This PDF file contains the front matter associated with SPIE Proceedings Volume 13228, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

The widespread deployment of IoT devices in various fields exposes them to increasing security threats. These devices often store and process large amounts of sensitive data, including personal privacy information, trade secrets, and even national security data. Attackers attempt to tamper with binaries executed by device processors or redirect jump addresses to hijack control flows and execute malicious code. In the face of this challenge, this paper discusses the current status and development trend of controller control flow integrity (CFI) protection of processor processes, and proposes a more complete control flow integrity protection strategy based on RISC-V. This strategy emphasizes that it is difficult to address complex and changing security challenges with a single solution, and it is more prudent to adopt a diversified architecture to design security strategies. Therefore, this strategy proposes an innovative encryption authentication mechanism for data and instructions and combines it with Control Flow Integrity (CFI) to effectively improve the security performance of the processor.

In this experiment, RISCY is selected as the experimental kernel, and the lightweight encryption algorithm ASCON is used as a part of the encryption authentication module, combined with the extension of the RISC-V instruction set and the support of the LLVM compiler framework. This article not only provides a detailed introduction to the cryptographic authentication process in the processor core, but also shows how to achieve control flow integrity through compiler automation. In terms of experimental validation, Embench-IoT was selected as the relevant suite of experiments and tests, and further through the performance evaluation on the FPGA platform, the overhead in terms of code size and number of cycles actually depends on the program, which drops to between 0% and 36% in the benchmark, and the integrity of the data and code sacrifices part of the memory usage, and the computation time is increased by 2 to 3 times. The experimental results show that this strategy not only provides new ideas and methods for the security protection of IoT devices, but also provides a valuable reference for the development of processor security technology in the future.

In order to achieve real-time acquisition of a 3D force sensor, a data acquisition system is designed for the existing 3D force sensor. This system utilizes Xc7a100tffg-484 as the main control chip, where the analog voltage signals are converted into digital signals through AD7606 and transmitted to the main control chip. The data is then sent to the host computer via Ethernet UDP protocol, where it is stored. Through experimental testing, the measurement error between the acquired values of the system and the actual applied force is controlled within 2.5%. This design system offers advantages such as high sampling accuracy, low development cost, and fast response speed.

The categorization of brief Chinese texts, a critical area for extracting insights from data with limited information content, presents unique challenges such as limited word count, ambiguity, and non-standardized information. These factors complicate the extraction and representation of textual features. This study introduces the BERT-based BRLC (BERT Recurrent Layer Composition) model, customized for brief text categorization assignments. The model employs BERT as its foundation, utilizing the Transformer architecture to encode input text and produce a semantically rich vector representation. The BRLC component then captures both global and local sequential features of the text, culminating in classification outcomes through a non-linear transformation process. Evaluated against contemporary deep learning models, the BRLC model demonstrates superior classification performance and efficiency. Utilizing the THUNews dataset, our model attains a precision level of 94.92%, outperforming the original BERT by 2.19%. Comparative experiments confirm the BRLC model's effectiveness in enhancing the accuracy of short text classification tasks.

In recent years, the construction of New Power System has become the main work of the State Grid Corporation of China (SGCC) to help achieve China's carbon peaking and carbon neutrality goals. Due to the high speed, low latency and large connectivity features, the 5G technology is expected to play a significant role in the future power grid and large-scale 5G preliminary application has been carried out by the local power companies in the current stage. According to the research on the transmission requirement of collection information in the low voltage station area, a 5G communication module was designed by the team and the details will be discussed in this paper with testing results, which will provide practical reference for relevant research and application.

With the rapid development of internet technology, cybersecurity faces an increasing number of challenges, among which network intrusion detection systems are crucial. Due to its flexibility, regular expression matching has become a key technology in this area. The fast growth and increasing complexity of modern networks require that regular expression matching technology be both efficient and storage-saving. This article provides an overview of the classification of regular expression matching technologies, from software algorithms to hardware implementations. The content includes an introduction to automata theory, detailed discussions on software and hardware implementation methods. Finally, the article looks to the future development trends, emphasizing the significant value and potential impact of regular expression matching technology in the field of information processing.

GPS/IMU multi-sensor fusion algorithm is of great significance in the auto drive system. GPS has high precision, but the sampling frequency is low and prone to failure; The IMU sensor has a high sampling frequency and is relatively stable, but it is prone to error accumulation. Therefore, the two have good complementarity, and integrating them can obtain a navigation solution with better performance than a single navigation system. In recent years, many algorithms based on Kalman filters (KF) have emerged, and some scholars have proposed using artificial intelligence to fuse GPS/IMU data. This article aims to effectively fuse multimodal sensors and deeply analyzes the advantages and disadvantages of existing algorithms based on Kalman filters, machine learning algorithms, and neural networks.

Symbolic execution generates test cases that cover all possible paths by symbolizing program code and executing program paths in symbolic form. However, this technique is not good in code coverage due to expensive constraint solving, which takes up most of the time of symbolic execution. In view of the problem that constraint solving takes too much time on certain path constraints, this paper proposes to select the best solver for path constraints based on transformer model, so as to minimize the overall solution time, effectively alleviate the problem of excessive solution time for constraint solving, and improve the efficiency and availability of symbolic execution.

Most existing fuzzing techniques rely heavily on coverage guidance. However, such approaches still face a significant limitation: triggering certain vulnerabilities not only depends on code coverage but also on program states. Furthermore, identical code coverage may lead to the discovery of useful new program states for vulnerability detection. To address this issue, this paper proposes Program State and Coverage Feedback Fuzzing (SACFUZZ), which is guided by program states. Specifically, SACFUZZ automatically identifies program state variables crucial for vulnerability detection, which can distinguish between program states. It then recognizes the current program state based on these critical variables, records combinations of value ranges of recently accessed critical program state variables to represent the program state, and utilizes this recorded program state information to guide the iterations of fuzzing. Test cases that trigger new combinations of critical program state variable values or discover new critical program state variables are both added to the seed queue. Experimental results demonstrate that, compared to coverage-guided fuzzing tools, SACFUZZ can discover more crashes in state-intensive programs.

Aids remote monitoring system is an important way to aids maintenance, which plays an important role of improving the maintenance quality of the aids. The communication effect would affect the management level of aids remote monitoring system. However the application of IOT communication and the BeiDou-2 Short Message technology broaden the scope of Aids remote monitoring on the sea, the image of environmental information of the damaged buoy nearby could not be transmitted back to the Aids remote monitoring system. The application of Beidou-3 data transmission technology on the light of the buoy could transmit the images and the voice data to the system which would form pressure to the people who damage the buoys on the sea psychologically. At the same time this application on the buoy light would get image or voice of the environment information concerned which would provide more information to the maritime accidents investigation and marine environment monitoring.

In current offshore fishing operations, there is a growing demand for long-distance operations by small and medium-sized fishing vessels. However, due to the remote distance from the land, terrestrial communication infrastructure such as landbased communication stations cannot communicate with these vessels in a timely manner. Installing satellite communication equipment on fishing vessels is a common method for communication, but the limited bandwidth capacity and expensive cost of satellite communication make it impossible to maintain continuous real-time communication with all fishing vessels. Utilizing the LEACH (Low Energy Adaptive Clustering Hierarchy) networking protocol for vessel networking communication can effectively address such issues. This involves integrating information from cluster head vessels within a cluster and then communicating with satellites, thus enabling satellite communication for vessels within the cluster and extending the lifespan of the communication network. However, the main challenges during offshore fishing operations are the long communication distances and significant delays, compounded by the limited energy carried by communication terminals. Therefore, the traditional Leach protocol's frequent cluster iteration leading to communication energy consumption and substantial time overhead cannot be ignored. To address these challenges, this paper proposes an improved protocol named LEACH-FV. This protocol optimizes cluster head selection based on parameters such as remaining energy and location information of communication nodes, specifically tailored for scenarios with large delays and long communication distances. Furthermore, optimizations are made during subsequent cluster iteration rounds, effectively enhancing the survival period of communication networks for fishing vessels. Simulation conducted using Matlab, comparing the proposed LEACH-FV protocol with several clustering networking protocols including LEACH, LEACH-C, and TSI-LEACH, validates the advantages of the improved Leach protocol in terms of node lifespan. This study focuses on improving the networking performance of small and medium-sized fishing vessels during offshore operations. Through the improved LEACH protocol, it provides an optimized solution for communication among small and medium-sized fishing vessels at sea, holding significance for the sustainable development of offshore fisheries.

Orthogonal Time Frequency Space (OTFS) modulation technology can effectively address the significant challenges brought by fast time-varying channels to signal detection at the receiver, thus providing reliable system transmission performance for high-mobility communication environments. To tackle the problem of high complexity in common OTFS receiver signal detection algorithms, we propose an iterative enhanced detection algorithm based on the delay-time domain. In the delay-time domain, the channel sub-matrix exhibits diagonal characteristics, and the transmission signal and channel matrix appear in the form of vector dot products, effectively reducing the complexity of the receiver signal detection algorithm design. Based on this property, simultaneous utilization of iterative enhanced detection method is employed for interference elimination and Maximum Ratio Combining (MRC) estimation symbol of Least Square Minimum Residual (LSMR) equalizer, aiming to obtain channel delay multipath diversity gain and reduce detector computational complexity. Simulation analysis results demonstrate that compared to standard OTFS signal detection algorithms, such as Minimum Mean Square Error (MMSE) and Message Passing (MP), this detection scheme exhibits certain advantages in terms of computational complexity and system demodulation performance.

Using time and polarization multiplexing technology for transmission can reduce the interference of free space light on quantum signals. However, because of the presence of the free space channel, during light propagation, the polarization of light undergoes random fluctuations, causing substantial light leakage during direct polarization demultiplexing at the receiver, which can overpower the quantum signal. In this paper, we present an enhanced polarization control strategy leveraging a sophisticated algorithm, improving upon an existing dynamic polarization control approach. Through simulations, our proposal shows promising performance improvements. Our work focuses on the optimization of the control algorithm within the context of the previous dynamic polarization control strategy. After identifying limitations and issues in the original algorithm, we implemented a new optimization algorithm to address these problems. Simulations indicate that our optimized algorithm effectively reduces light leakage and provides fast and accurate response to changes in the polarization state of light.

This paper deeply analyzes the ant colony algorithm and its application basis in routing optimization, and clarifies the principle of the ant colony algorithm and its importance to WSN routing optimization. The basic workflow of the ant colony algorithm is summarized in detail, including the key technical points such as the pheromone update mechanism and the path selection strategy, which lays a theoretical foundation for the subsequent improvement of the algorithm. Then, the current research status and existing challenges of wireless sensor network routing optimization are comprehensively analyzed, including energy efficiency, network life cycle, data transmission reliability and so on. The comparative analysis indicates the shortcomings of existing studies in routing strategy, algorithm efficiency and adaptability. After research and analysis, a wireless sensor network (WSN) routing optimization scheme integrating ant colony algorithm is constructed, and the significant advantages of this model in improving WSN routing efficiency are verified through experiments. The performance advantage of the improved algorithm is further confirmed by the comparison and discussion of the results.

In the process of reliability evaluation, there may be inaccuracies in data collection and processing, leading to large deviations in the evaluation results. Therefore, the reliability evaluation method of digital main distribution network communication equipment is proposed. The power supply reliability evaluation indexes of communication equipment include the average power failure frequency of communication equipment, the average power failure duration of communication equipment, the average power failure duration of communication equipment users and the average power supply availability of communication equipment. The experimental results show that the proposed method of digital main distribution network communication equipment performs well in average iteration time and evaluation time, and has high efficiency and practicability. The high annual failure rate can improve the accuracy of reliability evaluation of communication equipment in the main distribution network.

In the context of the continuous development of information technology, the 5G era has ushered in a new intuation, the rapid development of mobile Internet and the continuous richness of business types, the data generated by users in the process of using various types of business has multiplied, the emergence of massive data and the change of data structure. It brings great challenges to operators' data management, data collection, data statistical analysis, data processing and data operation and maintenance. How to deal with these data effectively and efficiently has become the key to operators' improvement of operational efficiency and service quality. Edge computing, as a new computing mode, assigns computing tasks to the edge of the network, making data processing and analysis more real-time and efficient. This paper provides strong technical support for the implementation of edge computing through the characteristics of 5G network, such as high speed, low latency and large connection. It proposes a method to improve the data processing efficiency of operators based on 5G edge computing technology. It adopts theoretical analysis and experiments to carry out comprehensive research, effectively improve the processing of massive data, reduce cost and increase efficiency. And realize the security and efficiency of data, enhance the market competitiveness, and thus improve the user experience.

As drone technology rapidly evolves, the limitations of current communication systems, particularly regarding flight distance, bandwidth, and latency, become increasingly evident. This study addresses these challenges by integrating 5G technology with unmanned aerial vehicle (UAV) systems. Our approach employs a 300-meter aerial 5G network coupled with an advanced onboard 5G intelligent terminal, significantly improving drone communication. The paper details the terminal's hardware and software architecture, emphasizing enhancements in processing, interface configurations, environmental adaptability and cutting-edge software design. Extensive testing across various conditions demonstrates the system's robustness, efficiency, and adaptability, notably in maintaining stable communication across diverse environments. This research contributes to global sustainable technology trends and paves the way for future developments, such as the integration of artificial intelligence, machine learning, and the Internet of Things. This represents a pivotal advancement in UAV technology and its applications.

With the rapid development and widespread application of mobile Internet of Things technology, people have put forward higher demands for its efficient and reliable communication network. Taking into account node energy consumption and network load conditions, the routing of each node can be dynamically adjusted to achieve load balancing and energy-saving optimization of the system. Genetic algorithm and cross validation strategy are used in routing design to improve search efficiency and convergence rate. Under various network load conditions, the throughput performance of the system can be evaluated. When the network load is 10Mbps, the throughput is 8.5Mbps, the packet loss rate is 0.5%, and the delay is 25ms; When the network load is 20Mbps, the throughput is 16.8Mbps, the packet loss rate is 0.8%, and the delay is 30ms. The development of this article is of great theoretical and practical significance for promoting innovation and development of communication network technology in China, as well as building an intelligent and sustainable society.

In the field of key-value storage performance optimization, WiscKey has enhanced the write performance of key-value storage engines based on the Log-Structured Merge-tree (LSM-Tree) by introducing the concept of key-value separation, achieving significant progress in performance optimization of key-value storage systems. Building on this design concept, this paper proposes a new flexible and dynamic partition storage scheme based on key-value separation, named FlexiKV, aimed at optimizing range query performance in key-value separation systems. FlexiKV introduces dynamic value log partitioning and designs a comprehensive scoring mechanism based on access frequency and the number of key-value pairs, achieving effective data organization and storage. The scheme dynamically adjusts partitioning strategies to achieve a balanced distribution of data load and improves the efficiency of range queries in scenarios of data load skew.

Mapping IP addresses into geographic locations is crucial for location-based services (LBS), such as online targeted ads, localized content delivery, and online account protection. However, existing methods are inaccurate in inferring the targets’ locations when their network paths are incomplete. In this paper, we focus on enhancing the geolocation accuracy on IPv6 networks when targets’ paths are incomplete and propose Fingerprint-Geo, an IPv6 geolocation method based on the analysis of regional path fingerprints. Each path to a target region has its unique features associated with this region, which are referred to as regional path fingerprints. By analyzing the regional path fingerprints, an overview of the network measurement characteristics in the region can be obtained, and the targets within the region can be estimated by matching fingerprints even when the paths are incomplete. Experiments conducted in ten metropolises in China and the U.S. IPv6 networks demonstrate that Fingerprint-Geo can geolocate the targets with incomplete paths and is more accurate than the typical method PBG.

In distributed systems, there exists a bottleneck in asynchronous BFT protocol: the broadcast primitive, which occupies a large amount of bandwidth resources, will drag down the nodes with poor bandwidth and slow down the system to reach consensus. In this paper, we propose to decouple the broadcast phase of BFT protocol into consensus phase and message downloading phase. The consensus phase consumes less bandwidth, while the message downloading phase requires more bandwidth. Nodes with limited bandwidth only need to participate in the consensus phase, which will not slow down the overall process. We design PAVID-M protocol to realize the above decoupling, and combine it with multi-value validated Byzantine agreement (MVBA) protocol to build a complete asynchronous BFT protocol system. Experiments show that compared with HoneyBadgerBFT protocol, our scheme can reduce the delay by 43%, which significantly improves the performance of asynchronous BFT in bandwidth-constrained environments.

As the number of satellites in the global satnav system increases, so does the computational load for satellite positioning, which consumes more time and impacts real-time positioning. To enhance real-time positioning and reliability, adopting an appropriate satellite selection strategy is crucial. This paper proposes a fast satellite selection algorithm based on a BDS/GPS combination. It considers the minimum geometry dilution of precision (GDOP) and the time of satellite selection, selecting all visible satellites of the BDS/GPS dual system. Experiments indicate that, within the same time frame, the star selection results obtained by the proposed fast star selection algorithm outperform other star selection algorithms. The algorithm effectively reduces the computational load of the positioning solution while ensuring positioning accuracy.

The traditional communication data encryption methods rely on complete time series to extract encryption features, making them susceptible to data noise and resulting in high packet loss rates. To address these issues, this study introduces an enhanced MD5 algorithm for encrypting network communication data. Initially, a Huffman single item reassembly encryption structure is established to create a deployment of multi-objective fine-grained encryption nodes. Subsequently, a sliding homomorphic mixed encryption model is developed, along with derived network communication encryption instructions. This is followed by the use of overlapping deep learning algorithms to achieve mixed encryption of network communication data. Compared to encryption methods based on ring topology and VPN technology, the proposed encryption structure is more flexible, variable, and exhibits dynamic accounting verification characteristics. The findings indicate that there are fewer encryption discrepancies in practical applications, and the data maintains higher stability during the mixed encryption process.

IP geolocation uses IP addresses to determine the geolocation of network devices, supporting location-based services (LBS) such as online advertising, content access control, and cybercrime tracking. Current city-level IP geolocation algorithms are designed for IPv4 networks, their key assumption of relationships between geolocation and network measurement data, such as static city boundary routers, may not be suitable in rapid-growing IPv6 networks. Therefore, transferring current geolocation algorithms to IPv6 networks often lead to much lower accuracy. To offer reliable city-level geolocation services for IPv6 networks, this paper introduces a city-level IPv6 geolocation algorithm based on self-attention-driven router sequence learning (SRSL-Geo). Instead of finding static mapping rules, SRSL-Geo models dynamic mapping patterns in router sequences of IPv6 targets. SRSL-Geo leverages self-attention-driven deep neural networks to place more importance on routers which are related to city-level geolocation. The experiments in real-world IPv6 networks covering 22 cities of China, Germany, and Japan show SRSL-Geo improves the accuracy of city-level geolocation accuracy averagely by 46.38%, compared with mainstream algorithms.

This article delves into the intricate mechanisms of information dissemination in online social networks, exploring the complex interplay between network structures, user behaviors, and algorithmic influences. The study adopts computational methodologies to dissect the nuances of how information circulates within these digital ecosystems. By examining the multifaceted dynamics of digital communication, this research aims to provide a comprehensive understanding of the factors that shape the flow of information in social networks. It highlights the evolving nature of network architectures, the diverse behavioral patterns of users, and the critical role of algorithms in moderating content distribution. The study emphasizes the importance of computational tools in deciphering the vast and intricate data generated by online social interactions, offering insights into effective strategies for managing digital communication and countering misinformation in the age of social media.

Aiming at the problems of low recognition accuracy, missed detection, and false detection of targets in autonomous driving scenarios of the original YOLOv7, an improved YOLOv7 object detection algorithm is proposed. In the improved algorithm, the SimAM attention mechanism is first introduced at the neck of the network, which can evaluate feature weights in the 3D dimension without adding additional parameters to enhance important features, suppress invalid features, and achieve high detection speed while meeting accuracy requirements. Secondly, choosing EIoU Loss helps the model better learn multi label classification tasks, improving the performance and generalization ability of the model in cases of imbalanced and overlapping labels. The experimental results show that the Precision (P) of the improved algorithm is 75.3%, and the Recall rate (R) is 56.6%. The mean Average Precision (mAP) reached 54.2%. The improved algorithm demonstrates good performance in real-scene detection tasks, effectively reducing the missed detection rate and false detection rate, while significantly improving the detection ability and accuracy of the model.

With the continuous advancement of information technology, remote education exams are facing increasingly diverse cheating methods, and traditional monitoring methods are no longer sufficient to cope with these challenges. This study aims to design and implement a comprehensive anti-cheating system for exams. The core of this system is an intelligent monitoring system that integrates head posture recognition and eye tracking technologies. By monitoring the multidimensional behavior of examinees, it effectively identifies and prevents cheating behaviors. The system also enhances security through functions such as anti-screenshot, disabling clipboard, and preventing remote answering. Through practical application testing, the system can effectively identify and prevent various cheating methods, significantly improving the monitoring efficiency and fairness of remote exams. This research provides a comprehensive and effective technical solution for modern educational exam supervision, which is of great significance in improving the security and fairness of remote online exams.

Aiming at the issues of small surface defects on aluminum profiles and low detection accuracy, this paper proposes an improved aluminum profile surface defect detection method based on YOLOv8. Firstly, the Ghostv2 module is introduced to minimize the number of parameters and volume of the model, and the convolutional layers of YOLOv8 are replaced with the Spatial Depthwise-separable Convolution module (SPD-Conv) to facilitate feature fusion and cross-scale information interaction in the Convolutional Neural Network (CNN), thus improving the ability to extract small target features. Secondly, a detection module for small targets is integrated into the Head section to improve small target detection effectiveness. Then, the LskBlock attention block from LSKNet is introduced into the neck network to dynamically adjust the spatial receptive field to better simulate the range context of different objects. Ultimately, the CIoU loss function is substituted with Wise-IoU, which has stronger positioning ability for detection targets compared to the original CIoU, improving the accuracy of target detection. Experiments show that the improved network model achieves an mAP@0.5 of 84.3%, representing a 5.8% improvement compared to the original YOLOv8 algorithm.

This study proposes an improved YOLOv8 algorithm based on the network model framework in response to issues in grading accuracy, slow speed, high false alarm rate, and the excessive workload of monitoring staff encountered in the tiered monitoring of unsafe banking conditions. These challenges lead to less effective monitoring results. The study utilized a dataset that includes PASCAL VOC2007 and a section of a self-built dataset, collectively comprising of 8,511 images. Initially, a Partial Convolutions module was introduced to the backbone section, enhancing the ability to recognize occluded objects in images and improving the resistance to interference. Following that, an enhancement was made to the Bottleneck structure, integrating Large Kernel Attention (LSKA) and enhancing its detection capabilities for small targets. For large target detection, Multi-Scale Dilated attention was employed to enhance the model's processing efficiency and detection accuracy. Lastly, an ASFF head was incorporated. Overall, the improvements resulted in a 3.0% increase in mAP@0.5, boosting detection accuracy and speeding up the detection process.

Attributional network anomaly detection is increasingly becoming a focal point of academic research due to its applications in critical domains like social networking and financial fraud. This task faces many challenges due to the differences in attributes between anomalous nodes and others, as well as their complex interactions. While existing shallow methods fall short in concurrently addressing both attributes and structures, leading to suboptimal performance, methods based on deep learning have made significant advances in enhancing anomaly detection. However, they still lack sufficient exploitation of anomaly information. In response to the aforementioned issue, this paper introduces an attribute graph node anomaly detection method based on training strategy optimization (ANATSO). The model constructs a secondary view through edge perturbation and samples subgraphs from each view. These subgraphs are then used to initialize networks for subgraph-node and node-node instance pairs. Additionally, a phased training process is implemented, where the outcomes from the initial phase are used to optimize the node input processing in the subsequent phase, ultimately calculating anomaly scores for each node. This research was evaluated on five benchmark datasets, and the results demonstrate that compared to traditional baseline methods, our model significantly improves the precision of anomaly detection.

With the continuous innovation of wireless communication technology, fog computing-based distributed storage serves as a useful complement to centralized cloud computing and is trending towards large-scale deployment driven. However, for the distributed storage scenario in fog computing, there is currently a lack of effective data auditing methods. This article proposes a distributed data storage auditing scheme in fog computing environment. In our scheme, we first design a distributed storage model based on fog computing. In this model, we use threshold secret sharing technology to divide the outsourced data into multiple blocks and upload them to multiple fog servers to effectively solve the single point of failure and achieve privacy protection. In addition, we have established a fog device audit consensus group based on the verifiable random function (VRF), which ensures the correctness of audit results through consensus among members. In addition, this article proposes a lightweight validator generation algorithm to address the issue of limited resources at the user end. Finally, experimental results show our distributed data storage auditing scheme is efficient and effective.

JavaScript engines serve as the cornerstone of the modern web ecosystem, and fuzzing, as one of the most advanced vulnerability discovery techniques, is often used to detect vulnerabilities in JavaScript engines. Test cases generated by fuzzing JavaScript engines are essentially JavaScript code. However, randomly generated JavaScript code by fuzzing is often invalid and unable to detect vulnerabilities in JavaScript engines.

To ensure the syntactic and semantic correctness of test cases generated by fuzzing JavaScript engines, this paper proposes a code block combination strategy based on semantic constraints. Specifically, the paper first decomposes Proof of Concept (POC) into syntactically correct code blocks. Simultaneously, to satisfy semantic constraints and ensure the correctness of test cases, it is necessary to record the preconditions and postconditions of code blocks during the decomposition process. Then, the originally syntactically correct code blocks are combined to form test cases under the premise of satisfying semantic constraints. This approach not only ensures the syntactic correctness of test cases but also ensures their semantic validity.

To validate the effectiveness of the proposed method, this paper designs and implements the fuzzing testing system ValidJSFuzz. ValidJSFuzz is experimentally compared with three classic advanced JavaScript engine fuzzing tools, namely jsfunfuzz, CodeAlchemist, and DIE, on four mainstream JavaScript engines: JavaScriptCore, SpiderMonkey, ChakraCore, and V8. The experimental results indicate that the ValidJSFuzz system can improve test case correctness by 25% compared to the best-performing tool, DIE.

There are performance deviations between model and system due to incompleteness of logs when directly using logs generated by the system to modeling. Existing mining methods rarely consider the incompleteness of the logs. This paper starts by exploring the potential evolution of local behavioral relationships in the model, seeking ways to express the potential evolution concisely. Firstly, the restriction of the provided logs on the evolution between the transitions and the blocks are analyzed. Then, construct parallel structural blocks that are not affected by the integrity of the log by using of concurrence relation. Next, search the needed transition set for each block with the help of adjacent relation, cut them to exclusive block by inductive mining algorithm and embed them to blocks respectively. Then constitute the exclusive block appearing in other positions further. Finally, a complete process model generated by integrating all blocks and evolution signs used for showing the potential evolution in the process model. The paper run an example that verifies the feasibility of mining method of blocky process model based on incomplete-logs.

To address the challenge of insufficient detection rates and precision for miniature objects in infrared road scenes, we propose a refined object detection algorithm called SSTD-YOLO built upon YOLOv8 tailored for infrared road environments. Our approach begins by leveraging SPD-Conv to mitigate the loss of detailed information stemming from the prevalence of small objects and low image resolution. Furthermore, we integrate a triplet attention mechanism within the C2F module to enhance the identification of complex samples. Additionally, our model incorporates Dysample in the feature fusion module, surpassing alternative upsampling techniques in inference latency, memory consumption, FLOPs, and parameter count. To augment the detection accuracy of small targets in infrared images and alleviate instances of undetected small targets, dedicated layers for small target detection are integrated into our model. Experimental results demonstrate that our proposed model outperforms the YOLOv8n baseline, increasing mAP@0.5 by up to 14.16%.

When there are a large number of branch structures in the program, the path explosion problem will occur in symbolic execution. Because the path space is too large, all paths cannot be executed one by one, resulting in low coverage. The path with higher priority execution coverage can improve the symbolic execution coverage. Therefore, this paper proposes a path search strategy based on graph centrality analysis. By constructing a code coverage-oriented Katz path graph, and using Katz centrality score to calculate the node's centrality score, path priority is determined according to the node's centrality score. This strategy can prioritize the exploration of important paths and improve the efficiency of path selection, thus alleviating the problem of path explosion and improving code coverage. The experimental results show that compared with the existing search strategy, this strategy can explore more paths, and the coverage rate is improved by 15.7% on average.

Fuzzing is a technique for detecting vulnerabilities in target programs by generating a large number of test cases randomly. In this approach, code coverage serves as feedback to guide the evolution of seed queues. However, such evolutionary strategies often utilize context-insensitive branch metrics, which fail to differentiate the execution of the same branch in different contexts, potentially overlooking new internal program states. To address the aforementioned issues, this paper proposes context-sensitive branch coverage-guided fuzzing and implements a prototype tool called CSBFUZZ (Context-sensitive Branch Coverage Feedback Fuzzing). Specifically, CSBFUZZ utilizes context-sensitive branch coverage as feedback information by: (1) CSBFUZZ automatically identifying variables crucial for representing the sequence of branch executions, referred to as critical variables in this paper; (2) CSBFUZZ identifying the current branch execution sequence based on recorded critical variables' order; (3) incorporating test cases that trigger new branch execution sequences into the seed queue to guide subsequent fuzzing. The experimental results demonstrate that, compared to fuzzing tools guided by context-insensitive branch coverage, this paper can uncover more crashes and vulnerabilities, and explore more code in context-sensitive real-world programs.

The rapid development of Internet technology makes online rumors increasingly rampant. The current researches on rumor detection rely on large-scale labeled instance data. However, in the face of emergent events, obtaining large-scale labeled data becomes extremely difficult. To address this issue, this paper proposed a few-shot rumor detection model based on prompt learning and meta learning, named MPRD (Mate Prompt Rumor Detection, MPRD). MPRD combines hard prompt and soft prompt to explore universal rumor prompt templates that are not limited by specific topics, thus offering greater flexibility. Moreover, it employs meta-learning methods to sample meta-tasks based on past events, trains prompt models on these meta-tasks, and ultimately applies them to perform rumor detection tasks in emergency events. Experiments on two publicly datasets show that MPRD outperforms state-of-the-art baseline models on detection accuracy, and exhibits significant performance improvement on few-shot rumor detection under different topic.

To fully leverage the complementary advantages of the Artificial Bee Colony (ABC) and Differential Evolution (DE) algorithms for various optimization problems, this paper introduces an ABC-DE hybrid evolutionary algorithm based on individual preference for population selection. This method divides the population into subpopulations using a preference-based selection mechanism and employs ABC and DE for diverse subpopulation search strategies. Additionally, it implements a new resource allocation mechanism that increases computational efforts for promising solutions. Comparative analysis on the CEC2014 benchmark suite with five other current ABC and DE improved algorithms confirms that the proposed hybrid evolutionary algorithm effectively enhances convergence speed, accuracy, and robustness.

In the rapidly evolving landscape of Mobile Edge Computing (MEC), Deep Reinforcement Learning (DRL) emerges as a potent solution for task offloading, resource allocation, and network optimization challenges due to its exceptional decision-making and adaptability features. This paper commences by reviewing key technologies and challenges within MEC and proceeds to analyze the current applications and breakthroughs DRL has fostered in this realm. Subsequent sections present empirical research and data analyses to showcase the performance of various DRL algorithms in MEC settings, laying a quantitative foundation for further investigative directions. Synthesizing extant research, the paper concludes by discussing open issues and future prospects of DRL in the MEC domain, offering insights and projections for related studies.

Keyboard acoustic side channel attacks exploit audio leakage from typing to deduce typed words with a certain level of accuracy. Researchers have been improving the accuracy of these attacks through various techniques of feature extraction and classification. They also apply machine learning methods and deep learning techniques to enhance the accuracy of their results. At the same time, defense mechanisms against these attacks have not kept up with the increasing precision of the attacks. In this study, we introduce a practical defense strategy against keyboard acoustic attacks during password and text typing. We evaluate its effectiveness against multiple attack vectors. Our defense strategy involves generating unique background sounds using GANs to mask sensitive audio leaks from the keyboard, thwarting side channel attacks from extracting usable information about typed content. The background sounds are produced by the device used for text input. We assess the usability of our approach for short and prolonged usage durations, demonstrating that the addition of background sounds does not impede users’ ability to enter passwords or perform computer tasks effectively.

In this paper, the proposed FLB-YOLOv8 model address issues in current traffic sign recognition, such as leakage, false detection, low accuracy, and excessive model parameters. Firstly, a small target detection layer is added to improve semantic information and feature expression for small targets. Secondly, the FasterNet Block structure replaces the Bottleneck structure in the C2f module of YOLOv8, reducing parameters and computation while maintaining accuracy. The introduction of the LSK module adjusts the spatial receptive field dynamically, improving feature extraction capability. Lastly, a weighted bidirectional feature pyramid network enables efficient bidirectional cross-scale connection and feature fusion, balancing feature information of different scales. Experimental results on the CCTSDB dataset show that the improved model achieves 78.3% mean average precision (mAP), a 4.7 percentage point improvement over the original model. It also reduces model parameters by 56% and computation by 26%. This algorithm effectively improves detection accuracy, reduces model complexity, and achieves better performance in traffic sign detection.

With the increasing amount of data, more and more users choose to store data in the cloud server, and the security problem of data in the cloud server begins to come into people's vision. In order to protect data security, users will encrypt the data and upload it to the server. How to search the encrypted data efficiently is an urgent problem. The oblivious dynamic crosstags scheme achieves efficient search on encrypted data at the cost of partial leakage. However, the leakage of oblivious dynamic cross-tags scheme has been proven to have security issues. This paper proposes an oblivious dynamic cross-tags scheme based on hash subset membership check, which not only solves the common problem of oblivious dynamic crosstags schemes-x-item leakage problem and keyword pattern result problem, but also supports dynamic database and multikeyword conjunctive search, which improves the security of the scheme while maintaining efficient search.

The Artificial Bee Colony (ABC) algorithm excels in solving optimization challenges but faces slow convergence, low precision, and local optima issues.To mitigate these limitations, a novel hybrid optimization algorithm, termed the IABC algorithm, was developed to improve the global search capability and speed up convergence. Initially, the employed bee phase adopts a sine law-based dimension-changing strategy to boost search and exploitation capabilities, accelerate convergence, and prevent premature convergence; in the onlooker bee phase, it follows the global optimum, altering its step size—large for broad searches in the early stages and small for precise searches later on. Finally, in the scout bee phase, the hunting strategy from the SWO algorithm (Spider Bee Optimization) is utilized to strengthen global and local search abilities. Experimental conducted on the CEC2017 dataset confirm enhanced strategies' success and performance efficiency. Comparisons with other ABC algorithms and competitive algorithms show that the IABC significantly enhances the performance of the ABC. In nearly all test functions, the IABC attained the superior solution quality, quickest rate of convergence, and most robust performance. Its effectiveness is further proven by successful application in pulsation suppression for the HPCF(high-pressure constant flow) pump pulse systems, indicating its potential for solving diverse optimization problems.

IP geolocation technology is very important for many network services. However, the existing IP geolocation methods can not accurately give the error range of geolocation results, and it is difficult to meet the actual geolocation needs. This paper proposed a new IP location range inference method based on router location estimation (RLG). The first step is to infer the location of the router according to the delay between the landmark and the router, and the next step is to infer the location range of the target according to the delay between the target and the router. Experiments based on 7045 landmarks show that RLG can improve the geolocation accuracy while estimating the geolocation error. Compared with SLG, Hop-Hot and DDG methods, the geolocation errors of this method are reduced by 53.1%, 45.6% and 38.7% respectively.

Along with the development of new-generation technologies such as Big Data and Artificial Intelligence, education has attracted much attention as an important practice area for cutting-edge technologies. In this context, the concept of student portrait has emerged. Through in-depth analysis of student profiles, it can help educational institutions implement differentiated teaching more effectively and improve the quality of education. However, in the process of applying student portrait, some problems have emerged: on the one hand, although data mining algorithms are already very mature, they are less used in the field of student portrait and their accuracy is low. On the other hand, different educational institutions and platforms build different student portrait labels based on their own needs and understanding, which lacks universality and interoperability. Aiming at the above problems, we take the construction of student portrait labeling system as the starting point, and construct the student portrait model through the improved K-prototype algorithm, and finally, the experiments show that the constructed student portrait can accurately describe the students, and to a certain extent, it provides data support for the decision-making of the school education administrators, and reduces the cognitive burden of a large amount of data.

A PCB surface defect detection algorithm based on YOLOv8 is proposed to address the challenges of small object size and low detection accuracy. This algorithm effectively enhances detection accuracy while ensuring real-time detection speed. Firstly, the conventional convolutional layers in YOLOv8 are substituted with the SPD-Conv module, enabling finer feature extraction for small defects. Secondly, to improve the detection accuracy of small targets, an extra small detection head is incorporated into the Head section. This modification enables the network to prioritize small objects by combining shallow and deep feature maps. Meanwhile, in the neck section, the lightweight upsampling operator named CARAFE is utilized to gather contextual information across a broader receptive field, thus improving the network's ability to fuse features. Finally, this original CIOU loss function was improved to Wise-IOU. Compared to traditional CIOU, Wise-IOU considers not only the relative positions and size differences between objects but also introduces an intelligent weight adjustment mechanism that can adaptively adjust weight coefficients, thereby improving the accuracy and robustness. Experiment shows that the improved model reaches an mAP@0.5 of 91.6%, a 10.1% progress over the original YOLOv8, with a model size of 6.80MB, meeting lightweight requirements, and a detection speed of 108fps, meeting the real-time requirements of PCB defect inspection applications.

For 4G/5G voice services, rapid restoration of IP Multimedia Subsystem (IMS) voice services during signaling storm is fundamental to ensuring stable operation and improving the user experience of 4G/5G networks. However, IMS solutions for recovering from signaling storm are still inadequate for real-wild networks. This is evidenced by the recovery time from global network outages caused by signaling storm in recent years. Therefore, we propose an IMS Disaster Restoration Enhancement (IMS-DRE) mechanism using a called restoration-application server to effectively and quickly recover the IMS registration and call services at a low cost. Based on China Telecom's real-wild data of the network, experimental results show that the proposed mechanism was resistant to signaling storm and superior to existing recovery approaches.

In order to further improve the accuracy of latent fault diagnosis, this paper proposes the latent fault diagnosis method of GIS equipment based on multi-component data fusion. Through multi-interface transmission technology, transfer GIS device multi-factor data flow, eliminate the collected data with zero bias errors and eliminate the data of different GIS devices, analyze the multi-component data flow, extract the basis for the latent fault diagnosis model based on Elman neural network to realize the latent fault diagnosis of GIS devices. The experimental results prove that this method has high goodness of fit and diagnostic sensitivity and low misdiagnosis rate, which realizes the improvement of the diagnosis accuracy of latent faults.

With the rapid development of information technology such as cloud computing and Internet of Things, outsourcing computing has received more and more attention. Outsourcing computing provides a way for resource-constrained clients to solve complex computing problems. Matrix inversion is a basic mathematical operation widely used in scientific and engineering fields. However, there are some problems in the existing matrix inversion outsourcing computing, such as privacy security challenges, the limitation of parallel computing in multi-server mode and poor verification efficiency. In order to solve these problems, we propose a three-step decomposition based matrix inversion multi-server parallel computing algorithm, and designs a corresponding matrix inversion multi-server security verifiable computing scheme. In this scheme, the matrix partitioning and matrix 𝐿𝑈 decomposition techniques are used to decompose the matrix inversion process into three steps: matrix LU decomposition, triangular matrix inversion and triangular matrix multiplication. At the same time, the number of servers is extended from two to multiple, which improves the efficiency of server parallel computation. Security analysis and simulation experiments show that the proposed scheme can improve the efficiency of outsourcing computation while meeting the security goals.

In today's digital age, there is a fierce game between the performance pursuit of computer system and information security. However, with the development of technology, various forms of attacks continue to emerge. Transient execution attack, as a subversive hardware architecture vulnerability, can obtain sensitive information of the system by exploiting the internal design vulnerabilities of the processor in the transient execution caused by some errors of the processor, which poses a threat to the credibility and robustness of the overall system.

From the perspective of branch predictor, this paper deeply analyzes how transient execution attack exploits branch prediction structure to achieve incorrect program flow execution and illegally steal secret data. Aiming at the structural vulnerabilities of processor branch prediction unit, this paper proposes an improved mechanism of secure branch predictor to resist transient execution attacks through the hardware architecture level. The experimental results show that the performance overhead brought by the proposed improved scheme is less than 6%, and the false prediction rate of the branch predictor is reduced by about 2%, which achieves a better balance between security and processor performance.

The natural resources big data platform was constructed using cloud technology, IoT, 5G, and other technologies to integrate and utilize various operational resources of natural resources-related departments. The platform has real-time connectivity to field hand-held devices, observation stations, monitoring facilities, and other terminals, enabling the convergence, management, and sharing of a large amount of sensitive data within the organization. The protection of network security for the natural resource big data platform is complex and challenging, with a significant risk of data leakage. Therefore, it is crucial to propose more effective security solutions. This paper analyzes the architecture and implementation of a zero-trust network security system for the natural resource big data platform. It explores the design path of the network security system based on the zero-trust architecture and proposes a design scheme to meet remote access.

In recent years, a significant increase in cybersecurity incidents has been observed, characterized by the proliferation of malicious software across mobile networks, frequent cyber-attacks on domestic websites originating from abroad, manipulation of industrial control systems and smart devices, and a rising tide of data breaches involving website and personal information. Moreover, the emergence of fraudulent ransomware and advanced persistent threats (APTs) poses a direct risk to national security and societal stability. This paper offers an in-depth examination of the prevailing state of cybersecurity and suggests strategies for the development of information security surveillance and early warning systems, as well as measures for enhancement.

Due to the significant asymmetry between the high frequency of access and the low protection level of government networks and their edge devices, there is a considerable risk of being targeted by hackers. This study initially presents an overview of government terminal network edge devices, proceeds to analyze the security aspects of these devices, and subsequently puts forth a security model design aimed at mitigating the identified issues and vulnerabilities that could potentially result in unauthorized intrusions. Finally, by utilizing some common technologies through simulation experiments, a cost-effective, high-performance, practical, and close-to-real-world security system is designed. The simulation experiment results demonstrate that this model design can achieve automatic defense against network intrusions, ensuring the reliability of government terminal networks and their edge devices, thereby strengthening the security of government terminal networks.

Vulnerability is an important threat to current cyberspace security, and the vulnerability exploitation code is the carrier for attackers to attack information systems through vulnerabilities. By analyzing the exploit code, security personnel can understand the vulnerability location of the information system and the common exploitation techniques used by attackers, which is of great significance to cyberspace defense. In the common arbitrary code execution and other vulnerability exploitation code, the control-flow hijacking point is the key location where the attacker uses the vulnerability to modify the execution process of the program. Locating the control-flow hijacking point can provide insight into the attacker's exploitation path and can help security personnel formulate a more complete vulnerability defense strategy. This paper proposes a method to locate the control-flow hijacking point of vulnerability exploitation based on jump-oriented features. The algorithm analyzes the program's jump-oriented features during indirect jumps, invocations and function returns, and detects and locates the control-flow hijacking point of vulnerability exploitation based on the different jump-oriented features. The experimental results of five real vulnerability exploitation cases show that the method in this paper achieves good results on the real vulnerability exploitation cases and has a high degree of accuracy and performance.

With the global energy structure transition and increasing emphasis on environmental protection, new energy vehicles, especially electric vehicles, have become an irreversible trend. In order to efficiently and conveniently provide power for the rapidly growing fleet of electric vehicles, multi-network interaction serves as a solution. It relies on smart grids to connect electric vehicles, charging stations, and parking facilities, optimizing energy supply. However, user usage across multiple platforms presents challenges in password management. Traditional password vault schemes pose security risks such as synchronization service failures and data breaches. Blockchain technology provides a more secure storage method. This paper proposes a model and scheme for cross-platform password management based on blockchain and smart grids, introducing an oblivious key management server to support smart grids in assisting large-scale users. The scheme enhances security through re-encryption strategies and honey password vaults, achieving secure and efficient password management in the multi-network interaction scenario. Security analysis demonstrate that the scheme is both practical and secure.

JIT (Just-In-Time) compilers are among the intricate components within JavaScript engines, harboring numerous potential security vulnerabilities. Fuzzing, an advanced vulnerability discovery technique, is frequently employed for detecting vulnerabilities in JavaScript engines. However, employing fuzzing to detect vulnerabilities caused by JIT compiler optimizations poses significant challenges.

To facilitate the testing of JIT compiler components in JavaScript engines using fuzzing, this paper introduces a template combination strategy. This template guides the generation of test cases capable of triggering vulnerabilities in JIT compiler redundancy elimination optimizations during the fuzzing process, resulting in the implementation of the AimJITFuzz system. Specifically, by analyzing existing CVEs generated by JIT optimizations, the vulnerability mechanisms arising from JIT compiler optimizations are summarized. Subsequently, code block combination templates are designed based on these vulnerability mechanisms. These templates are utilized to guide the splitting of existing proof-of-concept (POC) codes into a code block pool and to guide the combination of code blocks to form test cases capable of triggering JIT compiler optimizations.

Experimental results demonstrate that the AimJITFuzz system can generate test cases capable of triggering JIT optimization vulnerabilities based on the vulnerability mechanisms. Within one week, AimJITFuzz triggers six times more crashes caused by JIT optimizations compared to FUZZILLI. Moreover, AimJITFuzz achieves the maximum branch coverage within the same timeframe compared to other tools.

As encrypted traffic becomes increasingly prevalent in modern network communications, effective classification of encrypted traffic is crucial for network security and management. However, existing classification methods face significant challenges when dealing with encrypted traffic, especially in capturing the complex temporal features and contextual information of traffic data. For this reason, this study proposes the GBERT model, a novel hybrid neural network structure that integrates the contextual understanding capabilities of the BERT model with the temporal processing advantages of Gated Recurrent Unit (GRU). Validated across several public encrypted traffic datasets, the GBERT model demonstrates exceptional performance, notably achieving classification accuracies of 99.03% on the ISCX-VPN-Service dataset and 99.32% on the USTC-TFC dataset. Additionally, this study explores various GRU configurations, offering valuable insights for selecting the most effective model architecture for specific tasks.

Blockchain is a shared digital ledger that, once transactions are recorded and verified, cannot be altered. Since most blockchain nodes maintain a copy of the ledger, blockchain brings the benefit of decentralization but also introduces privacy concerns. This paper explores the privacy issues associated with blockchain transaction data, primarily using zero-knowledge proof technology to design a privacy protection scheme. This scheme utilizes Pedersen commitments and Bulletproofs zero-knowledge proof technology to prevent the disclosure of transaction information while ensuring the correctness and legality of transaction data. Security analysis of the scheme demonstrates that it effectively addresses problems such as double-spending, payment authorization, and dishonest miners, thereby enhancing blockchain performance while protecting the privacy of blockchain transaction data.

Network intrusion detection technology is one of the important technologies to ensure the security of industrial production, civil and other fields. In recent years, the amount of Internet data is growing rapidly, and the ways of network intrusion behavior are also evolving. An intrusion detection algorithm with strong detection ability, rapid response, good generalization and active defense function is designed and developed, which has broad application development space. Considering the characteristics of network traffic data in time and space, an intrusion detection model based on Convolutional Neural Networks (CNN) and Gated Recurrent Unit (GRU) is proposed. It can quickly detect the spatial and temporal characteristics of traffic data. And for the problem of low classification accuracy of network data, the three-way decision theory is introduced into the study of intrusion prevention system, and an intrusion detection algorithm based on CNN-GRU and three-way decision is designed. The models are verified by CSE-CIC-IDS 2018 data set. Experimental results show that the model based on CNN-GRU-TWD can achieve better results in the classification of sufficient training samples, and the accuracy, precision, recall and F1 value are higher than other comparison models. Compared with the model CNN, the model based on CNN-GRU-TWD has an accuracy of about 0.7 percentage points, and the precision, recall and F1 value are about 2 percentage points higher. It can be seen that the proposed CNN-GRU-TWD solves the problem of over-fitting in training and effectively detects network traffic data.

With the development of wireless communication technology and sensor technology, wireless sensor networks are widely used in various fields such as industry, agriculture, medical care, etc., and have become an important part of building smart city networks. However, due to the openness, distribution, and resource constraints of wireless sensor networks, this network is more likely to become a target of attack. In this paper, based on the idea of trustworthiness metrics, we screen and isolate malicious nodes in the network by measuring and determining the static, dynamic, and transmission effectiveness of the nodes. Experimental validation shows that this scheme can exclude malicious nodes to a certain extent to ensure the security and trustworthiness of other nodes in the network.

Big data network has the characteristics of large scale and diverse types of data. How to quickly capture and accurately judge the types of network attacks and effectively defend against illegal intrusions is a key problem to be solved in network security. As a key research direction in the field of machine learning, ensemble learning has higher detection accuracy than single classifier, and is widely used in anomaly intrusion detection. Aiming at the challenges brought by the feature redundancy and class imbalance of the high-dimensional massive data generated in the big data network, the ensemble learning is adopted to integrate the classifical algorithms such as random forest, XGBoost and LightGBM into the network intrusion detection, and determines an optimal model through tuning and comparison. The experimental results show that the accuracy rate of the fusion model proposed in this paper can reach more than 95%, and its performance is better than the optimal submodel. The fusion model method based on ensemble learning can detect and identify the intrusion data more effectively, which is of great significance for maintaining the security of cyberspace and ensuring the sustainable development of society.

Given that heterogeneous information networks contain richer information and more complex semantics compared to homogeneous graphs, this paper utilizes heterogeneous graphs to depict the diverse relationships among users, reviews, and stores within a fake review dataset, thereby more effectively revealing the associations between review publishers and their reviews. This structural advantage facilitates the detection of collusive behaviors among fake review groups and captures global features. Moreover, we employ a Heterogeneous graph Attention Network (HAN) for automatic feature extraction of reviewer nodes. Within this framework, node-level attention learns the interactions between nodes and their neighbors defined by meta-paths, while semantic-level attention focuses on assessing the importance of different metapaths in the heterogeneous graph for specific tasks. Through the learning of these two levels of attention, our model can hierarchically optimize the combination of neighbors and meta-paths, resulting in node embeddings that more accurately capture the complex structures and rich semantics within the heterogeneous graph. Based on these advanced features, our approach effectively detects and identifies collusive fake reviews, demonstrating superior performance.

In response to the problem of low success rate in identifying abnormal access security in network databases, a network database abnormal access security identification method based on GSA-SVM algorithm and deep features is proposed. By combining deep feature extraction technology and learning high-level abstract features from raw data, the accuracy and efficiency of anomaly access detection are effectively improved by utilizing the global optimization ability of GSA algorithm and the strong classification performance of SVM algorithm. The experimental results show that this method has high accuracy and stability in identifying abnormal access to network databases, providing an effective technical means for database security. This method has broad application prospects in the field of abnormal access detection in network databases, and can provide strong support for the security protection of network databases.

With the development of informatization, the value of data is rapidly escalating. The security and controllability of data have become increasingly important. Digital watermarking is an effective method for data tracing and copyright protection. However, most digital watermarking methods focus on images, audio, and video files, with few applications dedicated to document file protection. Therefore, we present an imperceptible and cryptographic digital watermarking method for document files, inspired by the image digital watermarking based on the Least Significant Bit (LSB). Firstly, the watermark information is divided into segments which will be encrypted into ciphertexts separately. Then, these ciphertexts are converted into a binary byte stream and embedded into the LSB of the characters’ color information in document files. Finally, we design an algorithm combined with the chaos theory to execute the embedding process, ensuring its controllability and randomness. Meanwhile, we design a watermark extraction method with a voting mechanism to restore and verify the watermark information. This method can resist the destruction of document files (such as adding, deleting, or modifying some characters) and enhance the stability and traceability of watermark information. The results demonstrate that the accuracy of information extraction is still nearly 80%, even when up to 40% of the document's characters are deleted, enabling quick verification of document copyrights and rapid tracing of the data source.

This paper proposes a new high-reliability IPOE authentication method, Based on actual cases, the original traditional IPOE authentication method is upgraded and transformed. Through a detailed analysis of some problems existing in the original IPOE authentication method, three Session creation and recovery methods based on Sec-Initator-triggered authentication, DHCP Request and IP trigger are designed, and the original IPOE authentication method is completely redesigned to improve the reliability of IPOE authentication. The method is verified from multiple angles to ensure the availability and high reliability of the proposed IPOE authentication method.

Security command vehicles are used for emergency response command in counter-terrorism, emergency response, post disaster rescue, and other emergency situations. There are many types of installed equipment, obvious differences, difficulties in integrated control, and a lack of a unified operation and maintenance data management platform. To address these issues, this paper proposes a comprehensive operation and maintenance system for security command vehicles based on embedded Linux. The system is based on vehicle comprehensive operation and maintenance equipment(VOME), combined with multi-source heterogeneous data fusion, Internet of Things, and Web technology, to achieve data collection, centralized control, data storage, and operation and maintenance management functions. At the same time, the application was carried out on a certain security command vehicle, and the effectiveness of the proposed system was verified through network signal testing, CAN interface testing, and system testing. This has important engineering practical significance for improving the operation and maintenance management capabilities and emergency command capabilities of security command vehicles.

With the rapid development of information technology, network security has become an increasingly important focus. As a key means to protect data security, encryption technology is crucial to enhance its security. However, a single encryption technology often has limitations. This study proposes a secure encryption system that combines Advanced Encryption Standard (AES) and Secure Hash Algorithm (SHA-512). It introduces the design concept and implementation method of the encryption scheme combining the two algorithms in detail, and conducts an in-depth analysis and evaluation of the system's security. Experimental results show that the system provides high security performance and excellent performance indicators.

This paper proposes a self-organizing mechanism for service mesh proxies based on Named Data Networking (NDN), called NS-FIB. NS-FIB utilizes the Forwarding Information Base (FIB) maintenance mechanism in Named Data Networking to optimize the maintenance of FIB in service mesh, enabling proxies to autonomously negotiate and adjust network topology and reduce dependency on the controller, thereby better adapting to distributed systems.

As encryption technologies have become pervasive in safeguarding user privacy, encrypted traffic now constitutes a significant portion of network traffic, posing challenges to conventional rule-based classification methods due to the randomized nature of encrypted contents. In this paper, we deploy the MobileNet within the realm of encrypted traffic classification and propose ET-MobileNet. ET-MobileNet is a lightweight model that specifically designed for devices with limited computational resources. It extracts the packet length sequence and packet arrival interval sequence as features from the session flow, and reshape them into a two-channel picture for input. Our experiments on the ISCXTor2016 dataset encompassing 13 applications indicate that ET-MobileNet can achieve remarkable results while maintaining a low computational cost.

After the application of anti leakage methods, it is easy to re encrypt, resulting in poor anti leakage performance of data flow across domains. Therefore, it is necessary to design a new cross domain cyclic anti leakage method for sensitive network data based on homomorphic encryption. We have constructed a model for cross domain flow and leakage prevention of sensitive data, and designed a cross domain flow and leakage prevention algorithm for sensitive network data using homomorphic encryption, thereby achieving cross domain flow and leakage prevention of data. The experimental results show that the designed anti leakage method for cross domain data flow has high accuracy and detection rate, as well as low false positive rate, false negative rate, memory consumption, and CPU execution time. This proves that the method has good performance, reliability, and certain application value, and has made a certain contribution to reducing the security risk of sensitive data transmission.

Community search is a technique used to identify and retrieve groups of nodes with specific attributes or associations by analyzing the connections between nodes in a network or graph structure. Existing researches mainly focus on single secure search requirements, making it difficult to support complex graph searches. Therefore, this paper proposes a secure search scheme that simultaneously satisfies k-core and keyword attribute similarity requirements. Specifically, to enhance search efficiency, we utilize an improved core decomposition tree to index community graph containing keyword attributes. Additionally, based on the triangle inequality, we design an efficient pruning strategy. To enhance security, Paillier homomorphic encryption and matrix encryption are adopted, presenting a secure and efficient attribute community search scheme that protects the privacy of outsourced data, query requests, and query results. Security analysis and performance evaluation demonstrate that our proposed solution is both secure and efficient.

Neural networks have demonstrated remarkable potential in a multitude of domains, such as image recognition, speech recognition, natural language processing, and medical diagnosis. However, with their widespread application, privacy concerns regarding user data have become increasingly prominent. During the process, neural network models may involve vast amounts of sensitive personal information. The critical research question now lies in how to harness the value of data through neural networks while safeguarding users' privacy from being compromised. This paper presents a semi-honest three-party privacy-preserving inference framework for neural networks, designed to conduct neural network inference while ensuring the protection of user data from being disclosed. The framework effectively shifts the computational overhead to the preprocessing stage. Within this framework, our multiplicative protocol requires only 2l bits for online computations, significantly reducing communication complexity and runtime, thus enabling both efficient and secure model inference.

In the contemporary landscape of the battlefield, the enhanced cyber warfare capabilities of adversaries mandate that routing protocols are equipped with endogenous security features. These features must be capable of detecting threats promptly and addressing them effectively to ensure both security and control. They also provide secure configuration options for the intrinsic security architecture of communication nodes within communication systems. This paper introduces a novel, inherently secure identity verification system named ORCA, which utilizes the concept of noninteractive zero-knowledge proof. ORCA streamlines the process of node identity authentication by minimizing communication exchanges, thereby safeguarding network topology integrity and establishing trust upon network formation. Leveraging this framework, the ORCA-OLSR security protocol was developed. Comparative simulations on the OPNET platform between the ORCA-OLSR and the challenge-response mechanism-based CA-OLSR security protocol demonstrate that ORCA-OLSR outperforms its counterpart in metrics such as routing overhead, end-to-end delay, and routing load.

To overcome the challenges of high feature redundancy, strong subjectivity in manual feature selection, and low model adaptability in malicious encrypted traffic, this paper proposes a hybrid feature selection method based on mutual information and an improved Harris hawk optimization algorithm. This method combines the advantages of filters and wrappers, selects important features in malicious encrypted traffic through two-stage screening, and inputs them into a deep learning model for classification. Experimental results demonstrate that this method can select higher quality feature subsets and improve the classification ability of malicious encrypted traffic detection models.