This PDF file contains the front matter associated with SPIE Proceedings Volume 12714 including the Title Page, Copyright information, Table of Contents, and Conference Committee Page.

Batch computing services have the characteristics of providing services required by users on demand. Furthermore, failure factors such as resource constraints and scheduling can lead to service violations. The accountability mechanism is essential to be established, which integrates analysis, Network generation, Determination, and Tracing. There are functional modules that have done some research on the parts of the accountability mechanism. However, in order to integrate them into the accountability mechanism, the accountability framework is essential for batch computing services. Therefore, this article aims to present a survey of functional modules of accountability mechanism, presenting and comparing main features described in the literature and challenges of integrating the accountability framework. At the end of this work, we provide suggestions for future research directions.

With the increasing number and categories of cyber-attacks in the era of big data, intrusion detection techniques are constantly updated and optimized. In order to address the shortcomings of traditional intrusion detection methods and single neural network models in intrusion detection, this paper proposes a deep learning-based intrusion detection method. By combining a one-dimensional convolutional neural network with a bidirectional gated recurrent unit, a new CNN-BiGRU network model is formed to fully extract intrusion detection data features and improve the accuracy of intrusion detection. Using the public dataset CIC-IDS2017, multiple sets of comparison experiments were conducted on the proposed method in this paper in the same environment. Firstly, the ablation experiments yielded that in terms of detection performance, the accuracy of this paper's method improved from 99.50% and 99.34% to 99.58% compared with CNN and BiGRU-based models respectively. In terms of running time, the detection time of this model is reduced from 1921s to 907s compared with that of BiGRU, demonstrating the good detection performance of this method. Finally, through comparison tests, it is verified that the intrusion detection method proposed in this paper has better detection performance compared with other methods.

With the rapid development of Internet technology, the network environment has become more complex, and all kinds of campus information applications are constantly enriched. The security of the campus network has become more important. To monitor and evaluate network security effectively, network security situation assessment technology comes into being. In this paper, data fusion technology is used to integrate all kinds of equipment and system log data on-campus network, and a principal component analysis algorithm is used to reduce the dimension of the data. A security situation assessment model based on the entropy weight method and analytic hierarchy process is built to evaluate the threats the whole network suffered and analyze the network's security state.

With the development and popularization of IPv6 technology, cyber security is extremely important to national security. The evolution of campus network architecture to IPv6 has become an unstoppable trend, but we do not know much about the access security issues under IPv6 architecture. There are some security threats in IPv6 campus network, such as address acquisition spoofing, address resolution spoofing and flooding attack. In this paper, SAVI technology is proposed to solve the security problem of address acquisition spoofing network in DHCPv6 and SLAAC address acquisition mode under IPv6 architecture.

At present, smart grid information platforms mostly use containers to realize the miniaturization deployment of power services. However, containers interact frequently with the outside world, their performance is fragile, and they are easy to be paralyzed and become the target of attacks, which seriously affects the safe operation of the information platform. Therefore, this paper proposes a Docker based security protection system for container applications, designs the system architecture and each sub module, and focuses on the key technologies of container information collection and exception detection. The security system can record the behavior of Docker containers, use artificial intelligence technology to conduct real-time analysis on the recorded data, find exceptions in advance, identify attacks, and deal with security threats according to the pre-established rules and strategies, so as to provide a comprehensive guarantee for the safe operation of the smart grid information platform.

All data for training and evaluating recommended systems are subject to selection biases. In much research, principled approaches have been found to manage selection biases by adapting estimation techniques and models from causal inference. However, no matter what kind of method is adopted, the problem of data randomly missing always exists. This paper tries to discover whether the deeper model can effectively bring better prediction results and debias. We theoretically and experimentally examine whether the models are robust or not.

With the trend of networking of storage systems, storage systems have gradually become an important target for malicious attackers, and the result of attacks is often serious data accidents or encryption accidents. Therefore, the security of the storage system has become an issue that has yet to be resolved. In order to ensure the security of the data in the storage medium, a cryptography algorithm is used to encrypt and decrypt the data. A key distribution mechanism based on the K-means algorithm is designed to provide a reliable security guarantee for key transfer between various entities in the system. By comparing the time-consuming and storage efficiency ratio of the K-means algorithm and the rough set algorithm, the experimental results show that when using K-means, the rate of the storage device is slightly higher than that of the algorithm, indicating that the performance of K-means is slightly higher than that of the rough set algorithm when using stream encryption.

In the computer age, the rapid development of network technology has had a profound impact on human society, and brought many conveniences to people's life, work and study. At the same time, the endless network attacks have also brought great harm to the society, not only causing great property losses to individuals and enterprises, but also seriously affecting and threatening the security of the government and the country. As an active security defense measure, intrusion detection technology can effectively prevent the harm caused by intrusion. Based on the processing and analysis methods of big data and the characteristics of intrusion detection technology, this paper puts forward the basic method and detection process of computer intrusion detection, which has certain reference significance for the research of computer intrusion detection technology.

In intrusion detection based on traditional BP neural network, the BP neural network algorithm model has some defects, such as easily falling into local optimum and random initial value. The selection of the initial value directly affects the training effect of BP neural network, and a better initial value is conducive to the BP neural network skipping the local optimum, thus improving the training efficiency. Aiming at the defects of BP neural network, this paper puts forward a method of optimizing the initial value of BP neural network by genetic algorithm, so that BP neural network can get a group of better initial values. The experimental results show that the application of BP neural network optimized by genetic algorithm in intrusion detection can significantly improve the detection rate and convergence rate of the algorithm.

In order to realize real-time dynamic monitoring, remote visualization in different places, and improve the efficiency of data collection, transmission and processing in the process of subway security risk monitoring, a data security risk monitoring system for Internet of Things is proposed. Firstly, this paper puts forward the application of Internet of Things technology in subway safety monitoring. Through the application of information perception, processing, security and other technologies, a real-time online perception and intelligent management system of underground environment is constructed, which provides a safe and controllable real-time monitoring function. Secondly, in each module of the monitoring system, this paper mainly makes discriminant analysis for the early warning module, and establishes the subway safety early warning model by using fuzzy comprehensive evaluation method, and compares it with other comprehensive evaluation methods such as BP neural network. The results show that the fuzzy comprehensive evaluation method has strong theory and application in subway safety early warning. The similarity between the evaluation results and the results of multi-index grey interval number correlation decision, artificial neural network and regression analysis is above 82%, and it is in line with the actual situation on the spot. It can reliably provide scientific and reasonable theoretical basis for subway safety early warning and formulating risk response measures, guide site construction, and provide reference for other similar projects.

Program static analysis can be utilized to automatically investigate the execution procedures of programs. However, the proof of program completeness verification is less concerned by utilizing current static analysis tools, which only focus on the execution results verification. Therefore, we initially propose a novel completeness verification mechanism to provide a method to illustrating the targeted program is completed with required functions. We transfer the programming language into abstract symbols and establish the control flow graph of program, which can apply in all programming languages. Our proposed schemes significantly demonstrate completeness of program and more efficient than existing static analysis methods. From our experimental and evaluative results, proposed static analysis mechanism can effectively proof the completement of target program compared with existing methods and the computation cost is reasonable for the analysis period, which is decided by the size of programs.

Compilers play a very important role in the development of software. More and more researchers are studying how to improve the reliability of compilers. An effective method is to test the compiler with equivalent programs to check whether the running results of equivalent programs are still equivalent after being compiled by the compiler. Generating equivalent program sets is the key to solve this problem. This paper proposes an equivalent program generation tool DeCsmith, which is based on the development of Csmith. Through the equivalence relationship, the program generated by Csmith is equivalently changed, and the equivalent program set is automatically generated. The effectiveness of the method is verified by an example. The work of this paper is beneficial to saving human resources in compiler testing and improving the efficiency of testing compilers.

With the rapid development of the Internet and technological innovation, the number and resources of computers accessed in P2P (Peer to Peer) peer-to-peer networks are increasing. Another important feature of P2P is to change the current ethernet centric state of the Internet, return to "decentralization", and return power to users. Its openness, anonymity and autonomy increase the possibility of security attacks by malicious nodes. For example, malicious nodes providing false services will affect normal transactions between P2P network nodes. This has affected the development of P2P applications to a certain extent. In view of the existing P2P network node security problems, cannot deal with the node collusion attacks, joint fraud and other issues, this paper proposes a new trust calculation method. This method draws on the basic elements of social psychology about establishing intimate relationship between people, considers the influence of node behavior on trust, and uses aggregated trust, current trust, feedback trust and cumulative abuse trust to calculate and update the trust value, so as to improve the dynamic adaptability and fault tolerance of peer-to-peer networks.

The block cipher stands out among the reliable methods for data security. Key expansion is a crucial step in the block encryption algorithm and is thus important to develop secure round keys that are statistically independent and sensitive. Using Nonlinear Congruential Generators (NCGs), we propose a key expansion algorithm that helps to design more secure block encryption algorithms or hash functions. Despite the advancement in digital technologies, NCGs remain the effective method of Pseudorandom Number Generation (PRNG). However, conventional linear congruence generators have difficulties in applying the key expansion algorithm. In contrast to the conventional linear congruence generator, the round constant extension algorithm requires a random sequence of 01, which produces random integers within a certain range. To improve the Advanced Encryption Standard (AES) key expansion algorithm, we propose an NCG over a Galois field GF ( 2⁸ ). Our analysis includes analyzing a key expansion of 128 bits in length and a key of other lengths performed similarly. The experimental results show that the proposed algorithm is feasible and resistant to side-channel attacks. Our findings can be used to improve existing block cipher algorithms and make them more secure.

Floating garbage on sea surface has always been a key issue in the long-term research of environmental pollution. In order to effectively solve the problem of marine garbage pollution, this paper conducts in-depth research on the existing VGG16 network model and proposes an improved lightweight VGG network model. Instead of the fully connected layer, our model uses the global average pooling layer to reduce the number of network parameters and adds a residual module to the convolution module to improve the accuracy of the model. The experimental results show that the accuracy of the improved lightweight VGG network model is as high as 97.8% in the self-built sea surface waste data set. Compared with the traditional VGG16 network model, the number of parameters is reduced by 98.5% and the calculation amount is reduced by 78.5%, achieving the goal of rapid and accurate classification.

The loss of precision in floating-point programs is necessary to detect the loss of precision in floating-point programs because of the computational errors in floating-point operations, which can lead to the loss of precision in floating-point programs and affect the reliability of the upper-level software systems that call them. The inherent storage form of floating-point data makes it impossible to represent all real numbers fully, and the rounding error and error accumulation in floating-point operations make it difficult to detect the precision loss in floating-point programs. In this paper, we propose a dynamic analysis method to detect the precision loss of floating-point programs. First, we analyze the organization of the source code with the front-end Clang tool of LLVM to locate the floating-point data declarations and floating-point operations. Then we run the floating-point operations simultaneously with the higher precision MPFR operations with the help of the Dyninst staking framework to detect the precision loss at the statement level and record the precision loss of the floating-point. The precision loss of each step of the program is recorded. The precision loss change graph is generated to locate the floating-point precision loss. By testing the commonly used functions in the GSL library and some test cases in FPPench, the relative error of the method in this paper is improved by 43% on average compared with the randomly generated input data method to detect GSL functions. The relative error is improved by 1.8 bits on average compared with the Herbgrind tool for testing against FPPench.

Due to the increased sharing between modern autonomous vehicles and weak protocol design principles, the on-board Controller Area Network (CAN) bus needs more security mechanisms to identify anomalous access and protect itself from internal or external network attacks. Current security solutions, such as encryption, authentication, or deployment of Intrusion Detection Systems (IDS), improve vehicle security to some extent but still have limitations. This paper designs a new CAN admission control method that uses voltage and bit time as Electronic Control Unit (ECU) fingerprinting. Compared with existing fingerprinting solutions for CAN, this method improves the effectiveness and efficiency of device fingerprinting, thus identifying more types of attacks. Experiment results show that the method can identify access sources with less than 1.6% error rate.

With the increasing number of Internet of Things (IoT) access devices, security incidents in the IoT field have occurred frequently in recent years. Traditional intrusion detection techniques can no longer meet the current cyber threat discovery needs in the IoT environment. For this reason, we proposed a deep reinforcement learning (DRL)-based IoT intrusion detection system using an intrusion detection dataset produced by an IoT-based platform to enhance network security in the IoT environment in this study. The system applied a feature selection algorithm based on Pearson's correlation coefficient to extract the most effective feature set, applied a multilayer perceptron containing four hidden layers as the deep neural network structure shared by the value network and the policy network in this intrusion detection system, and constructs an intrusion detection system based on DRL proximal policy optimization algorithm. Based on the experimental results show that the proposed intrusion detection system obtains 99.96% accuracy in the face of different network attacks on IoT and outperforms current deep learning models based on LSTM, CNN, and DQN in terms of accuracy, precision, recall, and F1-measure.

Program static analysis is of great value of source code software vulnerability detection, but it is often limited by scalability bottlenecks. Constraint solvers are inefficient due to complex program dependencies on millions of lines of program source code. A single solver is difficult to get the balance between the accuracy and the time cost. This paper discusses the program dependence and constraint solving of static value-flow analysis, and specifically implements a solver rating system based on static taint analysis, which selects the most efficient solver for program dependence of critical path to reduce the false-positives and time cost of static vulnerability detection. Through testing for Juliet test sets and several real-world projects, we found that the overall performance of the system was better than other single SMT solvers or default scheduling strategies.

Deep neural networks are widely used increasingly in the real world, such as machine learning as a service (MLaas) which makes the owner of the model can deploy their fully trained model in the cloud for others to use. By paying remuneration and inputting the object into the model, customers can receive the predictions. Unfortunately, these network models are often threatened by model stealing attack. This attack uses the predictive information given by the original model to train a surrogate model with similar functions, so as to achieve the purpose of infringing the intellectual property of the model owner and the white-box adversarial sample attack against the clone model can transfer to the original model to a certain extent, which will result in more far-reaching damage and impact. But the model stealing process is often ineffective because the training data or related agent datasets of the original model cannot be obtained, for such data is often hidden by the model owner due to its privacy. On this basis, the model stealing attack without data has gradually entered people's view. The data generator will generate data to query the model to achieve the purpose of stealing. In this paper, a data-free model stealing attack frame based on gradient prediction and uniform sample generation has been proposed. Experiments show that this method can obtain a high imitation accuracy in a given query cost range. The performance (0.94x-0.99x) of the stand-in model is better than other MS attacks that rely on part of the original data set (such as JBDA, an attack that uses a part of the original data and other synthetic data generated from the original data by Jacobi data enhancement method to train the clone model, clone accuracy 0.13x-0.69x) or rely on the proxy data set (such as KnockoffNets which uses the surrogate data which has the homologous distribution with the original data to train the clone model, clone accuracy 0.52x-0.97x). At the same time, compared with other data-free MS attacks proposed recently (such as DFME, the SOTA method that uses the data randomly generated by generator model to train the clone model, clone accuracy 0.92x-0.99x), it also has the same or even higher performance. At the same time, the rate of success for adversarial examples attack through the surrogate model is also better than other data-free black-box attack.

In recent years, the demand for security in public places and buildings has been increasing. With the continuous development of science and technology, intelligent security gradually integrates into social life. Because the security system of computer vision generally has the problems of large storage resources and cannot be stored for a long time, the intelligent analysis of audio signals takes up the advantages of narrow bandwidth and less storage space. This paper will propose an audio security system based on the Internet of things, which realizes the monitoring and early warning of security events through the acquisition, storage, transmission and analysis of audio signals. After testing, the recognition rate of the system for audio types is about 95%, and the positioning accuracy is high in short distance. It will be able to provide auxiliary monitoring and early warning for security in different scenes.

The current network is closely related to life and work, and a smooth network environment provides a good operating environment for the majority of users. However, due to the wanton attacks of network viruses, it has brought great harm to network users. The ARP security problem in the network security is particularly troubled by the users' life and work. This paper discusses the ARP attack problem and defense methods in the network security based on actual work experience.

Data is the most valuable resource of the Internet, attackers often use SQL injection attacks to destroy the database in order to obtain important data information in the database, and today's attack scene is complex, dynamic, multi-channel, non-linear, the existing defense detection technology cannot cope with unknown attacks, the existing instruction set randomization method may be broken by force. Aiming at the above problems, an active defense system of SQL injection attack based on randomization method pool is proposed. The randomization method pool and parallel executor are introduced to build the system framework. The result is decided whether to forward to the database after the decision maker votes, which no longer depends on prior knowledge. The attacker cannot use the system information obtained before to carry out the next effective attack. The formal representation and experimental results show that this method can effectively defend against SQL injection attacks.

The machining error of parts is the result of the coupling action of various error sources, such as the spatial error of machine tools in the processing system. Machining error modeling is to establish the machining error relationship model according to the source of machining error or the measurement result of the machining error to compensate and improve the error. The accuracy of the machine tool itself is difficult to meet the requirements of the product process size. Therefore, it is necessary to carry out targeted process research to improve the processing accuracy and solve the problem of insufficient production capacity. In this paper, the data verification technology of CMM is adopted to preprocess point cloud, such as point cloud elimination, curvature analysis, point cloud division, regression analysis and point cloud regeneration. Firstly, the maximum allowable error position is found in the error model to carry out the subsequent compensation research. Finally, the proposed method is tested in a simulation environment to verify its effectiveness. The research tells that it plays an important role in improving the position accuracy, shape accuracy, surface integrity and consistency of machining.

Based on the theory of GJB5000B and software engineering, this paper puts forward "Software Responsible Person System"(SRPS). The SRPS can better solve the problems in software development processing, control the quality of software development and software development progress. Through the specific application of the system, the practice of software responsible person system and the advantages of the method are explained specifically.

The web system has gradually become the information platform for carrying the internal work of the unit. In order to enhance the monitoring and management ability of the system and ensure the long-term efficient operation of the system, this paper studies the software performance testing method based on Web applications. This paper puts forward a process model of Web software performance testing, clarifies the standard process that should be followed in the process of system performance testing, points out the index parameters that should be paid attention to when testing monitoring and result analysis, and establishes a complete set of testing tools. Under the guidance of the model, the test and preliminary performance tuning of the actual system are completed.

This paper proposes a document printing control method based on virtual components. This method first converts the print files output by various application systems into PDF file format, and then approves the print output of the PDF file. This method realizes the process control of document printing and the security protection of document content and solves the problem of document security printing. In addition, this method can also hide the security identification information used for document source tracking in PDF documents, thus providing an effective technical way for tracing the source of lost secrets due to illegal dissemination of documents.

The implementation of water system function in aircraft involves several systems, and the comprehensive system architecture has high complexity. The number of interfaces between multiple systems is huge and the interface forms are diverse, and the working logic between multiple systems is complex. To handle this complexity, a model-based system architecture design method is proposed. SysML modeling language is applied into the architecture design and behavior analysis of aircraft water system, which can realize the unified definition and decomposition of system architecture elements, the unified interface definition and analysis of multi-systems integrated architecture. At the same time, it also can realize the synthetic design of water system architecture. The architecture model and behavior model of aircraft level and system level are established, and the analysis of water system function and the multi-systems integration design is effectively realized. The practice shows that the architecture design method based on SysML can effectively deal with the complexity of multi-systems integration design and improve efficiency and quality of integrated design.

The mining and exploitation of security vulnerabilities have always been the focus of offensive and defensive confrontations. In recent years, with the application of technologies such as fuzzing in vulnerability mining, the number of discovered vulnerabilities continues to increase, which seriously threatens the security of the network world. With so many vulnerabilities, it is unrealistic to rely solely on manual analysis by security experts, which is not only costly, but also time-consuming, and cannot meet the needs of vulnerability assessment and defense. Therefore, an automated exploit solution for vulnerabilities is urgently needed to solve the problem of low efficiency of manual vulnerability assessment. Existing research has proposed some solutions and achieved certain results. The purpose of this paper is to review and analyze the existing typical research results.

Stack Overflow is an IT technology question and answer website in the computer field. In order to realize the sharing and utilization of knowledge in the question-and-answer website, this paper first uses the Scrapy crawler framework technology to obtain the structural data in the Stack Overflow question and answer website and stores them in the relational model. Then it uses the ontology modeling tool Protégé to build an ontology, and then use the D2RQ tool to achieve the knowledge extraction of the relational database and convert the relational schema into a triplet ontology model. In this paper, we find expert examples from the ontology model by rule inference Experiments show that expert examples extracted from Stack Overflow ontology model can improve the accuracy of API call sequence mining.

Automatic protection is an emerging topic in the field of network security. With the increasingly large scale of computer network system, the network threats we face are becoming more and more serious, and the complexity of network environment is also increasing. Obviously, manual security management is no longer suitable. Therefore, many scholars point out that automatic configuration of security policies can be carried out by building automatic generation system of security policies, and then can manage network devices easily. This paper proposes an automatic policy construction method based on security configuration intention and policy mapping model, including the establishment of standardized security configuration intention database, formal description of security configuration policy, and standardized security configuration intention and policy mapping.

In the process of power transmission and transformation, there are problems such as low efficiency and poor patrol effect in the way of personnel to the site for safety supervision. However, the on-site video system cannot fully monitor the construction process due to factors such as angle and monitoring height. Therefore, the use of Unmanned Aerial Vehicle (UAV) to patrol the whole process of power transmission and transformation project has become the main means of power infrastructure management and control. According to the requirements of the whole process patrol of UAV power transmission and transformation project and the standardization of UAV infrastructure patrol operation process, a UAV management and control platform for power engineering infrastructure is proposed and designed. This paper introduces the overall design scheme and deployment scheme of the system platform, including the network security protection scheme. The system is conducive to improving the safety of infrastructure work and promoting the construction of digital and intelligent infrastructure as a whole.

In order to strengthen the teaching quality of ETL and promote the improvement of students' ESL (English; All of the following with ESL instead) quality, this paper analyzes the feasibility of information technology in college ETL (ETL, All of the following with ETL instead). It also analyzes the application of information technology in college ETL. Explore the application of information technology major in the field of TESOL (English teaching; All of the following with TESOL instead), classroom TESOL, specialized ESL textbooks, writing and examinations in colleges and universities. Explore the application of information technology major in the field of TESOL, TESOL in the course of TESOL, specialized ESL textbooks and examinations in colleges and universities. With the support of high-tech information resources, college TESOL will be combined with new technology resources to create a new pattern of modern TESOL in colleges and universities. It will be a new task for college TESOL. In the context of information age, college ESL education is characterized by multi- culture. Information technology can play an important role in college ETL.

In recent years, deep learning has been applied to the field of malware detection to improve the detection accuracy. However, many malware detection models based on static features and visualization methods do not consider the impact of code confusion, and the detection rate of packed malware is low. In order to solve this problem, this paper proposes a model SE-MHSA combining channel attention and multi head attention mechanism. With ResNet as the backbone network, we combine the SE module and MHSA module to extract channel, local and global features, and weaken the interference caused by code obfuscation by fusing multi-level features, so that the model can capture the correlation information of each partial feature and improve the detection capability of the model. The experiment is carried out on the public data set Malimg, and the accuracy is 99.6%. The experiment is carried out on the packed data set Virushare-Packed, and the accuracy is 95%. Compared with other models, this model achieves better results, and verifies the generalization and anti-confusion ability of the proposed model.

Based on open-source hardware, a set of automatic evaluation system of endurance running items was designed, which aims at the problem that the hand timing of endurance running items is not accurate enough. The system can be applied to the daily physical education training as well as the test every school year and Senior High School Entrance Physical Examination. The system is divided into basic version and test version. The basic version is mainly composed of UNO, LCD screen, voice module, clock module, rotary potentiometer, infrared beam module, etc. The test version is mainly composed of ESP32-CAM and human body infrared sensor. In Arduino IDE, C language program is written to achieve the design and development of the whole system. The system uses variable and array flexibly to record various rating tables and marking patterns, and the system test data is better than manual timing. The whole system is simple and practical. The future improvement direction of the system is to further optimize the system and add the function of automatic circle recording.

Compared with the macrokernel, the microkernel operating system has higher kernel stability and security because the microkernel only includes the key parts of the operating system such as memory management. But once there is a bug with the microkernel, it will affect the entire operating system. In order to improve the security and stability of the microkernel operating system, this paper proposes a fuzzing scheme for the microkernel based on the relationship between system calls. The scheme deeply analyzes the correlation between system calls in the microkernel based on static analysis. Then the generation process of subsequent test cases is guided by the correlation. Finally, this paper conducts experimental analysis on the scheme to verify the correctness and effectiveness of the scheme.

The safety of decision-making in Autonomous Driving Systems (ADSs) is a challenging issue, which is very important for ADS development. As a highly acceptable decision method, Bayesian Network (BN) has attracted more and more attention, therefore, its decision confidence and robustness have also become a focus. Since BN is a form of reasoning based on probability, for the decision results with low confidence (that is, the probability values of several optional decision actions have a small difference), when the data is slightly disturbed, it is likely to change, resulting in serious consequences such as car crash. To generate safe decision-making, we innovatively propose a rational Bayesian network decision-making approach based on BDI model. It helps to improve the decision confidence of the traditional BN. BDI model is a well-known theory of inferencing agents' mental states (belief, desire and intention). We use it to guide the decision-making process of Bayesian network. According to the domain knowledge of ADS, we introduce and design rule-based intention inference for the decision agent to build a Bayesian network with BDI-layer. And for other uncertain agents in the environment, we utilize LSTM model to predict their intentions and provide scenario information for the construction of the above network. To sum up, we propose a rational decision-making approach based on Bayesian network guided by BDI model. Our novel approach makes the traditional Bayesian network decision more humanized and improves the confidence of decision results. Finally, we take the lane-change and overtaking scenario as an example to illustrate our approach in detail and demonstrate the effectiveness in improving decision confidence.

The top oil temperature index of transformer is an important indicator of transformer load capacity and service life. In this paper, the grey prediction model is used to predict the transformer oil temperature initially, and then the autoregressive differential moving average model is used to fit and train the deviation sequence for its prediction deviation. Finally, the optimized prediction model is obtained. The obtained transformer top oil temperature value is divided into training set data and test set data, and divided according to corresponding proportion, so as to achieve accurate prediction of transformer top oil temperature. The experimental results show that the optimized grey-autoregressive differential moving average model prediction algorithm has good prediction effect.

In this paper we first analyze the relevant transmission tower detection systems in China and abroad, understand the pain points and needs in the use of the system, then design an Internet of things-based transmission tower mechanical state monitoring system. The system adopts the hardware design of distributed host-worker machine, through the sensor of the worker or host machine to realize monitoring the tilt angle of each dimension of the pole tower and the climate environment information such as wind speed, temperature and humidity of the surrounding environment, transmitting them to the host machine by Bluetooth, and the host machine performs edge calculation to assess the operational fault risk of the transmission line pole tower using the improved analytic hierarchy process. We can achieve detection and prevention of risks much earlier. Then the host uploads the detected and processed data to Ali-Cloud Internet of things platform through MQTT protocol for data visualization. Risk data visualization enables the staff to grasp the operation of the transmission tower in real time and provide a reference for inspection of the tower operation mechanics state and the basis for inspection.

With the continuous development of digitalization, informatization and intelligence in the discrete manufacturing industry, the importance of data in the discrete manufacturing industry has become more and more prominent. Various aspects of the whole product life cycle, design, development, procurement, production and other processes generate complicated types and huge amounts of raw data. These data are distributed in their respective business departments to guarantee the normal operation of their respective business, record detailed operation, and are continuously generated during the production process. There are often complex logical relationships behind the business data of different departments. How to do a good job of data standardization and promote the sharing of manufacturing data in the whole life cycle is an important part of intelligent manufacturing construction. By investigating the business, data and information environment of enterprises, clarifying the business collaboration relationship among business departments, taking inventory of core data resources, promoting the standardization of data resources, building a framework of discrete manufacturing master data management platform, and supporting the efficient collaboration of data within manufacturing enterprises.

With the rapid development of modern society, higher requirements for the capacity and performance of distributed storage systems are gradually put forward. If the conventional DAS and SAN storage distributed storage methods are still adopted, it will lead to the system being difficult to play its maximum role and affect the user experience. Therefore, in order to meet the needs of modern users, it is necessary to develop a distributed storage system with better performance. The results show that when the system runs, the failure occurs in about ten seconds, and the system recovers automatically after 30 seconds. Data volume was 4kb when random read and write, and 1024kb when consecutive read and write runs occurred. According to the test, the system can still provide a service, both in the fault phase and in the recovery phase. Based on this, this paper designs a high-performance distributed storage system to provide support for further improving virtual machine performance.

Based on the requirements of UN/WP.29 R156, it puts forward the requirements of the whole life cycle software update management system construction for the implementation of software upgrade activities in enterprises, and RXSWIN is a very important point. In order to standardize the RXSWIN management process, a set of universal RXSWIN coding rules is designed, which can not only meet the requirements of enterprises to effectively identify and record RXSWIN related information in the whole process of software upgrade, but also achieve the universality and effectiveness of RXSWIN coding, guide enterprise internal management of RXSWIN to achieve smooth software upgrade process.

As a key component of network applications, data file extraction has become an indispensable part of software development. This paper This paper analyzes the performance and reliability problems involved in the data transmission in the network layer protocol. Combining the advantages and disadvantages of the existing protocols used for data transmission, this paper proposes a data file transmission mechanism based on UDP protocol, that is, the reliable transmission mechanism designed in the protocol layer ensures the data transmission performance.

Finite element analysis software is the main tool for the analysis and verification of complex lug. The commercial software represented by ANSYS occupies the main position in the field of finite element. Based on the open source finite element software SALOME and Code_Aster, using Python language and Pyqt5 framework, a lug simulation system is developed, which has the functions of guide type lug rapid modeling, automatic high-quality mesh generation, automatic acquisition of lug linearization path, automatic evaluation and report issue. The system is completely independent and controllable and replaces the conventional method of manually defining the stress linearization path, which effectively improves the evaluation efficiency of the lug. At the same time, the calculation results are compared with the ANSYS calculation results, and the error is less than 3%, which meets the accuracy requirements.

Container technology can improve the service deployment efficiency and resource utilization of the cloud. However, the popularity of container technology also brings more potential security risks to various container-based cloud service systems. We proposed an intrusion behavior detection system for the container cloud. It first collects the system call traces of the processes in the container from the host or the management system as the training data and makes the augmentation to the data. Then we translate the system call traces to the short sequence mappings and graph structures both as the features of a new deep learning network fused graph neural networks and deep neural networks. Finally, train this model offline and deployment on the contains cloud server for intrusion detection. Evaluation experiments show that the proposed method outperforms all of the compared works significantly, which suggests it has a more accurate and robust performance in intrusion detection.

Abnormal behavior detection of ships is very important to improve navigation efficiency and ensure navigation safety. Most of studies focus on the abnormal behavior detection of ships sailing on a river or sea. However, not much attention is paid to the T-intersection area of the Yangtze River inland waterway. In this paper, a method which uses shore-based surveillance radar data is proposed to detect abnormal behavior of ships in river T-intersection area. In order to realize the identification of illegal behavior of ships in river T-intersection area, we establish the definition model of judgment regions and the detection model of ship behavior. Compared with the traditional method based on AIS data, the proposed method achieves higher accuracy and better engineering adaptability.

Earthquake disasters have had an important impact on people's lives and economy. As the national department for comprehensive response to various disasters, the Emergency Management Department has done a lot of work for earthquake relief. On the basis of the original earthquake disaster rescue and relief, the Emergency Management Department carried out comprehensive rescue and relief work by integrating monitoring and early warning, dispatching and command, analysis, research and judgment, and fire rescue. The Internet of Things, artificial intelligence, video conference, navigation and other means are used to carry out comprehensive rescue and relief work. This paper analyzes the advantages and disadvantages of the comprehensive dispatching system of earthquake relief at the national level, and looks forward to the comprehensive dispatching work of earthquake relief at the national level.

Practical Byzantine Fault Tolerance (PBFT) for alliance chains is limited by high communication cost, which makes it unsuitable for more consensus networks. In view of the limitation of PBFT algorithm, a consensus algorithm of RPBFT is proposed. The RPBFT algorithm introduces RSA (Rivest-Shamir-Adleman) signature on the basis of PBFT algorithm, which not only increases the security, but also solves the high communication problem of PBFT algorithm and reduces the communication complexity to O(n). RPBFT also optimizes the consensus phase, shortens the consensus process and improves the efficiency of consensus. Experimental results show that RPBFT not only can effectively reduce the consensus delay, but also can improve the transaction throughput and scalability, which makes RPBFT more suitable for complex networks.

In order to explore the problem of computer network research, a computer network research based on communication data crossing technology is put forward. Through the key technical problems and solutions of information recommendation based on communication data crossing technology, the research of computer network was explored. The research shows that the efficiency of computer network research based on communication data crossing technology is about 16% higher than that of traditional methods. In view of the correspondence and differences between different layouts of computer networks, through local adjustment of existing layouts, the multi-layer structure and its mapping relationship are clearly displayed under a unified view, which helps users to visually analyze the network communication flow bearing relationship.

Comprehensive evaluation problems can be seen everywhere in life, and there are many comprehensive evaluation methods. Among them, the intermediate truth value measurement method based on fuzzy mathematics is one of the methods to evaluate multi-index objects. It is a method that can effectively deal with fuzzy phenomena, and it is also a subjective weighting method. In order to make the evaluation results more objective, this paper proposes a comprehensive evaluation method based on the combination of entropy theory and intermediate truth measure, which applies the entropy theory to the acquisition of the first-level index weight, and it uses the analytic hierarchy process based on expert scoring to calculate the second-level index weight. This method effectively reduces the error caused by the completely subjective calculation of the weight value in the intermediary truth measurement method. Finally, the method is applied to the quality assessment of residential projects. Compared with the measure of truth scale and the entropy weight method, the proposed method in this paper is more feasible and effective.

With the development of information technology, the online learning and examination system platform of university laboratory safety education based on campus website is difficult to continuously arouse students' enthusiasm and interest in learning due to the untimely information transmission and poor cross-platform performance. Based on the mobile platform, combining the functional advantages of the WeChat official account and mini program, the university laboratory safety education system was researched and constructed. Practical data show that the design of this system is more in line with the current college students' mobile and social learning habits and behavior trends and can substantially improve the effectiveness of laboratory safety education in colleges and universities.

Imaging examination plays an important role in the clinical diagnosis and treatment of diseases. Magnetic Resonance Imaging (MRI) is one of the imaging examinations, which has been widely used in clinic clinical practice. The non-Cartesian Radial sampling method is used to obtain the image data, and it is interpolated to the uniform Cartesian coordinates using the gridding algorithm and then reconstructed. Finally, using the ability of GPU parallel computing to improve the efficiency of the grid, shorten the reconstruction time, and expect to accelerate the image reconstruction speed. The experimental results also show that under the same conditions of other external conditions, compared with the CPU reconstruction, the acceleration ratio of GPU reconstruction has reached more than 12 times.

Evaluating the consumption of students' canteens at school is helpful to accurately identify students with financial difficulties from families. To obtain the canteen consumption data, the daily average consumption of all students and students with difficulties should be calculated as the reference standard, and then the daily difficulty index of each student should be calculated. Accumulate a student's daily difficulty index to obtain the difficulty index, and then convert it into the difficulty degree through feature scaling. According to the scope of the difficulty degree, students are evaluated as poor, medium, and rich. By comparing the consumption evaluation results with the manual evaluation results, we can evaluate the funding compliance of colleges and classes.

Advanced Persistent Threat (APT) is an advanced persistent network attacks on target information systems by attackers. Most existing APT detection methods are implemented by manually designing detection rules based on the analysis of APT attack strategies (e.g., ATT&CK network attack and defense knowledge base). Such methods are difficult to detect unknown APT attacks. In this paper, we propose an attack detection method based on provenance graph semantic analysis, which analyzes sequences from provenance graphs to detect attack events. Firstly, we construct a provenance graph with low complexity and low redundancy based on preserving its semantics. Secondly, we increase the number of attack sequences by mutation to alleviate the problem of data imbalance between normal and attack sequences. Finally, the Natural Language Processing (NLP) model is introduced to realize attack detection by analyzing the semantics of attack sequences. Through our experiments, we found that the method can reduce the scale of the provenance graph by more than 95%. It correctly identifies attack entities with an average of 100% precision and an average of 93.42% recall, as well as identifies attack events with an average of 100% precision and an average of 99.57% recall. The results show that the method proposed in this paper can effectively detect the attack behavior.

The combination of Differential Privacy (DP) technology and Federated Learning (FL) can solve the problem of privacy leakage in FL. However, most of the existing schemes allocate the same privacy budget to clients during training. This also causes the problems of slow convergence speed and low training accuracy of FL. To solve these problems, we propose an adaptively privacy budget allocating scheme, in which clients are assigned different privacy budget according to their contributions. On the one hand, adding less noise on the client with higher convergence contribution value can accelerate the model convergence and improve the accuracy of model training; On the other hand, fast model convergence can reduce the number of training iterations, thus reducing the overall noise added and the communication cost of the model. We experiment on different data sets and network models. The result shows that the scheme proposed in this paper can improve the accuracy of model training and accelerate the convergence under the premise of ensuring the privacy of the model.

Browser fingerprinting has been used as a user-tracking technique in recent years. As a long-term tracking technique, it requires not only obtaining unique browser fingerprints but also linking fingerprints from the same browser instance in that browsers change rapidly and frequently. To improve the efficiency of linking the evolving browser fingerprints, in this paper, we propose a browser fingerprint linking method based on Transformer-encoder. Transformer-encoder utilizes an attention mechanism to focus on certain parts of the input sequence, enabling it to capture complex connections and interactions within the data more efficiently. To make the most of the parallel processing mechanism of the Transformer-encoder, we combine multiple fingerprint comparison vectors into an input vector to train the model. We conduct extensive experiments on a public dataset to evaluate our proposed model. The experimental results show that our model outperforms some existing models, which proves the effectiveness of the Transformer-encoder in linking browser fingerprints.

In recent years, information processing technology based on transfer learning has developed rapidly, and more and more data users do have not enough time to complete data labeling. Domain Adaptation (DA) information processing for unlabeled data is becoming increasingly important. To improve the performance of the model in the target domain information processing, we design a target domain information network and propose a collaborative learning model between the private network and the regional adaptive network based on the TrAdaBoost algorithm. Different from the traditional method, this model can avoid directly reducing the difference between domains as the model optimization goal, and the target is optimized for cluster regularization, driving the target domain data points closer to the cluster center. And further, promote the training of domain information networks. Through simulation experiments, it is concluded that under this model, the target domain private network is effectively trained based on the TrAdaBoost algorithm under the specified target domain, to achieve better information processing performance in the target domain.

Fuzzing is a vulnerability mining approach with high execution speed, but lacks data flow and program state information, resulting in it being difficult to pass complex branching conditions. The FTI (Fuzzing-based taint inference) method proposed by Greyone is lightweight and has faster execution speed and lower execution environment requirements than the traditional taint analysis based on contamination propagation. FTI can obtain critical bytes in the input corresponding to branching conditions and perform mutation for the critical bytes, which can effectively pass complex branching conditions and improve the mutation of fuzz testing. efficiency. In this paper, we propose a fine-grained mutation strategy based on critical bytes. We identify the critical bytes in the input by FTI (Fuzzing taint inference) and execute a fine-grained mutation strategy on these critical bytes, including input corresponding states based on critical bytes, linear search, and random mutation, so that we can pass more branching constraints and eventually improve the coverage rate. Experimental results show that the method in this paper increases the edge coverage by 9% compared to AFL++, effectively improving the ability of fuzzing to pass complex branching conditions.

A new hierarchical communication parallel computing algorithm for transient structural analysis is introduced based on the architecture characteristics of heterogeneous multi-core processors in order to increase the parallel efficiency of the transient analysis of the entire extensive structure system carried out by heterogeneous multi-core processors and distributed memory parallel computers. With the two-layer parallelization of the computational process, this approach remarkably increases the memory access rate through distributed storage of a considerable quantity of data and significantly accelerates the communication rate. By fully utilizing the architectural features of heterogeneous multi-core and distributed memory parallel computers, it is possible to increase the efficiency rates of parallel computing for large-scale transient analysis. In the end, the accuracy and effectiveness of the suggested method were proved using typical numerical experiments. Specifically, a high-rise building's parallel transient analysis was conducted using ten thousand core processors with more than ten million degrees of freedom.

The accurate identification of the grounding ring of the distribution network is the prerequisite for the grounding protection of the live working robot. In order to improve the recognition accuracy, this study improves the YOLOX network. In order to facilitate deployment on the host computer, this method is based on YOLOX-S, replaces the PAFPN feature utilization layer of the neck with a simplified BiFPN network, uses GIoU_Loss as the positioning loss function, and uses Focal Loss as the confidence prediction loss function. Experiments on the grounding ring of the distribution network verify the feasibility of the proposed identification method and meet the working requirements of live working robots.

In order to improve the performance of web service recommendation in small world heterogeneous network, this paper proposes a method based on reputation evaluation. The method defines the calculation method of reputation between users and between users and services, and then proposes a clustering algorithm based on WOA to cluster users and services, which greatly simplifies the calculation of recommendation. Experimental results show that the proposed method is effective in web service recommendation.

An compacted representative signature is a string that persuade the checker to believe that, for all 1 ≤ i ≤ n, the i-th representative signer approved a piece of information. The loss of clandestine keys threatens a security system. We come up with the idea of Identity-Based Key-Preserved Compacted Representative Signature (IBKPCRS) in which representative signing keys vary over time. The usage of IBKPCRS is to meet the key-revealing challenge. In the IBKPCRS system, the user sets his telephone number or other overt string as his public key. We give the IBKPCRS scheme as a teaching case of cryptography.

Temporal modeling has attracted the attention of a large number of researchers in the past few years. In this work, we propose a new video architecture, termed as Temporal Excitation Network (TEN). The core of TEN is Temporal Excitation Module (TEM) block, which consists of Temporal Convolution Module (TCM) and Temporal Difference Module (TDM). TCM applies a channel-wise convolution to supplements the short-range temporal information. TDM works by computing feature-level long-range temporal differences and then exploiting it to excite motion-sensitive channels. These two-stage modeling scheme can be fused into existing 2D CNNs to model temporal structures flexibly and efficiently. Extensive experiments demonstrate the effectiveness of the proposed TEN on several benchmarks (e.g., UCF101, HMDB51, Something-Something V1 and Jester). The proposed TEN can guarantee high recognition accuracy while maintaining high recognition efficiency on these datasets.

The existing hash functions based on quantum walk construction are all proposed under the ideal closed environment. In order to better conform to the actual situation and improve the relevant performance of Hash functions, we propose a quantum hash function constructed by continuous quantum walk on the basis of broken-line noise model. We first propose a continuous-time quantum walk model on cycle under the influence of broken-line decoherence, after that apply this model to the construction process of the quantum hash function and put forward a building method of the hash function on the basis of the broken-line decoherent continuous-time quantum walk on the one-dimensional cycle. The safety analysis and numerical simulation indicate that our hash function has greater anti-collision performance than the existing hash function based on the quantum walk.

Distributed Denial of Service (DDoS) is a huge hazard to Software-Defined Networks (SDN). Active defense technology is one of the effective measures to ensure the security of SDN. Active defense can increase the difficulty of the attacker's attack and reduce the probability of being attacked successfully. However, the active defense method based on port hopping has the problems of fixed hopping strategy, lack of flexibility and poor security (for example, it is easy for an attacker to grasp the law of server port hopping). Aiming at these problems, we proposed a Dynamic Moving Target Defense method based on Adaptive Port Hopping (DMTD-APH). The DMTD-APH combines the characteristics of SDN on the basis of port hopping and improves the flexibility of active defense by designing strategies such as hopping synchronization, hopping and forwarding, and adaptive hopping. At the same time, the DMTD-APH dynamically detects the network status through the source address entropy value and data flow rate method and performs time-adaptive or space-adaptive hopping adjustments to ports according to the detection results to build an adaptive active network defense model. The experimental results show that DMTD-APH enhances the anti-attack and service type of the network, and has stronger dynamics and security.

The information network of the power grid enterprise is the same as the information system of the common Internet, and its application layer has a lot of general IT software. In order to mine all kinds of software vulnerabilities timely and ensure the normal operation of the software system, the most obvious disadvantage of this method is the low efficiency of implementation, which requires manual dynamic testing after the end of software development. It also needs to track the location of vulnerabilities according to the test results. In this paper, the cooperative control of heterogeneous wireless networked robots based on parallel control is proposed, and the configuration of defense resources is studied from the perspective of protection. Firstly, based on the network security robot, the game model of both sides of the power grid attack and defense under coordinated attack is established, and the optimal defense resource allocation strategy is analyzed and solved. Then, for heterogeneous wireless networks, a step-by-step solution is proposed based on parallel control optimization, and the allocation method of defense resources is formulated. Finally, the proposed method is verified on the simulation test system and compared with AUKF and IMM-UKF. The experimental results verify that the proposed method performs well in error control when the execution state changes, and achieves high accuracy, good stability and strong security prediction ability in general. It can ensure the safety of power grid and promote the healthy development of the national power industry.

There are many problems in the geological laboratory, such as complex business processes, scattered storage of experimental data, difficult to retrieve and process experimental data, and low security of experimental data. The working mode based on manual management can't meet the needs of the geological laboratory. Based on the existing experimental process, the experimental data of the existing experimental projects are associated with the corresponding sample information, and the data are stored according to the data logic. The functional modules and system architecture were designed. Research and development of geological LIMS based on microservice architecture. It provides sample information registration, sample distribution, test report upload and management, sub-sample management, data integration and comprehensive display, system management and other functions. The system realizes the information management of production processes and data during the operation of geological laboratories, the information management of resources such as personnel, instruments and equipment, test items and test reports for the daily operation of laboratories, and the information management of the quality management system and daily information of laboratories. Compared with the traditional manual management mode, the system based on microservice architecture is easy to deploy, maintain and upgrade, provides a more convenient management mode and a more powerful fault-tolerant mechanism, improves the data management ability and work efficiency of laboratory personnel, reduces the laboratory operation cost, and improves the informatization construction level of geological laboratory.