This PDF file contains the front matter associated with SPIE Proceedings Volume 12605 including the Title Page, Copyright information, Table of Contents, and Conference Committee Page.

In future communication systems, in order to improve the channel capacity, high frequency communication (i.e. mmWave communication / THz communication) has drawn a lot of attention. The high frequency signals suffer from a severe path loss. Fortunately, massive MIMO has been proposed to realize directional beam with a high array gain to compensate for the high path loss. In order to realize accurate beamforming, the Channel State Information (CSI) is of vital importance. One traditional method of acquiring the CSI is beam training, where the whole space is searched by a series of pre-designed codewords in the codebook, and the codeword with the maximum received power is selected as the direction of the users. However, the performance of this beam training method is limited to the angle resolution of the codewords, and the beam training overhead will increase with the increment of antenna numbers. In this paper, we propose a novel beam training scheme, which can finish the beam training in only two time slots, and the accuracy of the proposed beam training scheme is unrelated to the angle-resolution of the codewords. Simulation results show that the proposed method can reach the near-optimal beam training performance with a very low beam training overhead.

Under the condition that the demand for emergency materials is uncertain after the occurrence of an emergency, this paper considers the minimum time and cost of emergency rescue at the same time, studies the location problem of pre disaster emergency materials storage. The box uncertainty set is used to describe the uncertainty of demand, and a robust optimization model with capacity constraints is established. Based on Bertsimas and Sim robust optimization methods, the nonlinear robust optimization model is transformed into a robust corresponding model by introducing auxiliary variables and dual transformation. We use ε constraint method to transform the multi-objective problem into the single objective problem. An immune algorithm is designed to solve the problem. The feasibility of the model and algorithm is verified by computational experiments. The research shows that the robust optimization model has good robustness and can better resist the risk induced by uncertainty. Decision makers should determine the emergency rescue time constraint according to financial level and expected response level; evaluate the risk status, weigh the relationship between reliability and emergency rescue cost, and determine the robust uncertainty parameter for scientific site selection.

With the continuous development of computer and communication technology, it is difficult to meet the application requirements of high concurrency and low latency of IO such as high-performance computing and big data analysis in terms of transmission bandwidth and communication delay based on Socket protocol. In this paper, through the research on Remote Direct Memory Access (RDMA) protocol, including RDMA development status, RDMA key technologies, and RDMA communication principal concept, a RDMA communication method based on FPGA to implement Roce V2 architecture is proposed. Through software simulation and verification, the scheme proposed in this paper achieves the design expectation and has strong practicability.

With the development of intelligent transportation systems in recent years, automatic License Plate Detection and Recognition (LPDR) for automobiles has raised significant research interests. LPDR can be applied in many fields such as in traffic management and security surveillance. There has been a lot of work done on LPDR for regular scenarios. However, rare research can be found for complex traffic scenarios, such as night scenes, where the images or videos often appear to be blurred, too dark or bright, or the distance might be too far away or too close. These situations may pose great challenges to detection and recognition. In this paper, we established an automatic pipeline system for real-time LPDR in complex traffic scenes using DeblurGANv2 for image deblurring, YOLOv7 for license plate detection, LPRNet for license plate recognition. We used the CCPD dataset as well as a synthetic license plate dataset to train our models. Our results indicate that the system works well, and the deblurring module can significantly improve the results of LPDR.

While autonomous driving has made great strides, there are still some scenarios that need to be solved. For different kinds of vehicles, it is necessary to classify them accurately so as to adopt different driving strategies. At present, due to the good performance of deep learning in the field of image recognition, this method is mainly used for image classification. Therefore, facing the problem of vehicle classification, this paper adopts the method of combining EcaNet and ResNet to classify ten common vehicles in automatic driving perception. The experimental results show that the classification accuracy of the proposed method is 75.83%, compared with 66.46% of the comparison method.

Students use a large number of Web security training scenarios when conducting Web security experiments. In this paper, the running environments required for a large number of Web security training scenarios is classified into several categories. Each category uses one virtual machine. In this way, the consumption of computer hardware resources is reduced, and students' time is saved when students switch between Web security training scenarios in the same category.

As a mainstream heterogeneous programming model, OpenMP has important practical significance for its uninstall performance research. Songshan supercomputer system installed in Zhengzhou Supercomputing Center is a new generation of E-class high-performance computer cluster independently developed by China, and the DCU chip installed on it is also home-made. In order to improve the offload performance of OpenMP on the platform and make full use of hardware resources such as registers, a redundant cycle optimization for thread iteration in DCU was proposed, so that the thread could release register resources in time after completing the calculation task, thus relieving the back-end register allocation pressure and improving the program performance. At the same time, based on the loop unrolling optimization algorithm in LLVM and combined with the hardware characteristics and instruction set characteristics of the domestic platform, a better algorithm for calculating the loop unrolling factor was proposed to improve the optimization effect of loop unrolling. Thread iteration optimization using SPEC ACCEL and Polybench resulted in an average 33.7% reduction in the overall register count and a 37% average performance improvement after loop expansion optimization.

Aiming at the change of task requirements during the period of laser communication payloads in orbit and the potential design defects of software, this paper puts forward a design scheme of software on-orbit dynamic reconfiguration for spaceborne “FPGA and DSP”. In the aspect of DSP, based on the hardware architecture of MRAM and "FPGA+DSP", under the instruction control of the load controller, the dynamic reconfiguration functions such as version switching and on-orbit update of spaceborne DSP software are realized. at the same time, the dual storage mode of PROM and MRAM is constructed, which realizes the basic version fallback function in the case of abnormal reconfiguration and ensures the reliability of on-orbit operation. In the aspect of FPGA, the anti-radiation read-back refresh circuit is used to complete the read-write operation of the program in the NOR FLASH chip, so as to realize the FPGA reconstruction and program read-back refresh, improve the anti-single particle overturning ability and ensure the reliability. The scheme has flexible function and high reliability and has been successfully applied in the project.

With the development of the times, the demand for executing programs that require massively parallel computing on embedded devices has become increasingly strong. To meet this demand, various heterogeneous systems have been built on SoCs. In order to better use and manage heterogeneous systems, and to facilitate the development of heterogeneous computing programs. People will implement a heterogeneous computing framework in the operating system. Unfortunately, the reworks real-time operating system does not provide us with a heterogeneous computing framework for writing heterogeneous computing programs. We chose OpenCL as the heterogeneous computing framework for the reworks operating system and implemented the kernel state of the OpenCL heterogeneous computing framework, also known as the OpenCL driver.

This paper introduces a method of internet behavior detection. With the rapid development of society, the pace of people's life is faster and faster, and the competition is more and more fierce in all aspects. For example, college students often encounter mental imbalance in employment, study, life, love and self-consciousness. Now, the network, data and other concepts have been very closely embedded in people's life, study and work. A college student using a mobile device can generate hundreds of megabytes data during one day. By using Intel's open source Hyperscan engine, we can analyze the data of internet behavior by using the keyword matching technology of internet traffic, so as to quickly and effectively identify and warn psychological crisis.

With the increasing proportion of renewable energy in the new power system, the energy interaction process between distributed generation units, microgrids and energy storage systems is becoming more and more complex. How to improve the integration and reduce the operation cost through the coordinated control of virtual power plants and promote the development of new power systems is a problem that needs to be solved. Firstly, an improved cost minimization model of virtual power plant is proposed in this paper. An objective function including configuration cost, maintenance cost and operation cost is constructed to meet the constraints of voltage and power and optimize the configuration and operation plan of distributed generation, energy storage system and microgrid in virtual power plant. Secondly, the relevant parameters in the standard particle swarm optimization algorithm are adjusted adaptively to improve the iteration speed and convergence accuracy of the algorithm. Finally, the proposed model is applied to an actual commercial load, and three simulation cases are set up with different decision-making requirements. The simulation results show that the proposed method can better balance the configuration and operation costs of virtual power plant components and create a reliable, economical, efficient and environmentally friendly distribution system.

Thread parallelism and single-thread’ performance are two important factors affecting the performance of kernel functions, and they are both closely related to register allocation. According to change the thread parallelism to optimize GPU register resource allocation can effectively improve the performance of heterogeneous programs. We obtain the required number of vector registers by counting the number of virtual registers during the compilation of kernel functions, and then combine them with the number of wavefronts used to launch kernel functions for overall performance analysis, proposing a RAW compilation method for collaborative optimization of register allocation and thread management for AMDGPU, which is implemented in the LLVM compiler. It is verified that the method has a speedup ratio of about 1.12x for the Rodinia test set and about 1.4x for the quda application.

With the rapid development of the mobile internet technology, m-learning is gradually becoming a key modern remote learning way. With the combination of mobile technology and digitalized technology, a learning mode based on mobile terminals is created to provide a learning environment unlimited by the time and space for learners, thus meeting the real-time learning needs of learners. However, the current mobile terminal-oriented learning system is still in the exploratory stage, and how to build an intelligent learning system for mobile terminal access is a hot spot concerned by numerous researchers. Based on this, this paper explores the Intelligent Mobile Learning System (IMLS) solution from the two technical dimensions of user terminal recognition and learning content adaptability and describes its development environment and key technologies. Experimental results show that users’ convenience and access can be enhanced to some extent by the system, which will further enhance their learning quality.

The design is a multi-branch power parameter monitoring system based on Lora Network. The system adopts the IM2812B power acquisition module to collect the branch power parameters, the data transmission mainly adopts the Lora and WiFi wireless communication technology as the transmission media, and the transmission protocol is based on the Lora private protocol and the MQTT transmission protocol. The power parameters and environment data of each branch are displayed on the serial port PC and APP through Lora Network and can analyze the current branch of the electrical appliances and predict the existence of the security risk, make up the existing defects. It is not only to solve the remote monitoring problem, but also to solve the problem of wireless network multi-branch. The system supports the power parameter acquisition with 65,536 branches, and it can greatly reduce the data transmission delay by using the star network and FIFO. After testing, the system's data transmission delay is low and stable. It is easy to layout, and with low maintenance difficulties, it has positive significance for the power parameter monitoring system.

In this paper, the precise clock synchronization technology of IEEE1588 is studied, and a hardware and software co-implementation method based on the Zynq-7000 FPGA chip is proposed. In the aspect of hardware, the jitter delay of the protocol stack is shielded by stamping in the GMII interface and obtaining the timestamp information. In the aspect of the software, the protocol engine is transplanted in the user layer, at the same time, to realize register reading and writing and configuration, the driver is compiled into the Linux kernel in the driver layer. In addition, the real-time clock adjustment module of the slave clock is designed to adjust the clock quickly. The experimental results show that the clock synchronization method based on FPGA hardware and software co-implementation can realize the marking of the timestamp effectively, and the accuracy of clock synchronization reaches the level of sub-microsecond. In the meanwhile, the realizability and portability of synchronous system are also improved.

In the information age, with the increase of data feature dimensions, the performance of Intrusion Detection System (IDS) in training time and classification accuracy is declining. A large number of researchers have conducted a lot of research on how to use machine learning algorithms for fast and accurate intrusion detection. This paper presents an IDS model based on Extreme Learning Machine (ELM). Firstly, the intrusion detection data set NSL-KDD is normalized. Then, uses the normalized data as input, and the hidden layer activation function of ELM algorithm suitable for intrusion detection is optimized by comparing the correct rate of the model under different hidden layer activation functions and the number of hidden layer nodes. Finally, the optimal number of hidden layer nodes corresponding to the optimal hidden layer activation function is obtained. Experiments show that the optimized ELM has better performance.

Sound waves are considered the only information carriers that can be transmitted at a long distance in the ocean, and marine research, resource development, and maritime military struggles are inseparable from hydroacoustic technology. The development of underwater acoustic technology demands the support of acoustic transducers, and the mission task of underwater acoustic transducers is to transmit and receive sound waves underwater, so underwater acoustic transducers are called "underwater acoustic equipment eyes and ears". The underwater acoustic transducer is a type of sensor that realizes the conversion of sound and other forms of energy or information in the aqueous medium. Tonpilz transducer is used for relatively low-frequency high-power acoustic emission and is a common transducer configuration in sonar applications, consisting of piezoelectric ceramic rings stacked between two mass blocks, prestressed by center bolt, and tail and head blocks to reduce the resonant frequency of the device.

Large-scale network models based on transformer architecture have strong versatility in many fields. Due to the computation-intensive and large-scale characteristics of the model, large-scale training on domestic heterogeneous accelerators is restricted by aspects such as computing and communication efficiency, resulting in poor training performance. Aiming at this problem, the hot functions and performance bottlenecks in the training process are studied and analyzed, and the corresponding performance optimization methods are proposed based on the hardware characteristics of domestic heterogeneous accelerators. In order to solve the problem of low performance in low accuracy training, the low accuracy package optimization is carried out for the underlying matrix multiplication core operator. To solve the problem of significant startup delay of kernel function caused by fine-grained core operators, the LightSeq framework is transplanted on the domestic heterogeneous platform for the first time, and the core fine-grained operators are specially optimized to adapt to the hardware structure according to the characteristics of network structure to accelerate the training process. In large-scale training, in order to solve the problem of low bandwidth in cross node communication, distributed communication optimization is carried out from two levels of data transmission and hardware topology, and communication efficiency is improved by reducing the frequency of communication and increasing the communication bandwidth. The experimental results show that using the WMT '14 English-German translation dataset, the performance is improved by two times after optimization on a single node without loss of training accuracy. The computing scale is gradually expanded to 128 nodes (512 accelerator cards) for large-scale distributed training and verification. Under the premise of ensuring performance improvement, the scalability can reach more than 90% in 256 accelerator cards.

Logistics and distribution services have a wide range of commodities. Mixing different types of commodities can lead to loss of commodities, incurring additional costs and reducing the quality of distribution services. Therefore, it is also necessary to consider the distribution target when planning the distribution route. In this paper, we propose a soft time window distribution vehicle path planning problem considering commodity classification and establish an integer planning mathematical model for this problem. Next, the heuristic function of the ant colony algorithm and the pheromone update rule of are improved. The effectiveness of the model and the improved algorithm is demonstrated by citing international test datasets. Experimental results show that the model is able to assign commodities that are not suitable for mixed transport to different vehicles and avoid cargo transport losses. Although the transport distance increases, it reduces the total logistics operation cost, satisfies the diversified service needs of customers and improves customer satisfaction.

With the improvement of people's living standards, people have higher clothing safety and artistry requirements. Plant dyeing artworks use plant sap to dye natural and woven fabrics, thus creating a print close to nature, gorgeous in style, and free from the harmful components of chemical dyeing, which is recognized and widely used by highly knowledgeable people. However, the type and the design of plant dyeing artworks need to incorporate the artist's creative inspiration, which is limited by human work efficiency. It constrains the creation of plant-dyeing artworks and limits the speed of innovation. Deep learning has made significant progress in the field of artistic creation. In this paper, we propose to combine the disco diffusion model with the creation of plant-dyeing artworks. The high-quality image generation capability of the diffusion model and multi-modal content description is utilized to design and generate plant dyeing artworks. The original style is firstly modeled after photographs of the original plant dyeing artwork. The style type is further expanded using text descriptions. The disco diffusion model is then used to create the final plant-dyeing piece of art. The experiment results show that generating plant dyeing paintings based on the disco diffusion model achieves better visual results. Through the controlled generation process and high-quality work generation results, the stylized plant dyeing design works is obtained automatically. It provides strong support for the creation of plant dyeing art.

In order to formulate reasonable port dispatching plan, integrated continuous berth allocation and quay crane scheduling is studied in this article, considering the constraints such as the arrival and departure time of large ships under the influence of tides, a mathematical model is established, which is solved by using two-layer integer coding and Particle Swarm Optimization algorithm (PSO). According to the relationship and constraints of the parameters, a mixed integer programming model is proposed to minimize the ship's time in port. Finally, the effectiveness of the model and algorithm is verified by numerical examples of different scales, and the feasibility and superiority of PSO are verified by comparing the results obtained by PSO and Gurobi. It can provide reference for container port operation.

This paper proposes a self-learning interpolation algorithm based on Edge Structure Classification (ESC) of a single image, where the resulting high-resolution image I^HR can retain texture and edge information of the low-resolution one. The basic idea is learning the structure and features of a low-resolution image I^LR itself. Firstly, it down-sample the given I^LR to obtain its low-low-resolution I^LLR. Secondly, it learns the interpolation algorithm for enlarging I^LLR to obtain I^LR, which is based on edge structure classification. Each edge structure of I^LLR is mapped to a coefficient matrix, which is used to compute the values of the pixels to be inserted in the I^LR. Finally, the edge structures of I^LR are classified and the corresponding coefficient matrix is applied for compute the values of the pixels to be inserted in the I^HR. From our experiments, more types of ESC, better performance of the interpolation algorithm. Experimental results show that the algorithm based on ESC can achieve great performances of both PSNR and SSIM. The results can achieve smooth edge and fine details in subjective vision.

To improve the efficiency of blockchain system, a PBFT consensus optimization algorithm based on group reputation value (GV-PBFT) is proposed to address the problems of time extension, low throughput and arbitrary master node selection of PBFT consensus algorithm. First, group the nodes, the fuzzy value of the Vague set transformation of each node is calculated as the node reputation value within the group, and the node with the highest reputation value in each group is selected as the representative node to participate in the consensus. Secondly, a verifiable random number function is used to select the master node among the representative nodes to increase the randomness of the master node and make the master node less vulnerable to malicious node attacks. Finally, the consensus process is simplified by replacing the three-stage consensus of the PBFT consensus algorithm with a two-stage one. Experimental comparisons show that the GV-PBFT algorithm has improved in terms of the number of communications, throughput and latency.

With the rapid development and upgrade of the social network and IT, information security has become increasingly significant. Hash function has become a research key-point in the field of information security. To meeting the requirements of various circumstances, this dissertation puts forward Multilevel Security Hash Function (MSHF) and conduct a comparison experiment between MSHF and traditional methods to generate Merkle trees. The experimental results show that the Merkle tree generated based on MSHF has smaller depth, shorter path length, and shorter time used to find the wrong data.

To solve the problem of low routing efficiency and limited link utilization when Service Function Chain (SFC) deployment adopts the traditional shortest path algorithm in the face of large-scale network topology due to the expansion of the path search scope and only considering the shortest path each time, this paper proposes a routing strategy based on meta-heuristic algorithm to achieve the optimization of SFC routing. Firstly, the inertia weights in the original PSO algorithm are dynamically processed to adapt to the dynamic characteristics of Software Defined Networking (SDN) network topology and improve the optimization ability and convergence speed of the original algorithm. Secondly, the crossover and mutation of genetic algorithm are introduced to improve the ability of the algorithm to find the optimal path. Simulation results show that compared with the traditional k-shortest path algorithm, the proposed method can effectively improve link utilization during routing, reduce the routing time of large-scale network topology, and improve the SFC routing efficiency.

For the service composition under the multi-resource and multi-service constraints of the satellite network, without considering the problem that service quality and service priority lead to poor Quality of Service (QoS) and low service priority matching degree, the priority inverse number of service sequence is defined, the approximation is established aiming at the greatest QoS and the smallest priority inverse number of service sequence, and an adaptive genetic algorithm is proposed to solve the model. The algorithm improves the selection mechanism based on elite-reservation and roulette strategy and designs adaptive crossover and mutation operators. It solves the shortcoming that standard genetic algorithm is easy to fall into local optimum and avoids the loss of optional solution. The simulation results show that the proposed method increases the QoS of service execution by 13.461% and reduces the priority inverse number by 26.77%. It effectively solves the service composition problem with multi-resource and multi-service constraints in satellite network.

The deployment of scientific computing software in a high-performance computing environment is very complicated and often requires many dependent libraries. How to quickly build a scientific computing environment is very important. Containers provide a flexible way to deploy software due to their lightweight and easy portability. Taking the deployment of material simulation software VASP as an example, this paper proposes a standardized process for building HPC containerized applications. Use Docker and Singularity container technology to realize containerized packaging of applications, solve complex application deployment, and realize the goal of "manufacture once, run anywhere". Based on the original Slurm workload on HPC, combined with the message passing interface technology, it can realize multi-machine parallel processing of containerized jobs in a high-performance environment without changing the user's habit of submitting jobs. The lab part starts with HPL benchmarks and then evaluates the performance of VASP applications on containers and physical machines. Experimental results show that the containerization process is feasible and flexible, and the performance overhead after containerization is within 4%. This research is expected to provide an effective solution for the construction and operation of containerized applications in the high-performance computing field.

Optical pulse coding technique can improve some performance such as sensing distance of Brillouin Optical Time-Domain Analyzers (BOTDA) with the compatibility of its conventional device, and it often apply the Field Programmable Gate Array (FPGA) technique on pulse generation. To meet the requirement of fast measurement and system integration of BOTDA, a FPGA-based pulse coding and decoding technique is proposed in this paper, which accelerates the decoding process by parallelization and pipelining. The technique is tested in a BOTDA system with 100km measuring distance, 250MHz sampling rate and 255-bit Simplex coding, and achieves a decoding delay of 72ns to realize real-time decoding. In the experiment, the technique has an effect on improving the performance of BOTDA on measurement speed, sensing distance and spatial resolution on long sensing distance, comparing with the same system without pulse coding.

Big data is characterized by large amount of quantity, complex structure and real time performance. The main difficulty in processing big data is to improve the processing speed and accuracy. In this paper, some fast and incremental technologies are introduced, which usually used to increase the processing speed of the big data. And this paper also illustrates the application fields of these technologies. In order to speed up the query and classification of big data, index technique is proposed. The application of indexing technique can optimize candidate data sets, which could be used in data clustering and reachable query problems. In view of the increasing dynamic data, an incremental technique is used to process the big data. This method only considers the data that has changed, rather than re-modeling all the data each time as traditional methods did. Compared with traditional algorithms, incremental technique could get faster speed and higher accuracy in data prediction and graph query. In order to process the mixed big data quickly, some mixed frame for big data is given in this paper. The main mixed frames include flow-batch mixing frame and graph-flow fusion frame, which based on mixed data processing technique. At last, some experiments are given to prove the feasibility of these techniques.

Fuzzing is a technique to detect vulnerabilities in target programs by generating testcases in random mutation phase. However, for path constraints such as magic numbers in the program, the fuzzing random mutation phase can easily generate a large number of invalid test cases that do not result in new code coverage and trigger new crashes, resulting in low code coverage for Fuzzing. In this paper, firstly, the random mutation process of Fuzzing is investigated, and it is found that 1) dictionaries help to improve the code coverage of Fuzzing and 2) the random mutation stage generates a large number of invalid testcases due to randomness. Then, we propose a novel method called Dict-MAB (Dictionary and multi-armed bandit). Dict-MAB improves the code coverage of fuzzing by reducing the number of invalid test cases generated during random mutation. Specifically, the extracted proprietary dictionary can assist fuzzing to pass complex path constraints such as magic numbers, while modeling the random mutation phase as a multi-armed bandit machine model can reduce the randomness of this phase and reduce the number of invalid testcases generated. The experiments show that the number of valid testcases generated by Dict-MAB is improved by 27.49% on average. In particular, the code coverage on lcms and libjpeg improved by 13.71 % on average.

In recent years, edge data caching has been concerned by more and more research scholars. Caching data, especially popular data, on edge servers can significantly reduce users’ latency, while also greatly reducing direct data transmission between users and remote clouds and slowing down the congestion of the backbone network. Most of the existing research on edge data caching is mainly aim at minimizing users’ latency, saving devices’ energy, and improving the hit rate of data caching, etc. From the perspective of an app vendor, how to formulate an edge data caching solution that can bring him the most profit while complying with a given budget constraint and considering the experience requirements of different users is a very realistic and challenging problem. In this paper, we formulate the Profit-Maximizing Edge Data Caching (PMEDC) as a constrained optimization problem. For PMEDC in small scale scenarios, we propose an optimization model named PMEDC-IP that can derive the optimal caching strategy. For PMEDC in large scale scenarios, we propose an approximation algorithm named MPF (Maximum Profit First) that can find an approximate optimal caching scheme in a reasonable time. We conduct extensive experiments on a real-world dataset, and based on the final experimental results, PMEDC-IP and MPF proposed by us significantly outperform the other three representative methods.

The construction of a training base is a systematic project that requires the joint efforts of the government, enterprises and schools, and it is necessary to achieve comprehensive planning and reasonable layout in the process, so it is necessary to continue to conduct in-depth research on this issue. This paper explores the construction of a practical training base based on artificial intelligence technology and uses intelligent neural network technology to build an artificial intelligence experimental platform, which is technologically advanced, highly reliable and meets teaching needs. The experimental process simulates the real process of the enterprise, which can give full play to the ability of the practical training room and the enterprise process of long and close combination and keep in step with the current from the new research, closely combined with the practical application, which cultivates solid artificial intelligence application skills talents.

Studies towards botnet activity detection methods have become a major focus over the years since many concerns have been raised due to botnets. The detection against Domain Name System (DNS) queries, which bots widely used for their connection with the Command&Control (C&C) server, was considered the main method of botnet prevention. However, this method does not apply to Domain Generation Algorithm (DGA)-based botnets, which dynamically generate a large number of Non-existent domains, combined with fast flux. The approach of using machine learning in which the models are trained to study the characteristics of DGA domains is thought to be more effective than the methods that rely solely on a set of fixed rules. For such reasons, this paper will explore a novel method of botnet detection using machine learning. The effectiveness of this method is to be assessed in a system we designed for botnet detection, fed with truth marked dataset. The results indicate a high performance of our method with 95.3% in accuracy and a 5.4% false positive rate (FPR).

With the improvement of image processing tools and the flexibility of digital image editing, automatic image inpainting has found important applications in computer vision and has also become an important and challenging topic of research in image processing. Through the analysis of the exemplar-based image inpainting method, we found the calculation of priority of image patches to fill is beneficial to the textures than the geometric structure, while the geometric structure is more important to the filling than the textures, resulting in the problem of texture extension or the discontinuity of the geometric structure in the image. The image decomposition model is applied to decompose the image to in paint into two components: texture and cartoon, which are then filled respectively. We improve the calculation of the priority of the texture component and the cartoon component, which make it suitable to the image character. In order to solve the mismatch problem, we propose a measurement based on global structure information of the target patch and the candidate patch in this paper. The mean and variance of the pixels in the image patch representing the global structure of image patches added to the measurement formula can improve the accuracy of the best matching patch to fill the missing region. We compared our model with the Criminisi algorithm and its improved one, through applying the algorithm to the synthetic images and the natural image. We also compute the PNSF and SSIM values between the original images and the inpainted images to verify the effectiveness of the proposed algorithm. The experimental results show that the proposed algorithm can get more perceptually aware in painted results than the Criminisi algorithm and its improved algorithm.

For gate lobe suppression of distributed array, a method based on genetic algorithm combining array layout optimization and frequency synthesis is proposed to suppress grating lobe. This paper first introduces the principle of gate lobe generation and the basic principle of genetic algorithm, then the method of gate lobe suppression by array layout optimization based on genetic algorithm and the method of gate lobe suppression by frequency synthesis based on genetic algorithm are given. Finally, the method of combining space and frequency to restrain grating lobe based on genetic algorithm is presented. Simulation analysis shows that the combination of space and frequency has better suppression effect than the single array layout optimization or frequency synthesis.

Heterogeneous runtime systems, which use multiple devices for different types of computations, are important in scientific research and engineering and are becoming more common. They provide efficient and effective solutions to complex problems. However, for some data-parallel applications, communication among heterogeneous devices within a compute node can easily become a performance bottleneck, hindering the collaborative execution of multiple devices. This is particularly true when the devices are of different types or capabilities and may require different communication protocols or approaches to achieve optimal performance. In this paper, firstly, for the current data copy APIs used in the runtime system, we suggest implementing an optimization for the small-size data copy between different devices using the multi-vector access instructions supported in the X86 architecture. Second, based on our analysis of the transfer rate during the two phases of the pipeline copy process for the page-locked memory buffer, we have found that the host-to-buffer transfer takes up a significant amount of time. To optimize the data communication between the host and page-locked memory buffer, we make full use of multi-vector access instructions to reduce the transfer time. This approach can help improve the efficiency of the transfer. Third, we analyzed the transfer latency for various buffer sizes and numbers of buffers for the pipeline copy of the page-locked memory buffer and determined the optimal buffer configuration for different transfer data, leading to a significant reduction in transfer latency. According to experiments conducted on current heterogeneous computing nodes, our proposed optimization has reduced the transfer latency to 40% of the original value for small data segments (4KB-4MB) and to 90% of the original value for medium-sized data segments (4MB-100MB). These optimizations allow us to achieve higher transmission bandwidth between different devices.

The Davidson method is a traditional large-scale iterative diagonalization method in computational quantum chemistry. The Blocked-Davidson method is a classical version of the Davidson method, which is an important part of VASP software, occupying 22.9% of the calculation time. When VASP is transplanted to the new Sunway supercomputer, it is incompatible with the many-core heterogeneous environment of Sunway processor. The Blocked-Davidson method does not use slave-core accelerators. This paper designs collaborative computing, double buffer, store variables in slave-core LDM, LDM level two access, and segmented transmission methods of Blocked-Davidson method based on Sunway many-core architecture. The test results showed that the function in the Blocked-Davidson method accelerated 30.65 times with a single core group in the silica test case. However, negative optimization of the Blocked-Davidson method happens in some test cases such as ZrTiAlVSiN5 and Hg50. Therefore, an adaptive algorithm of automatic selection of less time-consuming optimization paths is designed to solve the negative optimization. The maximum acceleration is 23.6 times compared with not using the adaptive algorithm.

At present, Graph Neural Network (GNN) methods usually follow the node centered message passing process and rely heavily on smooth node characteristics rather than graph structure. In view of this limitation, based on the heuristic method and graph attention mechanism, a feature fusion link prediction model (SAFL) combined with graph neural network is proposed. This model extracts the enclosing subgraphs around the target, combines the attention mechanism to assign neighbor weights to learn useful structural features, considers the impact of different nodes on the link, and fuses the graph neural network with the characteristics of input nodes to predict the link. The experiment on OGB dataset shows that the proposed link prediction model based on heuristic method enhances the graph structure characteristics, effectively represents the connectivity of enclosing subgraphs, and achieves better performance in link prediction.

In order to enhance the performance of wireless powered Mobile Edge Computing (MEC) systems, a multiple Intelligent Reflecting Surfaces (IRSs) empowered transmission scheme based on Non-Orthogonal Multiple Access (NOMA) is proposed. Specifically, multiple IRSs are deployed between the Hybrid Access Point (HAP) and Wireless Devices (WDs) for assisting the downlink wireless power transfer from the HAP to WDs and uplink computation offloading from WDs to the HAP. The WDs adopting the binary offloading scheme offload their computation tasks to the HAP via NOMA for remote computation at the MEC service embedded at the HAP, and the discrete phase shifts are considered at the IRSs for practical implementation. To maximize the system, sum computational rate, we formulate an optimization problem with respect to time allocation, computing mode selection, phase shifts for the downlink and uplink, transmit power at the WDs, and local computation frequencies, and propose a Block Coordinate Descent (BCD) based algorithm to solve it efficiently. Numerical results indicate that the proposed scheme can significantly improve the system performance in terms of sum computational rate and verify the impact of discrete phase shifts on system performance.

Metasurface-based terahertz absorbers have recently gained much attention in research on terahertz devices. However, the actual applications of such absorbers are severely constrained by issues such as difficult preparation procedures, constrained bandwidth, polarization, and incidence-angle sensitivity. This paper proposes a tunable broadband terahertz absorber scheme based on the metasurface of all-dielectric microspheres. The device consists of a single layer of strontium titanate microspheres compounded with a thin-film base. The Mie resonance of the microspheres works together with the metal film to achieve excellent absorption performance, while the modal coupling between the microspheres greatly broadens the absorption bandwidth. According to these findings, the device is capable of tunable absorption performance with a bandwidth of up to 0.18 THz. The terahertz broadband absorber proposed herein has a simple structure, superior performance, and low preparation cost, as well as potential applications in terahertz sensing, terahertz detection, and terahertz imaging.

With the development of the internet, enterprises and individuals pay more and more attention to network security. Various anti-virus software, active defense and other protection products have developed rapidly. It is not ideal to use traditional Anti Anti-virus technology. In this paper, we propose a Anti Anti-Virus (AAV) technology based on shellcode. First, generate shellcode using Cobalt Strike. Then, by using encryption obfuscation, sandbox evasion, API substitution call technologies, shellcode is encapsulated secondary. At last, compile it into executable file with Golang. Test the executable file on the local anti-virus software. At the same time, upload it to VirusTotal for multi-engine online testing. Testing results indicate that the executable file can run, and it is suitable for penetration testing in daily.

With the increasing scale of applications and system programs and the increased processing capacity of software, in order to improve the efficiency of software, more and more information-sharing technologies are adopted. There are more and more loopholes in large-scale information-sharing, mainly because of the uncertainty of information-sharing scheduling, which leads to great uncertainty in the control flow of programs. In This paper, the theory of fuzzy cooperation evaluation is applied to the research. Through the analysis of the participants in data-sharing and their fuzzy cooperation relationships, the fuzzy cooperation evaluation model of data-sharing is constructed based on the degree of participants' participation. The benefit allocation model of data-sharing subjects is constructed by using the fuzzy Shapley value and Banzhaf-Coleman right index. It provides a new research perspective and ideas for the smooth long-term practice of data-sharing and the equilibrium of its rights and interests.

Most of the deep learning-based image compressive sensing network models have been able to reconstruct good quality images, but at low sampling rates, the reconstructed images often lack details. To solve this problem, an information enhancement and refinement network model for block-based image compressive sensing (IERN_BCS) is proposed. The model first uses the block-based compressed sampling sub-network to compress images, then an initial reconstruction sub-network is implemented to recover initial reconstructed images, and finally a deep reconstruction sub-network consisting of the Information Extraction and Enhancement Block (IEEB) and an Improved Information Refinement Block (IIRB) are applied to obtain high-quality reconstructed images with rich details. Experiments show that, compared with existing state-of-the-art image compressive sensing network models, IERN_BCS can prominently improve the reconstruction performance.

According to the transmission capacity currently provided by the Beidou-3 system, a data transmission system based on the Beidou-3 Navigation Satellite System (BDS-3) was designed to address the problems of the original data transmission of the hydro-meteorological observation system. The data transmission protocol was defined, and the data transmission message was composed of multi-minute observation data superimposed and encrypted by DES. The active self-reporting data transmission scheme and the interactive query data transmission scheme were elaborated. Data security is ensured by both data encryption and BDS-3 satellite communication; the data transmission message consists of multi-minute observation data overlay to ensure the arrival rate of real-time observation data; and the data integrity is ensured by the interactive query method.

The cyber security event log method stipulated in the IEC 62351-14 standard defines and describes the cyber security events that might be encountered in the cyber security of the power system substation and gives an approach on how to map these cyber security events to the Syslog protocol for transmission and recording. Based on this theory, the cyber security process based on application-layer authentication has been deeply researched, cyber security events for application-layer authentication have been expanded and defined, the content of the standard has been improved, its application scope has been expanded, and its application scenarios have been enriched. Cyber security event log technology has been applied in actual substations, which has effectively improved the monitoring capability of cyber security conditions in power system substations, enhanced the level of analysis and diagnosis of cyber security failures, and made it more convenient to further promote the engineering of cyber security in substations.

Aiming at the shortcomings of short distance of micro-power wireless communication and high cost of fault indicator, this paper puts forward a LoRa solution with long distance and low power consumption. Firstly, this paper elaborates the technical matching degree between LoRa and fault indicator service requirements from the aspects of coverage, speed, delay and power consumption. Then, by comparing the cost of fault indicator based on LoRa and micro-power wireless technology, the economic benefits of the scheme are analyzed. Finally, the system is tested and analyzed. The test results show that this scheme has the characteristics of long communication distance, low power consumption, convenient networking and so on, and has a wide application prospect.

To improve the sensing accuracy of the power fiber optic communication system, an intelligent sensing method of fiber optic network threat based on multi-information fusion is proposed. The architecture is composed of an information acquisition layer and a quantitative sensing layer. Multiple sensors are used to extract the multi-channel signal data of the optical fiber network. After the data is de-noised, the artificial neural network is used to implement multi-information fusion on the de-noised data. The data with the highest network security impact in the fusion information is extracted, and the information is transformed into the threat state through quantitative perception technology. After using the abstract ability to create the change process of the situation, through continuous reasoning, the complete situation of the network threat is displayed to realize the intelligent perception of the network threat. The experimental results show that the method has a high data fusion calculation detection rate and fusion rate, a low false alarm rate, and a strong data fusion calculation capability; it can effectively sense the threat potential of different networks, and the perception correct rate is always higher than 97%.

With the increasingly fierce offensive and defensive confrontation in cyberspace, security threats have spread to the field of network terminal equipment data monitoring and collection, and the problems of network terminal equipment data collection being damaged and tampered with under the offensive and defensive environment are prominent. To this end, this paper proposes an SGX-based network countermeasure host log highly trusted collection method, adopts SGX hardware-level protection technology as the security reinforcement means, and conducts research on log collection methods, transmission methods, and storage methods. In this paper, the SGX-based network countermeasure host log highly trusted collection system is designed with full consideration of the collection strategy for host logs of multi-operating system platforms. At the same time, not only the physical host collection scenario is considered in the scheme design, but also the host log collection task for virtual machines in the cloud platform with the continuous development of cloud computing, which greatly broadens the application scenario of this method and satisfies most of the network scenarios in reality.

In this paper, a radio signal detection algorithm is proposed. This algorithm detects the radio signal from complex electromagnetic environment to obtain the center frequency, bandwidth and starting and ending time. First, to detect the time and frequency information simultaneously, the I/Q data in time domain is transformed into spectrogram. Then, the noise threshold is calculated by equidistant binning statistics, and the spectrogram is binarized based on the threshold. A kind of filter operator is designed for binary image denoising. Then we use DBSCAN clustering algorithm to cluster signal points in binary spectrogram. Finally, after separating the overlapping signals and filtering the false signals, the center frequency, bandwidth, starting and stopping time of each signal can be extracted. For experimental evaluation, Average Precision (AP) and average Distance-DIOU (ADIOU) are used as evaluation indexes. Radio signal data with center frequency near 2.4GHz and bandwidth 40 MHz is collected. And 420 segments of test data are extracted for manual labeling. Detection experiments on the 420 test samples reveal that the algorithm we proposed can achieve an AP value of 0.839, and the ADIOU value reaches 0.674. The result shows that the algorithm can detect radio signals accurately.

For the SIFT algorithm having great computational complexity caused by too many feature points and complicated descriptors, an advanced image registration method is proposed. At first it uses SIFT to extract the feature points from two images and eliminate edge points. Next it introduces image entropy into feature point threshold to select feature points with better quality and reduce the number of feature points. Then it uses ORB algorithm to describe the feature points and Hamming distance nearest neighbor method to match the feature points. Lastly it uses the advanced voting law based on the constraint relations of feature points to purify the matching points and the improved PROSAC algorithm to calculate the optimal transformation matrix. Experiments verify the performance of the algorithm.

Energy spectrum CT can provide higher CT imaging resolution and thus play an important role in clinical diagnosis. X-ray Absorption Spectrum (XAS) instrument based on photon counting detector has great potentials in energy spectrum CT. The main problems of PCDs are “Stacking effect” and “charge sharing”, which can cause spectrum distortion and therefore depress imaging quality. It is quite complicated to correct spectral distortion through hardware improvements and physical simulation. Hence, we proposed a method for x-ray absorption spectrum data correction based on adversarial neural networks. The generator of our proposed can generate corrected spectrum after training. Spectrum data for correction in our experiments was collected by laboratory equipment and the ground-truth was obtained by simulation software. Real data and simulated data were feed to the network to train the generator and discriminator simultaneously. Results of our experiment illustrated that the well-trained network can effectively recover the spectrum from the distorted spectrum data. We also evaluate the imaging quality with the corrected spectrum data. It can be seen that the quality of CT reconstruction image can be significantly enhanced through corrected spectrum data. Our method of using adversarial neural networks to generate noise-free x-ray spectrum provides new ideas for the clinical application of energy spectrum CT.

Aiming at the problem of pedestrian behavior recognition in infrared images, a method based on Improved GoogLeNet is proposed. Firstly, by analyzing the application scenarios and the characteristics of common network models, GoogLeNet with better comprehensive performance is selected as the backbone network. Inspired by NIN, a kind of 1*1 convolution kernel structure is introduced to reduce the number of channels and significantly reduce the number of parameters. Then channel padding and resize to adapt to the network requirements for the training set and test set of the infrared image human behavior data set. Next, the fully connected layer and the classification output layer of the network are modified according to the number of behavior types contained in the data set. The convolution kernel and inception parameter in the pre-training network are introduced to accelerate the network training and improve the generalization ability of the network. Finally, the quantitative index is used to analyze the experimental results and judge the recognition performance of the network. Experimental results shows that the Mean Average Precision, Average Recall and F1 score obtained by the proposed algorithm are better than the traditional methods.

Space-Time Adaptive Processing (STAP) technology has advantages in clutter suppression and is gradually applied to radar jamming suppression. In this paper, three algorithms of STAP have been compared: the Displaced Phase Center Antenna (DPCA) algorithm, Sample Matrix Inversion (SMI) algorithm, and the Adaptive Displaced Phase Center Antenna (ADPCA) algorithm. The simulation of anti-jamming by them is carried out through the new MATLAB Radar Toolbox. Experimental results show that the SMI algorithm and ADPCA algorithm can suppress interference.

The power communication network suffers from low evaluation accuracy when the error in the probability value, risk value and impact value of security events is higher than 0.02. This paper proposes a quantum key-based security analysis method for power service communication networks, with reliability, resilience, adaptability and self-healing selected as security assessment indexes for power service communication networks. Then, the corresponding weights of analysis indicators are calculated by combining hierarchical analysis method and entropy weight method, and the security evaluation of power service communication network is realized by multi-attribute decision method. The test result shows that the proposed method can accurately assess the possibility value, risk value and impact value of security events in power service communication networks, and the accuracy of the assessment results is high.

Emotion is an essential aspect of human life, and effectively identifying corresponding emotions from different scenarios will help promote the development of human-computer interaction systems. Therefore, emotion classification has gradually become a challenging and popular research field. Compared with text emotion analysis, emotion analysis of audio data is still relatively immature. Traditional audio sentiment analysis research is based on feature information such as MFCC, MFSC, etc. while using time-memory models such as LSTM and RNN for emotion analysis. Due to the rapid development of transformers and attention mechanisms, many scholars have shifted their research from the RNN family to the transformer family or deep learning models with attention mechanisms. Therefore, this paper proposes a method to convert audio data into a spectrogram and use a vision transformer model based on transfer learning for emotion classification. This paper conducts experiments on the IEMOCAP dataset and the MELD dataset. The experimental results show that the emotion classification accuracy of the Vision transformer in the IEMOCAP and the MELD datasets reach 56.18% and 37.1%.

With the development and application of Model-Based Systems Engineering (MBSE), it is more and more accepted and valued in the aerospace field, and System Modeling Language (SysML), as the most popular model language for MBSE, has become more and more important for the correctness and safety verification of its models. SysML views are described from different angles of the development system in MBSE, and it is very important to verify the relationship between each view, so this article summarizes how to extend its model verification part based on the open-source project Papyrus, and successfully extends the correlation validation between various views of SysML.

The simultaneous implementation of object ranging, and object calibration based on Received Signal Strength Indication (RSSI) using RFID is the main focus of this research paper. Ranging technology is a low-cost and low-complexity distance measurement technology that is widely used in distance-based localization technology for wireless sensor networks. Due to the influence of non-visible range and multipath transmission in indoor environment, the ranging error is relatively large. The linear fitting characteristics of RSSI ranging model with path loss model in indoor environment are analyzed for the basic method of wireless signal ranging, signal channel characteristics affecting ranging in indoor environment, RSSI model and the actual measurement values of the relationship between RSSI and distance in indoor environment. In addition, the object calibration is carried out by using the electronic tag of RFID, and the experiment shows that the determination of the distance of the object can be realized while the object calibration is carried out.

Charge pump is an important module in analog phase - locked loop. The mismatch current greatly affects the in-band noise of CPPLL. To reduce the charge pump mismatch current to improve the overall noise performance of the PLL. In order to reduce the influence of channel length modulation effect on current source, a compensation current is added to the current source of traditional charge pump, and the current is determined by feedback loop. The simulation results show that when the output voltage ranges from 0.4V to 2.8V after the addition of the structure, the charge-discharge mismatch ratio decreases from 7.9% to 0.34%. Thus, the noise performance of the PLL is greatly improved.

In most memristor circuits, a dual voltage is used to adjust the memristor value. This paper proposes a memristor change circuit in which only positive voltage changes the memristor value. A bidirectional change circuit of memristor resistance value composed of three transistors and two memristors is proposed, and a mathematical calculation method for approximate analysis of memristor changes is used. The dual synapse function can be realized by using the circuit and can be directly controlled by digital logic signals, which greatly reduces the number of devices and provides convenience for the integration of neural synapses.