm at different sampling rates are higher than that of MSRNets algorithm by 1 dB, 0.7 dB, 0.82 dB, 1.61 dB, and the Structural Similarity(SSIM) values are higher by 0.03, 0.04, 0.24, 0.10 units. The reconstruction time in the CPU is less than that of the MSRNet algorithm by 0.175 5 s, 0.399 8 s, 0.41 s, 0.396 s, respectively. Through big data sets and noise experiments, it is verified that the image reconstruction quality of this algorithm is significantly improved, the reconstruction time is greatly shortened, and it has a strong ability to resist noise attacks.

Short distance acoustic communication technology under the oil well is the key to transmitting the borehole information measured by the near-bit sensor to the Measurement While Drilling(MWD). In order to explore the transmission characteristics of the sonic carrier signal in the short-distance drill pipe, the finite element method is adopted to establish a 1.28?m long drill pipe model, and an excitation load with a frequency of 20?Hz and a duty ratio of 50% is applied to the drill pipe to simulate the calculation of the radial receiving point response of the rod and the characteristic frequency of the drill rod structure. The transmission characteristics of the acoustic wave in the short distance drill rod are analyzed as well. The results show that when the drill pipe is subjected to an excitation load under the initial conditions, the acoustic signal at the receiving point of the drill pipe has an enveloping curve that decays exponentially, and its interior is a cosine periodic signal with gradually decreasing amplitude, and finally stable at a new equilibrium position; when the drill pipe continues to be loaded, the acoustic signal will continue to decay on the basis of the previous steady state, and eventually decays and stabilizes in the initial state. The simulation results are in good agreement with theoretical analysis and experimental results. This work can provide basic model for the simulation of sound wave transmission in multiple drill pipes.

Aiming at the problems of complex distribution and the lack of comprehensive and effective pre-warning methods for power cables, a cable temperature pre-warning system based on the Gray Markov model is proposed. Firstly, the overall structure of the temperature pre-warning system and the data communication method are analyzed. The communication method combining Long Range(LoRa) and General Packet Radio Service(GPRS) is utilized to ensure the reliability of data acquisition. Secondly, the gray Markov model is adopted to predict the cable temperature. The prediction result is collected for fault identification in a cable temperature pre-warning model to improve the accuracy of the pre-warning system. Finally, the feasibility of the cable temperature pre-warning is verified by experiments and an early warning platform is built for cable temperature classification. The timely notification of various faults to enterprise management personnel is realized by this means.

A cubature information particle multi-sensors fusion algorithm is proposed to improve the precision of multi-sensors fusion. The algorithm combines Cubature Information Filter(CIF) with Particle Filter(PF), and adopts CIF to propagate particles of PF. The measurement information of multiple sensors is updated into particles of PF by introducing information contribution vector and information contribution matrix, to increase the approximation degree of particles to really posterior probability distribution, and to improve the precision of multi-sensors fusion. Meanwhile, the algorithm takes the state estimation of CIF as current particles, to eliminate the influence of random disturbance on multi-sensors fusion, and the precision of multi-sensors fusion is further improved. Simulation and experiment results show that, the algorithm can deal with the centralized multi-sensors fusion problem, and the filtering precision is high.

Wireless Sensor Networks(WSNs) consist a set of sensor nodes whose primary task is to sense and relay the data to sink. Sensor nodes are powered by a battery, and most of the energy is consumed for the relay of the data. The closer the nodes are from the sink, the more the data will be relayed, and the faster the energy will be consumed, which results in energy-hole problem. The introduction of a mobile sink for collecting the data from the nodes can avoid the energy-hole problem. Particle Swarm Optimization-based Rendezvous Point Selection(PSO-RPS) is proposed, which considers the data delivery delay and traffic rate constraints of sensor nodes for rendezvous point selection, and finds an optimal number of rendezvous points by particle swarm optimization. On this basis, the delay-ef?cient trajectory of mobile sink for data collection is built. Simulation results show that the proposed PSO-RPS algorithm can effectively control the path length and reduce the time delay of data collection.

Mobile users and wireless devices are now the sources of a large volume of data. In such data-intensive mobile and wireless computing systems, Delay-Tolerant Network(DTN) routing plays a critical role in data routing, dissemination, and collection. Therefore, Contact Duration-based Routing (CDR) is proposed in this paper. A new routing problem that the transmission of large volume of data cannot be completed in one contact is introduced ?rstly for data-intensive routing. Then the weight of the path is calculated by three kinds of information including contacting frequency, contacting duration and the available space of the cache area. Taking the path with the maximum weight as the shortest path, data is transmitted according to this path. Simulation results show that compared with Spray-and-Wait(SW) routing, CDR routing can effectively improve packet delivery rate.

The system architecture and function of an Internet+ based integrated intelligent power business hall are proposed aiming at the problems of incomplete system functions, weak information interaction, low service level, and scattered operation and management in the traditional electric power business hall. Firstly, the user requirements are analyzed to improve the service business architecture and network architecture of the intelligent business hall, and the functions that need to be realized by the intelligent business hall are analyzed comprehensively and systematically. Combined with the accurate matching function of the data of the intelligent power business hall, the reliability and stability of the service of the power business hall are effectively improved. This study will enhance the user experience, operational efficiency and the competitiveness of the electric power enterprises effectively compared with the existing research, and will provide reference for the future development of the electric power business hall.

Chua's circuit is a kind of chaotic circuit with simple structure and easy realization. Chua's diode is the core of Chua's circuit. Most of the existing Chua's diode models are implemented by dual-supply operational amplifiers and resistors. This paper sets the appropriate bias point for the operational amplifier and realizes the model of Chua's diode with a single power supply circuit. In the circuit implementation, the linear inductors used in the existing Chua’s circuit have the disadvantage of high price and difficult precision control. To tackle these problems, this paper takes the active simulated inductors to replace the linear inductors. The board-level circuit is designed and tested for two kinds of active simulated inductors. The test results indicate that using simulated inductor will not affect the chaos of the circuit under +5?V single power supply.

The model of interface traps and body defects caused by radiation ionization damage in oxides in semiconductor devices is studied. Using the Backward Euler method to deal with the time discretion and using the linearity method to deal with the nonlinear drift-diffusion reaction equation, the numerical simulation of the interface traps and body traps generated by the silica layer in the Metal–Insulator–Semiconductor(MIS) structure is completed. The algorithm is implemented on the high-performance parallel finite element software―Parallel Hierarchical Grid(PHG). The numerical results of the simulation are consistent with the Enhanced Low Dose Rate Sensitivity(ELDRS) in the ionization damage experiment and the data under different hydrogen ambient conditions. The corresponding models are analyzed aiming for the simulation results.