This PDF file contains the front matter associated with SPIE Proceedings Volume 12453, including the Title Page, Copyright information, Table of Contents, and Committee Page.

After data preprocessing, the feature dimension of NSL-KDD dataset increases from 42 dimensions to 122 dimensions. High dimensional data will make it more difficult for the model to learn the characteristics of the data, and there will be a lot of redundant data in the data set. Therefore, this paper uses the deep belief network to reduce the dimension of the characteristics of the intrusion detection data set after data preprocessing, and uses the stacking algorithm as the classifier to construct the intrusion detection model. Through comparative experiments, it is proved that the model has good performance in the four evaluation indexes of accuracy, precision, recall and F1 score, and effectively improves the performance of intrusion detection model.

With the application of information technology in the field of electric power, the emergence of smart meters(SM) has made the collection of electricity information data faster and more convenient, and its collection, processing and analysis of power data has become more intelligent and automated. The SM data acquisition system under the information technology can effectively reduce the input of manpower and material resources, monitor power data online, and feed back to relevant staff as soon as a problematic power meter is found, so that the power meter can be replaced quickly. SM data acquisition system online monitoring, data processing and analysis to quickly perform inaccurate replacement behavior can improve the user's power meter management level, improve the quality of power service, and improve the accuracy of meter measurement. The main research of this paper is the analysis of the collection system integrating the inaccurate replacement under the information technology, the analysis of the composition of the SM collection system, and the monitoring of the user's electricity consumption data. Next, the necessity of inaccurate replacement is explained, and the function analysis of the acquisition system that incorporates inaccurate replacement is carried out. This paper studies the data error of the electricity meter collection system, analyzes the current and voltage integration, and analyzes the electricity consumption data of the area where the SM collection system integrated into the inaccurate replacement is replaced. The research results show that after using the smart meter with inaccurate replacement, the abnormal situation of user power consumption data in this area gradually decreases. When the total number of lines is the same, the qualified rate of power consumption data gradually increases, from 95.89% in January to 99.32% in June.

Based on the semiconductor equipment data acquisition standard, this paper analyzes the communication method between the fabrication client applications and the equipments. The common equipment model and the method of self-description of equipment nodes based on metadata are firstly studied in this paper, and then we research the security authentication and authorization mechanism between equipments and fabrication client applications. A method for client acquisiting semiconductor equipments data timely and flexible is given by SEMI EDA, which improves the ability of fabrication client applications to collect equipment data.

An intelligent evaluation method of power information intrusion tolerance based on machine learning is proposed to solve the problems of selection of evaluation indexes, poor evaluation accuracy and efficiency in the evaluation method of power information intrusion tolerance. An intelligent evaluation system of power information intrusion tolerance is established, and an intelligent evaluation model is established by using random forest algorithm. The random forest algorithm determines the evaluation weight according to the quantitative value of the index, sets the value of the intrusion risk function, and obtains the intelligent evaluation results of intrusion tolerance. The method studied is applied to the comparative evaluation experiment of the survivability situation of the power information physical system. The evaluation mean square deviation of the intelligent evaluation method based on machine learning is less than 0.1, the average time is 140.72ms, and the evaluation efficiency is increased by 42.51%. At 1000min, the reliability value is still 0.46, which has practical significance.

To solve the problem of fusion scheduling of relay protection device functional safety and information security, the measures of relay protection device functional safety and information security are selected based on the analysis of relay protection device safety requirements. The priority of each measure is analyzed and a priority-based fusion scheduling scheme of functional safety and information security is proposed. On the STM32F407ZG platform, the safety tasks are managed using the μC/ OS-II embedded operating system, and the safety tasks can be executed correctly.

Among the many network protocols, the protocol responsible for obtaining the Mac address based on the IP address is called the Address Resolution Protocol (ARP). Since the ARP protocol is a stateless protocol and lacks an authentication mechanism for the data source of the request/response signal, any client can forge malicious ARP packets to poison the ARP cache table of the target host. Attacks based on ARP spoofing are very harmful, and will lead to reduced network transmission efficiency, network congestion, and even user information theft and privacy leakage. This paper uses wireshark software to view network traffic, and analyzes the implementation principles and characteristics of man-in-the-middle attacks (MITM), denial of service attacks (DoS), and MAC flooding based on ARP spoofing. A forensic method for ARP spoofing attack based on network data flow analysis is proposed.

The so-called law university is the establishment of an educational institution that trains high-quality students, where everyone's level is a law graduate student with a higher education. Based on the talents cultivated by legal education, and supplemented by the judicial examination and judicial training system, a "pipeline" process of preparation for employment is formed. The legal network education is mainly based on computer technology and multimedia technology, and adopts advanced multimedia intelligent technology to provide a new education method for a wide range of learners. Online education not only moves offline classroom courses to online courses, but also brings questions between libraries, training rooms, and students directly to our desktop, providing us with a convenient way. A modern education with initiative, creativity and novelty. The website education website is based on the Web, it can not only help us answer questions, but also conduct related discussions, solving the problem of informal face-to-face communication with students alone in online education. Therefore, the combination of Web and database technology , which applies a new type of non-dynamic education service system, which is an important prerequisite for the development of education websites.

Power information communication contains sensitive and private data, which may be illegally stolen and tampered after being leaked, posing a threat to the data security of power enterprises and users. Aiming at the problem of limited key space, a data leakage prevention method for power information communication based on chaotic mapping is proposed. The physical layer of power communication includes power data produced and distributed by power equipment, and the information layer includes operation, monitoring and dispatching data. There is an association between the two data. The dispatch and control center uses the association relationship to establish access rights for users and generate access private keys for trusted users. According to the user authority, the communication data is encrypted and decrypted by chaotic mapping, so that the two adjacent components of the reconstructed key space are relatively independent, and the properties of the chaotic sequence are maintained. The test results show that the proposed method can effectively reduce the computing and communication overhead of the power data encryption process and improve the encryption efficiency.

With the rapid development of information and communication technology, network attacks and sabotage methods emerge one after another, and the internal and external risks facing the critical information infrastructure in the communication industry are becoming increasingly prominent. In order to improve the effectiveness and pertinence of the security protection of critical information infrastructure, the security protection path of critical information infrastructure is proposed. Through the analysis of structural equation model (SEM), it is confirmed that the identification factors of critical information infrastructure in the communication industry include three kinds of factors: importance, harmfulness and relevance. It also points out the influence degree of each factor on the critical degree value and each secondary index on the three factors, which has a certain reference significance for the effective identification of the critical information infrastructure in the communication industry.

Botnets widely use DGA (Domain Generation Algorithm) technology to evade network security detection, and DGA malicious domain name detection has attracted much attention. Aiming at the problem that poor feature extraction effect and low detection accuracy of existing domain name detection methods, this paper proposes a hybrid neural network model based on CNN-LSTM. The model first uses multi-channel Convolutional Neural Network (CNN) to extract the NGram features of domain names; then uses Long Short-Term Memory (LSTM) to extract the contextual grammar features of domain names; finally introduces the attention mechanism to assign different weights for the extracted domain name features, focusing on more critical information. The experiment results illustrate the proposed model maintains an Accuracy of 99.02% in malicious domain name detection, which can obtain higher detection accuracy than the existing domain name detection model.

With the increase of network demand and the complexity of network environment, malicious attacks on network devices emerge in an endless stream. Compared with ordinary networks, edge networks have more limited resources and face more complex network environment. Edge networks need more efficient and accurate intrusion detection systems to ensure the security of edge networks. According to the characteristics of edge networks, an intrusion detection method based on multicenter incremental clustering is proposed in this paper. The method first uses the DBSCAN algorithm to cluster the initial data set. At the same time, the concept of multi-cluster centers is proposed. For a cluster with large number of samples and irregular distribution, the characteristics of the cluster can be described by multiple cluster centers. For the new incremental data, we determine the class of the new data by the location of the cluster center and the number of core sample points around it. The multi-center incremental clustering algorithm not only reduces the time of clustering new data, but also can effectively detect unknown network intrusion, which improves the efficiency and accuracy of network intrusion detection.

Starting from the research background and significance of the subject, this paper analyzes the data characteristics of smart grid and some data security problems it faces, analyzes the current research status of data confidentiality and integrity, and clarifies the practical significance and necessity of data security for the safe operation of smart grid. This paper mainly completes the following work: first, the shortcomings of the common hybrid encryption algorithm are analyzed, and the hybrid encryption algorithm is improved. Experimental verification and analysis are carried out to show that this method can improve the data Security. Secondly, through the analysis of the existing integrity verification methods and combined with the characteristics of smart grid, the improvement of the integrity verification method is completed. The experimental results show that this method can improve the accuracy and efficiency of verification. Finally, the smart grid simulation experiment is carried out to test the application of the improved method. The experiment shows that the improved method can improve the security of smart grid data storage to a certain extent.

Cyber attacks are often costly, and attack models can better help researchers analyze complex attack behavior. Traditional threat models focus on the logical description of attacks, thus neglecting other critical factors in attacks. Therefore, this paper proposes a multi-factor coordinated formal description method for network attacks. This approach can provide a more comprehensive description by combining network environment, attack nodes, target assets, and collaborative description of attack steps. First, this paper proposes a description method that approximates the network topology to similarly describe the complex network environment and attack nodes. Subsequently, MAL and ATT&CK matrices are used to describe the target assets and attack logic. The description methods from the two perspectives are combined to form a multi-factor coordinated formal description method. Finally, it is validated using actual attack cases. The method proposed in this paper can describe attack behavior more comprehensively.

Multi-variant Execution (MVX) technology can effectively resist various types of memory leak attacks. Since it was proposed in 2006, after years of development, as an active defense method, it is widely used in cyberspace security. in defense. The multi-variant execution technology distributes the program input to multiple functions through distribution agents, executes executions with different structures in parallel, sets up voting points, and detects whether it is attacked by comparing the output results of the executions of the executions by the monitor. When the multi-variant execution technology is applied to the actual scene, there will be some wrong judgments. For example, the execution body is not attacked, but because the execution bodies are isolated from each other and executed in parallel, random numbers will be generated when the execution body executes the program. These random numbers are included in the output results and are sent to the monitor for voting. Based on the consistency judgment voting, the monitor compares the output results to determine that the execution results are inconsistent, and the misjudgment system is attacked.A multi-variant system is modified by multi-variant execution technological. Such misjudgments are called misjudgment of random numbers. This paper summarizes and analyzes the voting misjudgment caused by random numbers in the multi-variant system, points out the reasons for the misjudgment of random numbers, lists common use cases, summarizes the previous solutions and ideas, and proposes a method to use the file system to synchronize random numbers. Experiments have shown that this method can effectively reduce the misjudgment of random numbers in multi-variant systems and improve the availability of multi-variant systems.

Metering asset equipment is an important basis for building a smart grid, and is mainly used for power metering at power supply terminals. Due to the big difference between the inventory management requirements of metering asset equipment and other materials of power supply enterprises, for example, it needs to be stored in special storage areas, warehouses and storage spaces. Therefore, it is usually necessary to use a dedicated inventory business for management. The main purpose of this paper is to optimize the intelligent supply chain of metering materials based on data mining technology. This paper mainly analyzes the organizational performance of the smart supply chain and the differences in metering materials. During the 10 iterations of the experiment, the maximum and average response times of the system gradually increased with the number of concurrent users. 1.965s and 1.224s, the maximum response time of which did not exceed the preset 3s target value, so the performance test passed.

Based on the current situation and problem analysis of the construction of new engineering disciplines in cyberspace security, this paper proposes a connotative construction of professional construction quality oriented to "collaboration", with the "third-level specialty certification" as the standard and two "Shuang Wan plan". To establish a "1+2+1" talent training model to promote the implementation of the construction of the teaching staff and curriculum system, and explore the new way of new engineering construction of diversified cyberspace security such as "Internet engineering + network technology security" and "network technology security + cyber culture security".

Federated learning (FL) enables decentralized data sources like mobile phones to joint training a neural network model without sharing the original data. However, shared local gradients make the privacy of local data in FL vulnerable. The aggregation server also may return incorrect results to clients due to unexpected error or the deliberately attack. In this work, we explore how to design a non-interactive and publicly verifiable aggregation scheme. The existing verifiable schemes are under semi-honest adversary model, in which the server is honest-but-curious but with additional power to counterfeit the aggregation result. We propose a scheme under stronger security model against malicious servers. The proposed scheme guarantees that as long as the two servers are non-colluding, even a malicious server cannot obtain input privacy of client. The malicious server will be detected by honest clients when it tries to tamper the result.

Data mining is a method of using massive data and using algorithms to retrieve massive data to discover hidden, previously unknown, and potentially meaningful information. At present, in many aspects, the application of data mining technology has been deeply discussed. The teaching system is a complex system that is combined and constrained by many factors. The evaluation of scientific systems is a high-quality teaching activity, therefore, evaluation must be carried out at every stage of education. In the systematic education of social people, schools have played a great role in education, taking classrooms as the main body. The scientific and standardized teaching behavior in the classroom will directly affect the quality of education. As the carrier of the overall education system, colleges and universities are an important node and carrier for college students to step into the society; and colleges and universities are colleges with the highest academic standards and scientific and cultural standards. Therefore, it is very necessary to evaluate and analyze college English classroom teaching.

The outbreak of the new crown pneumonia on a global scale has caused serious damage to every country, whether financial or human, and the death toll has also increased significantly. This enhances the importance of laboratory biosafety management, which is related to human life safety and should not be underestimated. In the past, biosafety has not received much attention, resulting in the biosafety management status of biosafety laboratories in my country is not optimistic. On the one hand, it is necessary to cope with reducing the number of people infected with pneumonia. On the other hand, there is an urgent need to match drugs against the new crown. Time is very short, and the number of infected people will increase rapidly according to time. Since viral nucleic acid testing is toxic, it has been reported in our country that many doctors have been infected with the new coronavirus, which has dealt a heavy blow to laboratory biosafety and the people of the country. Immediately afterwards, many medical universities were also exposed to lax safety management, lack of laboratory safety principles, and low safety factor for teachers and students. With the development of life safety and biotechnology, efficient implementation should strengthen research on biotechnology and management safety. The state also hopes that they can make rectifications. This also gives great trust to the management of biosafety laboratories in colleges and universities. I hope they can do their best. correct.

The rapid development and wide application of computer technology have promoted the development of all walks of life and played a positive role in promoting the construction of the government in China. As an important participant, manager and service provider of a country, the enterprise management system plays an important role in the process of informatization in our country. The automation and standardized management of enterprise management business, and give full play to the Internet and the existing information foundation, effectively carry out effective communication and cooperation to meet the needs of modern economic and government development. In accordance with the instructions of the national industry and commerce department, an online electronic information platform for enterprise operation and management shall be established. Enterprise management is a collection of enterprise name verification, private operation, domestic capital management, foreign capital management, trademark management, advertising supervision, case handling, enterprise early warning, credit file, case management, contract management, daily management, and fee management.On this basis, the specific design of several main business units, such as enterprise registration, daily management, and case management, is proposed.

With the rapid development and popularization of computer technology, the status of computer network in China has become increasingly prominent, and at the same time, it has become a society that needs computer network technology talents. Therefore, colleges and universities must vigorously cultivate and cultivate computer network technology talents. The training of computer network technology requires students to carry out a large number of practical applications, but at present, due to the limitation of hardware conditions and opening hours, many schools cannot realize the training of network technology talents. In order to provide learners with a good learning environment, many network equipment manufacturers and software companies have developed a variety of simulation software, allowing students to simulate at any time and anywhere, thus reducing the dependence on hardware devices.

With the advent of the era of big data, in order to actively cope with the massive and unstructured "information overload", recommendation systems are widely used in network products and have a profound impact on people's lives. At present, the research of recommendation system has been developed for more than ten years, but many of them are still in the incomplete stage, there are problems such as sparse data, insufficient mining of user implicit preferences, and inaccurate semantic description of item features. These pictures, videos and other content are full of users' emotions, which contain a lot of user emotions, which need further Excavation and application, resulting in huge social and economic benefits. However, there are few researches on sentiment analysis of this kind of information, and the multi-modal sentiment analysis of text, images and other information is particularly weak. Multimodal information has the same emotional characteristics and can describe the user's emotional changes from different angles, so as to obtain more accurate emotional identification.

Based on three tier heterogeneous network, we examine the problem of energy consumption optimization with latency constraints, establishing a hierarchical computing model with multiple layers under heterogeneous network, each layer representing its individual computing model. An offloading strategy will be optimized to reduce energy consumption while ensuring latency constraints. Lyapunov optimization theory will be utilized along with a heuristic algorithm called Coral Reef Optimization (CRO) to analyze the current workload and available computing capacity of edge and cloud servers, making an efficient compute offloading choice reduces the energy used by cloud servers while still adhering to latency restrictions.. Simulation result shows that compared to cloud-only and edge-ward strategies, the proposed strategy shows improvements in term of energy reduction as well as execution time.

With the development of the global Internet industry, the demand for the transmission capacity of communication system is increasing rapidly. Optical fiber has become the main transmission media in communication systems because of its unique advantages, such as large bandwidth and low loss. It is necessary to construct large capacity optical communication systems to satisfy the capacity requirements. In this paper, the overall architecture for optical short-reach pulse amplitude modulation (PAM) transmission system is analysed. Then the signal damage factors and corresponding compensation schemes based on digital signal processing (DSP) are studied. The different implementation schemes for clock recovery and channel equalization are described and compared in detail. Finally, the challenges and prospects of DSP technology in short-reach PAM transmission system are discussed.

Rough set theory is an efficient mathematical tool that can be used in real-life data compression, data correlation, similarity and difference. Through the induction and arrangement of attribute reduction, its general characteristics and characteristics are obtained, and they are used in the generalization mode of typical rough sets, thereby expanding its application field. However, rough sets have certain limitations, such as being sensitive to noise, so combining rough sets with other intelligent algorithms is a current research hotspot. Neural network is a complex network composed of a large number of interconnected neurons. This paper mainly introduces the method of combining rough set and neural network. Based on the properties of rough sets and neural networks, many methods to achieve an ensemble have been proposed. On this basis, a neural network model based on rough set technology is proposed.

The development of electronic computer technology has its origin in electronic music, which provides a certain support on the basis of creation. It can have a rich knowledge base and software usage functions to meet the needs of the outside world, and its advantages allow the system to show further development. For some time, due to interface problems between systems, software repairmen and electronic musicians have developed in unison to the good, and have satisfied the continuous improvement of electronic music and the interesting soul of notes. This reflects a different phenomenon in electronic music, that is, the collision of hardware and software creates a fierce spark, the difficulty of development is getting more and more difficult, and the price is rising. For many users, these very complex systems for them are not all of their own can be used, some functions may not be useful, but users need to pay a lot for these idle functions, for some users, complex systems are mostly less flexible, can not meet the special field of artistic operations and their needs. Based on the computer network era, there are network communication technologies and interface technologies to operate in a comprehensive manner; The absolute factor of the operation function makes it possible for different operations to be related to each other and the affirmation of the function, which can be more simply communicated between data, and the obstacle can be erased under sharing. The calculation results of different hardware systems are also different, and they also consume a considerable cost to achieve, because these technologies, we will use many different systems to implement the protocol between the various platforms, etc., so that we can more easily build the effectiveness of the electronic music creation platform. This paper attempts to discuss and study from many aspects of the system's software, system creation platform, communication protocol, etc., and has used to find out the uncertainty of different hardware and software to facilitate the establishment of electronic music multi-system platforms.

The establishment of the public legal service big data platform is to promote our country's laws to better become people's right to safeguard their own rights, and is the goal of promoting the work of the legal industry towards a new era and modernization. It has the ability to better change the development of my country's public legal services and the efficiency of people's livelihood, and improve citizens' ability to find powerful tools for themselves in the society. Further, it can provide the masses with more intelligent and meaningful legal services, make the law closer to the masses, and improve everyone's cognition of the law and its perception of its value and system. In order to carry out a series of public legal services to the masses, in the face of the gradual increase of public opinion on the Internet, it has become the fuel of discussion among the masses. For this reason, the legal service platform provides the composition of system modules and related technologies to jointly improve.

The traditional word vector model cannot generate word vectors with polysemous features. The paper uses the BERT model to obtain word vectors, analyzes the modeling ability of the model in text sentiment analysis, and improves the Bert output structure. In order to obtain more text vector information, the vector output structure of BERT model is improved, the original single-layer output is replaced with multi-layers hidden state output, and the hidden information of these different hidden layers is fused and embedded into the downstream task to perform text analyze. The results show that the method proposed in this paper has a certain improvement in the text classification task.

With the increase of data volume and the shortage of frequency band resources in sub6 band, it is difficult to meet the communication requirements for the existing mobile communication systems. It is worth noting that, the 5G millimeter wave communication technology has the advantages of fast speed, large bandwidth and low delay, which can meet the demands of future mobile communication. How to generate millimeter wave with high quality is the key technology. In this paper, 5G millimeter wave communication systems and several all-optical millimeter wave generation technologies are discussed, including direct modulation method, optical heterodyne method, frequency up-conversion method, external modulation method and four-wave mixing (FWM) method. Furthermore, the advantages and limitations of different methods are compared and analyzed. In addition, the promising future applications of 5G millimeter wave communication systems are discussed, and the potential application value of 5G millimeter wave communication systems in digital twin, mobile edge computing and other fields is summarized.

At present, fingerprint authentication is one of the main authentication methods on terminals, but its security level is not enough. SIM-based authentication services have high security level and are being used more and more widely. However, SIM card authentication requires users to input PIN code, which affects user experience to some extent. In this paper, the fingerprint authentication technology, SIM card and SIM with fingerprint technologies are studied, and this paper proposes an authentication framework which combines SIM card and fingerprint authentication technology, this framework greatly enhance the user experience of SIM card authentication, at the same time it also increased the level of security for fingerprint authentication, which can be applied in all kinds of application scenarios with high level security requirements.

With the increase of carrier frequency, bandwidth and antenna size of communication system, the communication system and radar system are gradually approaching. Integration of communication and sensing technology has become a research hotspot in recent years, and can be applied to many fields such as UAV low altitude airspace monitoring and traffic management. This paper provides a comprehensive overview of the latest technologies in the integration of sensing and communication. For different types of integration schemes, the classification and improvement details are given. This paper refines the development and application of communication and sensing in UAV, and provides abundant resources on methods, including the overview of advanced algorithms, systems and applications. This paper can be used as a hands-on guide for understanding, using, and developing methods. This paper comprehensively considers the difficulties in the field of unmanned aerial vehicles, summarizes the challenges faced by the integration of sensing and communication technology, and points out the development direction of this rapid development field.

With the rapid development of science and technology, the Influence of the Internet on all walks of life in today's society is increasing. All walks of life are using advanced Internet technology to achieve product innovation and upgrading, to find a new way for the sustainable development of the industry. The Internet is playing an increasingly important role in China's education sector. The teaching mode under the network environment represents the latest technology in practice teaching gradually infiltrated and applied teaching, with the support of advanced network technology, education and teaching work has also had new development, vocal music classification research has made great progress. This paper mainly describes the research of vocal music classification based on convolutional neural network, and expounds the research status of vocal music classification. Under the network environment, vocal music teaching has gained more space for development, and more teaching forms gradually appear in the actual teaching work. Therefore, vocal music teachers can be combined with their own actual situation for bold attempt and innovation, try more content. Its purpose is to use the technical power of the Internet to provide students with better vocal music teaching services, so as to better cultivate students' music literacy.

As some power companies have become benchmark companies in the power industry, actively responding to the State Grid Corporation’s innovative regulatory policies, setting up activity tracking data centers, conducting online surveys on key indicators, key resources, and industries, and controlling and analyzing company network activities. The company’s overall operating efficiency and profitability. However, the construction of power company activity monitoring data centers focuses more on data applications and does not pay enough attention to data management. Therefore, it faces many data management issues that directly affect data analysis and implementation. This article first clarifies the background, purpose and importance of the research materials. Next, it describes and analyzes the current status of data management in power companies, observes the operation of power data according to the unbalanced data classification algorithm of power data management, and produces power companies Data management flowchart. Then, this article deeply researches the accuracy of data operation, the degree of execution, and the satisfaction of related staff in the power company data management (data storage part, data access part, data security access management part). It also describes the application principles, processes and steps of the solution, and shows the results of the application. Finally, a relatively complete summary of big data management practices in the power industry will lay a solid foundation for the development of big data in the power industry and provide a reference and theoretical basis for the further extraction and implementation of big data.

The development of modern technology has made many enterprises affected by the network. Of course, the accounting profession has also suffered changes, so the accounting information system has set strict requirements on itself. In the intelligent accounting information system, even if the non-automatic accounting system can already achieve intelligent use, but with the construction of accounting, there is no way to practice in many aspects. As a result, the relationship between enterprises and external social groups is not close. When defects arise, they do not think of other ways to solve them, resulting in the value of information technology itself not being brought into play. Because of this, an in-depth discussion of the construction of an accounting information system will help the reform of enterprise accounting. Sui Chunlei proposed that big data technology can improve the relevance and accuracy of management enterprises, integrate information technology into enterprises, apply to accounting management information, and gradually improve its management requirements, while achieving the extensiveness and timeliness of information strengthening. Wu Zhongsheng clearly mentioned the development process of accounting informatization and planned the development prospects of accounting informatization. On the basis of this system, the computer system designed the accounting information system for it. The construction of computer network accounting information system is to sort out the collection and application of big data and data, in order to better realize the correlation between the network and reality, on the basis of the traditional accounting information system, is conducive to better development, will be the function of the data, the module of the system, the comparative results of the test to analyze separately, the purpose is to explore the computer accounting information system to study, the result will be the need to consider the node, convenient computer network on the accounting information system design and research.

By quantitatively describing and analyzing the changing laws of opinions in knowledge collaboration, in order to design more scientific and reasonable knowledge collaboration rules. Considering the differences of individuals and the complexity of collaboration, the uncertainty of individual opinion selection is difficult to represent by a completely deterministic function or set of functions. The individual opinion selection probabilities are considered as a time series, and a suitable ARIMA model is built to fit and short-term expectation of individual opinion selection probabilities. By organizing an experiment to simulate the knowledge collaboration process, an eye tracker is introduced to collect information on individual eye movements in order to obtain real-time changes in the opinions held by individuals, and twelve sets of individual opinion selection probability sequences were obtained and modeled separately. The empirical results show that the ARIMA model can well simulate and predict the evolutionary trend of individual opinions. This method can effectively overcome the difficulties that the influencing factors of individual opinion changes are difficult to fully grasp and data are not easily obtained in the process of knowledge collaboration.

Quantum signature is a classical digital signature technology, which realizes the security guarantee of the integrity, nonrepudiation and non-tampering of quantum information through the basic principles of quantum mechanics. Arbitrated quantum signature (AQS) is a hot spot in current quantum signature research. The article deeply discusses the basic principle of AQS system, and takes the research and design of AQS scheme as the main idea, researches and analyzes the existing AQS scheme, finds the existing problems, and gives solutions. At the same time, in order to further improve the arbitration function of AQS, this paper proposes a flexible tracking quantum group blind signature system based on AQGBS.

With the rapid development of drones, unmanned vehicles and robotics industries, VLAM has become a hot technology. In particular, the birth of 5G-powered UAV has promoted the emergence of more industrial applications, making it the most core and indispensable role in many scenarios. The loop closure detection can decrease the accumulative total of error during the process of VSLAM. Former loop closure detection methods always rely on artificially features, which are not robust, making it hard to deal with changing complex scenarios. The later deep learning-based methods are considered to be better solutions for loop closure detection. However, due to the simple network structure, there is still a lot of room for improvement. This paper proposes a more complex neural network to achieve loop closure detection. This approach adopts a fish-shaped deep neural network backbone, which can extract and fuse data features at different levels. Experiments demonstrate the feasibility and effectiveness of this backbone in loop closure detection problems.

A more important part of the field of deep reinforcement learning is the study of multi-agents, for the specific scenario of multi-cognitive networks, the choice of spectrum will be affected by two parts, a single cognitive network and the access device under the cognitive network. In view of this specific problem, this paper uses the improved multi-agent reinforcement learning to solve, through the use of multiple agents from the user link modeling, can solve the problem of different cognitive networks for action execution when the environmental state is constantly updated, compared with the original algorithm. In the scenario where spectrum allocation is required in multi-cognitive networks, the improved algorithm can better handle the relationship between master and slave users in multiple networks, so that the spectrum utilization and the overall communication performance of the system are further improved.

To solve the problems of computing in decoding and cheating by a distributor or a participant of grayscale visual cryptography (VC), grayscale visual cryptography based on cheating prevention was proposed. We chose two grayscale images; one of the two images was a secret image and the other was a verification image. In the encoder, the input grayscale images were transformed into multitone images via multitoning first. Based on the unequal relationship between dots per inch (DPI) and pixels per inch (PPI), we got a group of sub-modes that contained the different numbers of print points. Two multitone images were transformed into two groups of bit-planes via sub-modes. For each secret bit-plane, we operated (k, n) -VC encoding and got secret bit-plane shares; for each verification bit-plane, we operated (2, n+1) -VC encoding and got verification bit-plane shares. To prevent cheating, we randomly chose n verification bit-plane shares and connected them with n secret bit-plane shares to obtain n composite secret bit-plane shares. The remaining verification bit-plane share was connected with a blank image of the same size as the secret bit-plane share to obtain a composite verification share. In cheating prevention, a participant can judge the authenticity of the secret bit-plane shares held in his hand through the cheating prevention module. In the decoder, we operated OR operations on composite secret bit-plane shares of the same level and obtained reconstructed bit-planes. Only the reconstructed bit-planes were stacked to obtain the original secret image. Experiments prove the feasibility and advantages of this scheme.

Distributed Denial of Service (DDoS) attacks are one of the dangers of Software Defined Networking (SDN). In order to detect DDoS attacks quickly and accurately, this paper proposes a two-stage DDoS attacks detection method based on SDN, which combines a preliminary detection method based on information entropy and K-Nearest Neighbors (KNN) regression algorithm, and a depth detection method based on Gated Recurrent Unit (GRU). First, the two-stage detection method extracts their necessary data features from the flow table information respectively through the feature extraction module. Second, the preliminary detection module uses the information entropy of the six-tuple as feature vectors to train the KNN regression model. Third, the depth detection module uses the GRU neural network to fully learn the sequence features. Finally, the method proposed in this paper is verified in the experimental environment based on Mininet. When the first-stage preliminary detection module determines that the SDN network environment is suspected of being attacked by DDoS, the second-stage depth detection module is called for further detection. The experimental results demonstrate that the method proposed in this paper can effectively detect DDoS attacks in SDN.

As a new distributed database technology, consensus algorithm is the main difference between blockchain and traditional database technology. Consensus algorithm decides the safety and performance of blockchain technology. PBFT(Practical Byzantine Fault Tolerance) has the problem that the main node is relatively fixed, and when network scale is large, communication cost is unacceptable. To solve this problem, this paper proposes an improved PBFT consensus based on quality of service. By dynamically evaluating the service quality of nodes to divide the participation status of nodes in the blockchain network. Nodes with high service quality will have a greater chance to participate in the consensus, reducing the probability of low service quality and malicious nodes becoming block producers, thereby improving the security of the system; The FTS tree is used in the eligible nodes to randomly determine the block- producing node to ensure that the block-producing node will not be fixed for a long time; A node state transfer mechanism is proposed to dynamically divide the role of nodes. Simulation experiments show that, compared with the original PBFT consensus mechanism, the system proposed in this paper can significantly reduce the communication overhead in the consensus mechanism and increase the fault tolerance rate and security of the system.

Since the discovery of adversarial examples, the research on adversarial examples in the image field has caused an academic boom. In recent years, with the development of artificial intelligence, adversarial samples in the text field have also attracted more and more scholars' research interest. This paper proposes such an adversarial sample generation algorithm in a black box scenario: using a targeted word deletion scoring mechanism, it can find keywords that have a significant impact on the decision of the classifier when the internal structure of the model is unknown, and use the HowNet vocabulary to search the synonyms of these keywords are replaced to generate a set of adversarial samples that are semantically consistent with the original samples. Then combined with genetic algorithm to search for the best sample in the generated sample space. The results of testing LSTM and CNN on sentiment classification and news classification data sets show that the algorithm can greatly reduce the accuracy of the target model with less disturbance.

Due to the poor analysis of the evolution cycle and cycle theory structure, the prediction accuracy of public opinion in university social networks is low. Thus, a new prediction model for the evolution trend of public opinion in university social networks is proposed. By analyzing the theoretical structure of the public opinion evolution cycle and the life cycle of social networks in colleges and universities, the public opinion evolution cycle is determined. By using the E-Divisive algorithm, the evolution trend of public opinion is divided. The trust degree of different views on the network platform and the characteristics of public opinion events are abstracted, and the public opinion evolution trend prediction model is established to predict the evolution trend of social network public opinion in colleges and universities. The experimental results show that the relative error of this prediction model is lower than that of the traditional model. The error value is less than 0.5, which indicates that the prediction accuracy of this prediction model is higher, which is conducive to creating a healthy social network platform for colleges and universities and promoting the healthy development of college students' bodies and minds.

Image as an important communication medium, how to obtain clear and high-quality image data has become a hot topic and subject of research, and the quality of image will affect future image processing and use. Autofocus technology is of great significance in medical microscopes, scanners, satellite navigation, space exploration, robot vision and automatic monitoring. Compared with the conventional focusing technology, the image-based automatic focusing technology is a kind of relying on the collected image data without additional photoelectric detection device.

This paper analyzes the field format of the EtherCAT bus protocol, defines the flip rate concept and analyzes the bit-flip rate characteristics for different types of protocol fields, and proposes an EtherCAT bus protocol field segmentation algorithm based on the bit-flip rate characteristics; Analyze the correlation rules between different fields and propose an algorithm for correlation analysis of protocol fields based on association rule mining; Finally, the algorithm is verified using the EtherCAT bus message data set.

With the development of cloud-native technology, service mesh realizes reliable request delivery in the complex service topology of cloud-native applications. Aiming at the problems in the role-based access control model in the Istio service mesh, this paper adds a service mesh authorization control based on user behavior credibility. The purpose is to continuously monitor user behavior, ensure user identity is credible, regulate user behavior, and prevent others from malicious attacks and impersonating identity to steal information within the scope of user authority.

With the vigorous development of mobile communication technology and the increasing popularity of Intelligent Connected Vehicles (ICVs), the number of automotive electronic functions and software has grown rapidly. The emergence of Domain Controller Unit (DCU) makes vehicles gradually move towards a domain-centralized architecture. During this period, FOTA (Firmware-OTA, FOTA) technology has been gradually applied to the automotive field due to its ease of use and flexible download. FOTA can help DCU iteratively upgrade, allowing car companies to deploy new in-vehicle software at a faster rate. However, due to the lack of effective Ethernet encryption and authentication mechanisms in the in-vehicle FOTA system, there are still some hidden dangers in information security when the FOTA technology is applied to the DCU software upgrade. In this paper, a new in-vehicle domain controller architecture based on "FOTA DCU" is proposed, and a secure FOTA communication strategy for vehicle controller software oriented to automotive Ethernet is designed according to the national secret SM series algorithm. Finally, a simulation experiment platform of automobile FOTA system is built and tested. The experimental results show that the security upgrade method improves the confidentiality, effectiveness and real-time performance of FOTA upgrade for ICVs to a certain extent.

To address the problems of botnet stealthiness and difficulty in detection, this paper proposes a botnet detection model based on dilated convolution. The model first uses dilated convolution to increase the perceptual field of information and extract features from it, and then uses reflection padding to expand the extracted spatial features with samples, then uses squeeze-and-excitation networks to assign different weights to feature channels, and then uses gate recurrent unit to extract the temporal relationships preserved between features, and finally implements botnet detection. The model is validated on the UNSW-NB15 and CIC-IDS-2017 datasets with 99.4% and 99.3% accuracy, respectively, which verifies the effectiveness of the model for botnet detection.

With the expansion of smart devices in 5G, Internet of Things, mobile Internet and other technical scenarios, the number of devices on the edge of the network has increased dramatically. In the edge computing scene, there are a large number of heterogeneous devices, each of which has its own unique characteristics and attributes. For the edge scene, due to the increasing requirements of massive data on the timeliness, security and network dependence of computing facilities, the current cloud platform with Kubernetes as the core is not fully applicable. Therefore, many open source frameworks came into being, and KubeEdge [1] is one of the representatives. Aiming at KubeEdge, this paper proposes a cloud-edge collaboration scheme, which deploys the surface defect recognition algorithm based on YOLOv5 network to cloud edge devices to realize surface defect recognition and node autonomy in edge scenes, and provides a solution for cloud edge collaboration scenes.

With the continuous development of science and technology, artificial intelligence has played a larger role in different fields. Artificial intelligence is developing rapidly at a faster speed and scale, which brings great benefits to human beings as well as great risks. Responsibility, director of the social total management concept and system of artificial intelligence to build is an important part of the legislation of artificial intelligence, through collecting, collating of the cloud and the analysis of large data, to a certain algorithm, engaged in a series of difficult in multiple industries replace human characters, and evaluate people's social responsibility and make some difference. In this paper, based on the lack of responsibility theory system on the basis of a certain of responsibility theory system research, discusses the construction of legal liability system, artificial intelligence and artificial intelligence harmless treatment specification, so as to build the robot of artificial intelligence science responsibility to build the underlying logic research, designed to speed up the development of artificial intelligence.

The fuzzy vault algorithm is mainly used for data privacy protection in fault-tolerant environments, It binds biometric features and secret data together safely to form a vault, which can realize "fuzzy" unlocking. in the real-life, fuzzy vault is mainly used in biometric authentication, image capture, recognition, etc. However, At present, most fuzzy vault schemes have the problems of high computational complexity and low communication efficiency. To solve the problems of high computational complexity and low communication efficiency in identity authentication, we based on the fuzzy vault scheme and construct a multi-secret sharing fuzzy vault scheme. which splits the big secret value into multiple sub-secret values and uses the RS code multi-secret sharing decoding method to improve the efficiency of biological identity authentication. Firstly, splits the big secret value into multiple sub-secret values. Then, We construct a multi-secret sharing scheme. our scheme supports the sharing of multi-secret information. Finally, we analyze the computational complexity and communication complexity of the scheme. The analysis shows that our scheme can reduce the computational complexity and communication complexity by an order of magnitude.