Public Key Infrastructure (PKI) is the cornerstone of network trust, yet its centralized architecture suffers from inherent risks such as single points of failure, fragile trust chains, and inefficient certificate revocation. Although blockchain technology enables decentralized PKI (DPKI) by providing transparent, tamper-resistant auditing and automated execution, existing schemes in multi-Certificate Authority (CA) settings still lack robust and adaptive governance mechanisms that can cope with dynamic CA performance fluctuations and malicious behaviors. To address this gap, we propose DRL-DPKI, a learning-based governance framework that integrates deep reinforcement learning to support performance-driven CA selection, quality-aware load balancing, and malicious-behavior mitigation. DRL-DPKI formulates CA committee election as a sequential decision-making problem, where the agent's state space fuses normalized reputation, recent success rate and delay level computed over a temporal sliding window, and a "hunger" measure that enhances sensitivity to short-term behavior changes. Experimental results show that, compared with seven representative baselines, DRL-DPKI substantially reduces the malicious-CA selection rate under high adversarial presence; in our scalability experiments with CA pools of up to 6000 and a 50% malicious-CA regime, the malicious-CA selection rate remains within 0-13.33% across the evaluated scales, and DRL-DPKI significantly outperforms all baselines in terms of robustness and quality-aware load balancing. This work introduces an adaptive intelligent decision-making paradigm for trust governance in decentralized PKI, substantially improving the security and resource utilization efficiency of DPKI systems.

The exponential growth of Internet of Things (IoT) devices has led to the proposal of edge computing for data processing. The decentralized nature of edge computing servers and IoT devices makes blockchain ideal for connecting IoT users and servers. The consensus protocol, a core technology in blockchain, ensures node agreement and operational efficiency. However, as node numbers increase and spread geographically, traditional consensus protocols face deployment challenges, significantly reducing efficiency. To address this, we propose a hierarchical blockchain consensus protocol based on geographic location and a trust model called location and trust delegated-BFT (LT-DBFT). This protocol reduces global communication delays caused by wide geographic distribution by assigning consensus nodes to different clusters based on their locations. A trust model is also designed to elect active nodes for consensus voting, thereby reducing the consensus overhead within clusters. Subsequently, the primary nodes of each cluster form a global shared layer to achieve the final ordering and execution of transactions. Through theoretical and experimental analysis, our scheme demonstrates lower latency and higher throughput performance than traditional practical Byzantine fault tolerance and GeoBFT, making it more suitable for efficient deployment in large-scale and geographically widespread IoT environments.

The issue of transaction order fairness in blockchain systems has become increasingly prominent, especially decentralized finance (DeFi) scenarios, where malicious participants may manipulate transaction ordering undue gains. Existing solutions for fair transaction ordering either rely on synchronous clock assumptions suffer from high computational complexity and significant confirmation delays. This paper introduces Dikaios, fair consensus protocol based on position anchoring and replica reputation. We propose the concept of Position Order-Fairness, categorizing transactions into Fully Determined and Undetermined types: Fully Determined transactions are directly locked into their positions in the global sequence, while Undetermined transactions processed using a grouped ordering method combined with a node reputation mechanism to resolve potential Condorcet cycles. Theoretical analysis and experimental results demonstrate that, compared to existing solutions, Dikaios ensures order fairness while significantly improving system performance. Experimental results show that Dikaios achieves a 220% improvement in throughput and reduces confirmation latency a factor of three compared to the state-of-the-art order fairness protocols.

Today, Ethereum is the world’s largest open-source platform. However, because smart contracts hold a large amount of money and cannot be changed once on the chain, they have become the target of attackers. Users will undoubtedly suffer significant financial losses. To counter these attacks, various methods have been proposed to scan smart contracts for vulnerabilities before deploying them on the blockchain, but few of them determine smart contracts to be vulnerable by examining transactions. In this paper, we propose a framework called EAOS to detect attacks through analyzing the opcode sequences executed by EVM. We first obtain the opcode sequences with operands of smart contracts during EVM execution by replaying the historical transactions of Ethereum, and then extract the feature opcodes from the opcode sequences to generate the feature opcode sequences. Next, we provide some very useful APIs to make it easier for users to get the data related to the opcode sequences. Based on the APIs, users can develop various algorithms to detect attacks. Finally, to verify the effectiveness of EAOS, five algorithms are developed to analyze the replayed transaction opcode sequences. The extensive experimental results demonstrate the effectiveness and efficiency of EAOS and our detection algorithms.

With the development of Vehicle Edge Computing (VEC), vehicles can offload tasks to Road Side Units (RSU) for processing. However, optimizing latency is a challenging problem in high-speed environments. This paper investigates the latency optimization problem for task offloading in high-speed environments, where multiple vehicles on the user end can choose to offload their tasks to either RSUs or Base Station (BS) for computation. Meanwhile, based on the vehicle’s location and speed, we predict the time a vehicle spends within the communication range of an RSU to select a more suitable edge node. Then, an improved Group Priority Sampling-Double Dueling Deep Q Network (GPS-D3QN) algorithm is used to optimize the proposed task offloading, resulting in an optimal offloading strategy that reduces latency. Through simulation, by varying the number of vehicles and RSUs, the proposed optimization scheme is verified to have an advantage in terms of delay compared to other schemes.

Edge computing is gaining widespread attention, particularly in the domain of task offloading. Efficiently and accurately offloading tasks to edge servers presents a significant challenge due to its complexity. In this paper, we depart from the traditional user-to-edge offloading scenario and instead focus on a model where users offload tasks to base stations. These base stations, equipped with servers and possessing an extended communication range, centrally manage all tasks. A base station can either offload tasks to a specific edge server within its communication range or handle them locally. To address the scheduling challenge among multiple tasks, we propose a priority-based Time Slice Scheduling (PTSS) algorithm to calculate task processing latency. Additionally, we introduce the PTSS-DQN algorithm by integrating the Deep Q-Network (DQN) to jointly optimize the offloading strategy, resource allocation, time-slice distribution, and task execution order. Experimental results demonstrate the effectiveness of our proposed approach.

Ethereum is a blockchain network that allows developers to create smart contracts and programs that run on the blockchain. Smart contracts contain logic to transfer assets based on pre-defined conditions. With over 100,000 new smart contracts being deployed every day, the potential for coding errors is high, making the contracts vulnerable to exploits. A key limitation is that once deployed, smart contracts are immutable and cannot be updated, even if flaws are found. This inflexibility puts funds at risk of theft and loss. The rapid pace of deployment outpaces security audits, increasing vulnerabilities that put users’ cryptocurrency at risk. To reduce the risk caused by smart contract vulnerabilities, we applied deep learning techniques. To develop a deep learning model capable of detecting vulnerabilities, we first created a dataset by replaying real transactions on the Ethereum Mainnet, collecting opcode sequences from real Ethereum contracts, and labeling them using the SODA plugin. We pre-processed this opcode data by removing duplicates, normalizing sequence lengths, simplifying opcodes into representative groups, and converting sequences into numerical vectors to ultimately obtain an optimal representation of the data. We then trained and evaluated three different neural network architectures on this dataset. Our best-performing model achieved an average accuracy of 88

The continuous evolution of network threats has made Cyber Threat Intelligence (CTI) sharing an essential strategy to counter these challenges. However, existing CTI data-sharing models face challenges such as establishing and maintaining trust, lacking flexible sharing strategies, and lacking mechanisms for dispute resolution. To address these issues, this paper proposes a reliable and transparent CTI data-sharing framework based on blockchain (RTB-RM). RTB-RM utilizes a Multidimensional Reputation Model (MRM) to assess user trustworthiness, thus enhancing the credibility of CTI data. It adopts a selective data-sharing approach based on trust levels and sensitivity, balancing the demands of data-sharing and security considerations. Additionally, RTB-RM employs real-time arbitration for dispute resolution. Our experiments validate the cost-effectiveness of RTB-RM and the effectiveness and rationality of MRM in evaluating user reputations and parameter configurations. RTB-RM demonstrates the immense potential and advantages of blockchain technology and multidimensional reputation models in enhancing the reliability and transparency of CTI data sharing.

With the fast growth of blockchain technology, blockchain as a decentralized distributed ledger technology has become more widely used and is gradually changing our way of life. But it also raises more and more security issues. As there are more and more smart contracts on the blockchain, and smart contracts cannot be changed once they are added to the blockchain, there is an opportunity for hackers to attack smart contracts. If not handled properly, it will cause serious economic losses to users. In this paper, we introduce a unique method for identifying abnormal behaviors of smart contract vulnerabilities using opcode sequences. We aim to identify the control flow paths triggered by transactions to capture the abnormal behaviors of smart contracts. The control flow paths are the traces on which the transaction is executed. Using Geth instrumentation, we collect the opcode sequences executed on the traces to represent the control flow paths. It should be noted that the process of detecting abnormal behaviors introduces some additional time overhead. However, our experimental results show that this method achieves high abnormal detection accuracy with minimal overhead. This suggests that our proposed method is effective in identifying potential security issues in smart contracts without significantly impacting the overall execution time.

Ethereum is a blockchain platform that allows developers to create smart contracts. Smart contracts are programs that can automatically execute and handle cryptocurrency funds. However, over a hundred thousand new smart contracts are deployed every day and inevitably contain vulnerabilities due to programming errors. Once deployed, smart contracts cannot be fixed or changed, leaving funds at risk. To mitigate it, we use deep learning to detect vulnerabilities in smart contracts. First, we create our own dataset of labeled smart contracts based on opcode sequences, since few smart contract codes and labeled datasets are publicly available. We collect opcode sequences by replaying real-world transactions from the Ethereum Mainnet in our fully synchronized node while we leverage a plugin called "SODA" to label opcode sequences with vulnerability classes. Second, after data collection, we preprocess the data by removing duplicate opcode sequences, normalizing the sequences to the same length, and converting them into vectors. Finally, to detect vulnerabilities in smart contracts, we train a deep classification model using LSTM neural networks. Our model achieved an average accuracy of 82.63% and an F1-score of 79.74% across seven types of vulnerabilities, which is important for securing funds and logic in smart contracts.

Unknown vulnerabilities, also known as zero-day vulnerabilities, are vulnerabilities in software, systems, or networks that have not yet been publicly disclosed or fixed. If these vulnerabilities are ever discovered by hackers, intentionally or unintentionally, they pose a major threat to network security. This is particularly true in the blockchain field, as smart contracts hold a lot of money, and if they are discovered and exploited by hackers, the financial losses to users will be even greater. However, the current research on smart contract vulnerabilities mainly focuses on known vulnerabilities, and the research on unknown vulnerabilities has been limited. Based on this, we introduce a machine learning-based method for detecting unknown vulnerabilities in smart contracts. First, the method obtains the opcode sequences executed by smart contract transactions in the EVM by instrumenting Geth and replaying the Ethereum transactions. Next, we employ an n-gram model and a vector weight penalty mechanism to extract the opcode sequence features. We then use machine learning algorithms to detect unknown vulnerabilities based on the similarity principle. Finally, we test the effectiveness of our method with four machine learning models: the K-Nearest Neighbor algorithm (KNN), Support Vector Machine (SVM), Logistic Regression (LR), and Decision Tree (DT). The SVM model performs best at detecting unknown vulnerabilities, with an accuracy of 96%, a precision of 91%, a recall of 100%, and an F1-score of 95%. We also discuss the benefits of the method: timely detection of attacks due to unknown vulnerabilities, thus reducing user losses.

Data sharing has received much attention and research as an excellent way to unlock the value of data. Trusted data storage servers will participate in the data-sharing system to provide users with convenient data access and storage services. Currently, the vast majority of researchers design data-sharing systems based on centralized trusted authorities and key management centers, but they tend to ignore the problems of trust dependency and data leakage that exist in centralized trust and lead to the overall untrustworthiness of the system. To solve the above problems, this paper considers the use of public key infrastructure (PKI) to provide trusted authentication for data-sharing entities, but the traditional PKI has CA root trust and scenario adaptation problems, so we optimize the traditional PKI model for data-sharing scenarios and call it an improved PKI. Combining the decentralized trust property of blockchain, this paper proposes a TDS-NA scheme based on blockchain and improved PKI to build a distributed trusted, and secure data-sharing system in a semi-trusted network environment. TDS-NA can secure shared data in data sharing, and provide digital certificates that support entity-trusted authentication and reliable access control while designing digest blocks for efficient data auditing. In this paper, we demonstrate that the TDS-NA scheme is able to resist man-in-the-middle attacks and certificate forgery attacks through formal security analysis while satisfying the necessary security properties of data-sharing systems. We implement a prototype of the TDS-NA scheme in ethereum smart contracts and finally verify the security and feasibility of TDS-NA through experimental comparison and analysis.

The emergence of smart contracts has led to the wider use of blockchain, which involves a large number of virtual currency transactions. Because smart contracts are inevitably written with some vulnerabilities, which makes them vulnerable to attacks that cause property damage, and existing detection techniques and static analysis methods mainly target known vulnerability detection. We design a machine learning-based unknown vulnerability detection scheme using opcode sequences. The scheme first obtains opcode sequences of the execution path of contract transactions in the Ethereum virtual machine (EVM) by replaying them in Ethereum. Then we use a combination of N-gram model and a vector weight penalty mechanism to extract opcode features. Finally, we validate the effectiveness of our scheme by three machine learning models, namely the K-Near Neighbor Algorithm (KNN), the Support Vector Machine (SVM), and the Logistic Regression (LR). The SVM model achieves an accuracy of 91.4

Reinforcement learning is a machine learning method that relies on the agent to learn by trial and error to solve decision optimization problems. It is well known that an agent based on deep reinforcement learning in complex environments is difficult to train. Moreover, the agent will generate unsafe and strange actions due to the lack of sufficient reward feedback from the environment. To make the agent converge to a better policy and make its behavior safer and more controllable under sparse rewards, we propose a subgoal embedding method based on prior knowledge and hierarchical strategy that can make the training process converge faster. The subgoal embedding method can be combined with existing reinforcement learning methods. In this paper, we combine the subgoal embedding method with REINFORCE algorithm and PPO(Proximal Policy Optimization) algorithm to test the method in the MiniGrid-DoorKey game environment of the gym platform. The experiments demonstrate the effectiveness of the subgoal embedding method.

In recent years, an increasing number of reports have explored the wound healing mechanism of these two traditional Chinese herbal medicines- Panax ginseng and Panax notoginseng, but there is no systematic research on the related core functions and different mechanisms in the treatment of wound healing up to now. Based on network pharmacology and meta-analysis, the present work aimed to comprehensively review the commonality and diversity of P. ginseng and P. notoginseng in wound healing. In this study, a wound healing-related "ingredients-targets" network of two herbs was constructed. Thereafter, meta-analysis of the multiple target lists by Metascape showed that these two medicines significantly regulated blood vessel development, responses to cytokines and growth factors and oxygen levels, cell death, cell proliferation and differentiation, and cell adhesion. To better understand the discrepancy between these two herbs, it was found that common signaling pathways including Rap1, PI3K/AKT, MAPK, HIF-1 and Focal adhesion regulated the functions listed above. In parallel, the different pathways including renin-angiotensin system, RNA transport and circadian rhythm, autophagy, and the different metabolic pathways may also explained the discrepancies in the regulation of the above-mentioned functions, consistent with the Traditional Chinese Medicine theory about the effects of P. ginseng and P. notoginseng.

In the case of limited endurance of unmanned aerial vehicles (UAVs), in order to further improve UAV data collection efficiency, this paper puts forward EDC-UAVIIoT: an enhanced energy-efficient data collection optimization algorithm for UAV swarm in the intelligent Internet of Things. First of all, the algorithm optimizes the UAV cruise path through the intelligent Internet of Things routing mechanism, avoids the occurrence of data errors in the packet transmission process, and uses the end-to-end transmission error probability model. The error probability of data packets in the transmission process is calculated to improve the efficiency of data collection tasks and data throughput. Secondly, considering the relationship between energy harvesting and energy consumption balance, this paper uses semi-definite programming and a convex approximation algorithm to transform the non-convex optimization problem into a convex optimization problem and realize the mapping relationship between the UAV cluster node and the target node coordinates, which reduces the computational complexity. Finally, the simulation results show that the EDC-UAVIIoT algorithm is compared with other algorithms in network energy consumption, running time, network delay, and network throughput. The numerical values are increased by 7.03%, 10.16%, 12.39%, and 8.82%, respectively, thus verifying the effectiveness and stability of the proposed EDC-UAVIIoT algorithm.

The problem of information leakage is becoming more and more serious, and how to effectively protect the security of personal privacy information has become an urgent problem at present. To this end, this paper proposes an interactive query differential privacy protection model based on big data analysis. The model is based on the irrelevance processing of differential privacy association rules to find out the hidden association information among trajectory data sets. The model uses the shared prefix method to de-compress the trajectory dataset deeply. The model constructs prefix trees to clarify useful information and remove redundant trajectory data. The model constructs a parallel gradient descent matrix decomposition algorithm based on the low-rank mechanism and alternating direction multiplier method to remove the load matrix built from the initial results and decompose the corresponding irrelevant load matrix. The model then takes reasonable values according to the user authority level to limit the privacy budget maximum. The model uses Laplace’s principle to add the determined reasonable noise to the differential privacy trajectory data, reduce and delete the irrelevant attribute trajectory data, and finally feed the query results to the user. Through simulation and comparison experiments, we conclude that the proposed model can make the privacy information available to the highest degree and minimize the success probability of stealers, and thus the differential privacy track data can be protected to the best effect.

With the development of the Internet of Things (IoT) and mobile edge computing (MEC) technologies, increasing missions are transferred to edge servers for computing. An ensuing problem is how to offload tasks to each edge server to reduce time delay or energy expenditure. Therefore, the offloading strategy plays a vital role in the offloading calculation process. To maximize energy efficiency of the offloading system, a computational offloading strategy (ISSA) based on the sparrow search algorithm is proposed in this paper. Firstly, we modeled the actual problem as a delay model and an energy consumption model. Then a penalty function was introduced to balance the delay and energy consumption, and a fitness function was obtained. Secondly, we improved the sparrow search algorithm which will be later used to solve the model. Finally, through experimental simulation, the energy consumption data obtained by the local offloading strategy, the random offloading strategy, the offloading strategy derived from Particle Swarm Optimization (PASO) and the ISSA were discussed. The experimental results demonstrate the superiority of the modified method.

Artificial intelligence (AI) technology is booming in information society, and its role has become one of the hot research topics in the field of image identification. Flowers play essential role in daily life, but the identification and know well of some flower species is not clear. Therefore, the system is of great significance to help people obtain the detailed information of plant flowers. The proposed method is to use machine learning algorithm to process and identify flower images, mainly using transfer learning model, based on the trained model parameters, and taking 25 reptiles as the training base. After the optimal training and parameters adjustment, the extracted features were transferred to the flower model. The experimental results show that the effective recognition accuracy of different flowers is up to 98.20%. The proposed algorithm can be adopted to the small-scale training task, and can reduces the amount of machine operation. In addition, the algorithm also can obtain satisfactory recognition accuracy within a short time. The flower information is presented in the form of dynamic web pages, which improved readability of idenfication results.