Abstract With the rapid proliferation of wireless Internet of Things (IoT) devices, there is a growing concern about the potential misuse of wireless IoT devices for unauthorized sensing and monitoring of daily activities. Individuals are increasingly aware of the risks related to unidentified wireless IoT devices. Therefore, there is an urgent need for a comprehensive approach to improve the transparency of wireless IoT devices. In view of this, we propose a mutual detection and identification approach for autonomously detecting and identifying wireless IoT devices based on encrypted traffic analysis. We develop an adaptive device detection algorithm that uses a mutual detection mechanism among devices to facilitate distributed detection tasks and gather encrypted traffic data. We also design a two-stage device identification method to identify the devices. Through experiments conducted on two publicly available datasets, the study achieved a classification accuracy of $ \mathbf{100\%}$ in distinguishing wireless IoT devices from other device types, as well as a high accuracy of $\mathbf{99.59\%}$ in identifying specific types of wireless IoT devices. These results highlight the efficacy and potential of the proposed method in enhancing the security and transparency of wireless IoT ecosystems.

In recent years, deep learning has been widely applied in smart contract vulnerability detection due to its automatic feature extraction and strong generalization capabilities. However, existing methods still face challenges such as redundant information in graph structures, insufficient utilization of data flow information, and single-scale feature extraction. To address these issues, we propose a function-level smart contract graph representation, namely the Multi-relational Semantic Graph (MSG), which employs various types of data flow edges to represent data dependency information within contracts. Subsequently, we introduce a detection model, REA_DCN, which combines a Residual Multi-scale Dilated Convolutional Network with a Multi-head Attention mechanism to capture syntactic and semantic features in the MSG. The model comprises two key modules: the Residual Multi-scale Dilated Convolutional Network (RE_DCN) can extract node features from three different dimensions, while the Multi-head Attention Network (MEA) is utilized for edge feature extraction. Experimental results on real-world datasets demonstrate that the highest score of REA_DCN in terms of accuracy, precision, recall and F1 score exceeds 97%, proving its effectiveness and feasibility.

The widespread adoption of EOSIO blockchain technology has underscored the critical importance of securing EOSIO smart contracts. Vulnerabilities within these contracts can result in substantial economic losses, making their detection a vital area of research. However, existing methods for detecting vulnerabilities in EOSIO smart contracts predominantly rely on expert-defined rules, which are often susceptible to errors and lack scalability. To address this issue, we propose a vulnerability detection method for the EOSIO platform based on a teacher-student network architecture. This approach specifically focuses on detecting vulnerabilities at the function level. This approach consists of two networks. The teacher network learns both the syntax and semantics of source code and bytecode, while the student network takes bytecode as input. The teacher-student network extracts function-level features by incorporating an MCOAttention mechanism. The student network infers missing bytecode embeddings by learning from the teacher network. By combining the inferred source code and bytecode representations, the method achieves improved accuracy in vulnerability detection. We introduce a cross-modal mutual learning strategy to facilitate knowledge transfer between the teacher and student networks. We evaluated our proposed approach using a dataset comprising smart contracts from the EOSIO platform. Experimental results demonstrate that our method significantly improves accuracy in vulnerability detection.

Automated verification of interrupt-driven programs presents significant challenges, as interrupt signals can arrive at arbitrary times and preempt the execution of the current task. This requires considering a vast number of possible execution paths during verification. Furthermore, multiple interrupts may be pending simultaneously, with their service order determined by interrupt priorities. This can lead to nested preemption, further increasing the complexity of the program state space and the difficulty of verification. We propose a formal method based on partial order modeling to verify interrupt-driven programs. Our approach models the interactions among multiple tasks in interrupt-driven programs through partial orders, encoding program execution paths as logical formulas composed of partial order constraints. These formulas are then solved by an SMT solver capable of handling partial order constraints, enabling assertion checking within interrupt-driven programs. We have implemented the proposed method in a prototype tool called DIDP and conducted experiments using a benchmark dataset consisting of real-world embedded system code and device drivers to evaluate its performance. Experimental results demonstrate that, compared to state-of-theart verification tools, DIDP significantly improves verification efficiency while maintaining accuracy.

ABSTRACT Scene memory generation (SMG) refers to training AI agents to recall scene memories similarly to the human brain. This is the key work to realize the artificial memory system. The challenge is to generate scenes rich in motion and keep it realistic while ensuring temporal consistency. Inspired by the principles of memory function in brain neuroscience, this paper proposes a motion‐aware scene generation model named SMG based on motion‐aware temporal style modulation (SMG‐MATSM), which ensures temporal consistency by redesigning the temporal latent representation and constructing a motion matrix to guide the motion of intermediate latent variables. The motion matrix preserves motion consistency in the scene memory through both the cosine similarity and the Mahalanobis distance of intermediate latent variables of adjacent frames. Additionally, SMG‐MATSM uses a style‐based approach and enhances conditional features through the motion matrix during the scene memory synthesis process. Experimental results show that SMG‐MATSM has better effect of action‐enriched scene memory generation, and has varying degrees of efficiency improvement on different datasets with Frechet video distance and Frechet inception distance evaluation metrics.

Vulnerabilities in EOSIO smart contracts have caused significant economic losses. Although some approaches have been proposed to detect these vulnerabilities, they often face several limitations, such as inefficiency in path exploration, insufficient diversity of test cases, and path explosion, which col-lectively reduce code coverage and detection accuracy. Currently, there is a lack of hybrid fuzzing techniques specifically designed for EOSIO smart contracts to address these issues. To fill this gap, we propose a coordination-driven hybrid fuzzing approach for discovering vulnerabilities in EOSIO smart contracts. Our method employs a scheduling strategy using an online linear regression model based on stochastic gradient descent to reduce the edge redundancy detection in hybrid fuzzing and enhance the efficiency of path exploration during symbolic execution. Additionally, a synchronization strategy based on constraint domain abstraction and random walk sampling ensures uniform sam-pling in simplified scenarios, thus improving code coverage and mitigating path explosion. Furthermore, we design a function-level mutation strategy to rapidly diversify test cases in the seed pool, facilitating the efficiency of detecting vulnerabilities. We implement our method in a tool named CDHF and evaluate it on 3,440 smart contracts. Experimental results indicate that CDHF can detect vulnerabilities more precisely and efficiently, achieving an approximate 20 % improvement in code coverage compared to WASAI.

As an effective technology for detecting vulnerabilities, symbolic execution has been applied to detect vulnerabilities in smart contracts. However, it faces challenges such as low detection efficiency. Existing parallel symbolic execution techniques for vulnerability detection of smart contracts lack load balancing mechanism, resulting in unnecessary overhead. To address this issue, we propose a load-balanced parallel symbolic execution approach to accelerate the vulnerability detection of smart contracts. In the approach, the main process first explores and assigns an initial path to each worker process. The worker processes then explore different paths in parallel and utilize a work stealing algorithm to balance the workload among them. By allowing multiple worker processes to explore various paths on different CPU cores and ensuring the load balance, we enhance the efficiency of symbolic execution. We implement a parallel symbolic execution tool called PMyth based on Mythril and evaluate our approach by experimenting on two third-party datasets. The experimental results show that PMyth can provide up to 4.36x acceleration compared to Mythril.

Unsupervised attribute graph representation learning allows for embedding node information into compact vectors without relying on any labels, which greatly facilitates downstream tasks. Graph contrastive learning, founded on the principle of maximizing mutual information, has emerged as a pivotal technique in unsupervised graph representation learning. It achieves node discriminative representations by bringing positive samples closer together and pushing negative samples further apart. However, most existing graph contrastive learning methods primarily concentrate on node-level comparisons, capturing highly abstract node differences to discriminate them, while overlooking the wealth of information present in community substructures within a graph. Additionally, nodes within the same community in a graph often exhibit similar semantics, and considering all other nodes as negative samples unavoidably leads to sampling bias issues. In this work, we propose Community-aware unsupervised graph debiased contrastive representation learning (CAGDCL). Specifically, CAGDCL employs a novel edge-density driven contrastive objective on the augmented graph for community detection to generate robust community prototypes. We introduce the node-prototype contrastive objective based on the node-level contrastive objective. The primary purpose is to encourage the encoder to capture more community-related semantics, enabling the obtained embeddings to maintain intra-community alignment and inter-community uniformity in the embedding space. To mitigate the issue of sampling bias, we propose a weighting scheme of negative samples based on community assignment result and community prototype similarity. Through extensive experiments on several real-world datasets, we demonstrate the effectiveness of CAGDCL.

Message Passing Interface (MPI) is a standard communication protocol utilized in parallel computing. In MPI programs, communication deadlock is one of the most serious problems. To detect deadlocks, existing methods usually traverse all possible execution paths. However, with the increase of wildcard receives, these methods face the problem of path explosion, resulting in low detection efficiency. To address the issue, we propose a deadlock detection approach with path compression and focus matching. In the approach, consecutive identical send operations within a process in an MPI program are combined to a new compressed operation, which reduces the number of communication operations to be analyzed. Then, for each receive operation, a match-set containing all possible compressed send operations from which the receive operation can receive a message is formed based on the Lazy Lamport Clocks Protocol. Finally, a focus matching algorithm based on the match-sets is applied for deadlock detection. We have implemented our approach in a tool called PCMPI and evaluated PCMPI by experimenting on 16 test programs from the Umpire test suite and open-source projects in the real-world. The experimental results demonstrate that PCMPI exhibits higher efficiency in deadlock detection than the two most related tools SAMPI and PDMPI.

The existing anomaly detection methods for IoT devices suffer from several limitations, including inadequate and untimely implementation of security control measures, as well as an inability to promptly mitigate intrusion behaviors. To address these issues, we propose an anomaly detection model for IoT devices based on behavioral feature analysis (ADIoT). The model implements two methods, specifically multidimensional behavior feature selection and anomaly detection. The first method uses the principal component analysis based on weighted contribution(WCPCA) to identify the most representative feature combination from a large number of features and generate a feature dataset that accurately reflects the device behavior patterns. The second method uses an anomaly detection method based on isolation forests to score and identify anomalies in behavioral data and dynamically adjust the anomaly detection threshold. Experimental results demonstrate that ADIoT achieves a detection accuracy of 98.6% on the TON dataset, outperforming state-of-the-art anomaly detection models. Additionally, ADIoT significantly improves time efficiency, reducing test times by 30.79% compared to SVM-based models and by 20.11% compared to CNN-based models.

Symbolic execution is one of the most popular technologies for detecting vulnerabilities in smart contracts, however, the path explosion problem and the timeout problem in solving path constraints hinder the detection efficiency. In order to find possible vulnerabilities in smart contracts faster, we propose a reinforcement learning guided heuristic search strategy for symbolic execution of smart contracts. In the strategy, we employ Q-Iearning as the reinforcement learning algorithm, and use Q-table to provide some suggestions for the path selection of symbolic execution. In addition, to further improve the detection efficiency, an incentive path pruning strategy is also adopted to delete paths which are not relevant to finding vulnerabilities. Moreover, for the timeout problem in solving path constraints, we predict the solution time to determine whether or not to solve a constraint. We have implemented our strategies in a tool called MythrilQL, and evaluated it on benchmarks consisting of four publicly available smart contract datasets and 5 large-scale contracts with more than 800 LOC (line of code). The experimental results show that MythrilQL is more efficiency than the most related tools Mythril and MPro, especially in detecting smart contracts in large scale.

Attacks exploiting vulnerabilities in EOSIO smart contracts have caused serious economic losses. To detect these vulnerabilities, some approaches have been proposed, and concolic fuzzing is one of the most popular techniques among them. However, the existing concolic fuzzers have problems such as path explosion and adopting redundant constraint solving strategies, which reduce the detection efficiency. In order to alleviate these problems, we propose a demand-driven concolic fuzzing approach to discovering vulnerabilities in EOSIO smart contracts. In the approach, execution information is first collected to guide the execution of the system in a demand-driven manner. To improve the efficiency of vulnerability detection, we design a pruning strategy to eliminate the paths that are not relevant to the discovery of vulnerabilities and redundant paths to be explored. Meanwhile, an incremental constraint solving method is used to process only paths that can explore new branches. In addition, we also design a path prioritization method to preferentially explore paths which are more conducive to discovering vulnerabilities, so as to find vulnerabilities in smart contracts as early as possible. We have implemented our approach in a tool called WASAIUP and evaluated it on 3441 smart contracts. The experimental results show that WASAIUP improves the performance by 25.1% to 149.8% compared with the state-of-the-art tool WASAI in terms of efficiency, while maintaining high detection accuracy.

As one of the representative Delegated Proof-of-Stake (DPoS) blockchain platforms, EOSIO blockchain platform is developing rapidly in recent years due to its excellent features, such as the scalability of transaction speed and support for smart contracts and decentralized applications. However, vulnerabilities in EOSIO smart contracts have caused serious economic losses and moreover vulnerability detection tools for EOSIO contracts are limited. To overcome the above shortcomings, we implement a grey-box fuzzer called GFuzzer based on WebAssembly for smart contracts on the EOSIO platform considering that EOSIO contracts are not open-sourced. In order to generate more test cases for branches that are difficult to cover, GFuzzer selects test cases with the minimum distance to explore uncovered branches for mutation. We evaluate GFuzzer on 3963 real-world smart contracts and the experimental results show that GFuzzer can detect more vulnerabilities in EOSIO contracts than the existing tools EOSFuzzer and EVulHunter, and is efficient in achieving high branch coverage during vulnerability detection.

Deadlock is one of the critical problems in the message passing interface. At present, most techniques for detecting the MPI deadlock issue rely on exhausting all execution paths of a program, which is extremely inefficient. In addition, with the increasing number of wildcards that receive events and processes, the number of execution paths raises exponentially, further worsening the situation. To alleviate the problem, we propose a deadlock detection approach called SAMPI based on match-sets to avoid exploring execution paths. In this approach, a match detection rule is employed to form the rough match-sets based on Lazy Lamport Clocks Protocol. Then we design three refining algorithms based on the non-overtaking rule and MPI communication mechanism to refine the match-sets. Finally, deadlocks are detected by analyzing the refined match-sets. We performed the experimental evaluation on 15 various programs, and the experimental results show that SAMPI is really efficient in detecting deadlocks in MPI programs, especially in handling programs with many interleavings.

Smart contracts are commonly used to build finance-related decentralized applications. If a smart contract vulnerability is exploited by an attacker, the contract owner may suffer financial losses. We focus on a particular class of smart contract vulnerabilities that require a specific sequence of multiple transactions to trigger, which we call multi-transaction sequence vulnerabilities. Due to the combinatorial explosion problem caused by the huge number of possible transaction sequences, the efficiency and scalability for existing security analyzers to detect multi-transaction sequence vulnerabilities are limited. To alleviate the problem, we propose a vulnerability detection approach based on symbolic execution and inter-path data dependency. In the approach, we first traverse paths in a contract, and record read and write operations of each path. Then, we selectively execute paths which are conducive to discovering vulnerabilities during the subsequent detection process according to inter-path data dependencies. By pruning out most paths that are not relevant to vulnerabilities, we improve the efficiency and scalability of detecting multi-transaction sequence vulnerabilities. We evaluate our approach on 442 contracts collected from CVE reports and 104 contracts with Ether leakage and suicide defects. The experimental results show that our approach reaches an average 2x speedup comparing to Mythril.

Big data techniques are widely used in various fields. To deal with large data sets efficiently, a new programming framework MapReduce has emerged. Thus, new verification challenges arise to improve the reliability of big data processing. In this article, MapReduce processes are implemented by modeling simulation and verification language programs. Then, several data properties such as data soundness, nonconflict, nonduplication, cooperation, and completeness are taken into account. Moreover, these properties are specified by propositional projection temporal logic formulas. To verify these properties, a runtime verification approach at code level based on unified model checking is employed. In addition, two case studies are conducted to demonstrate our approach: sparse matrix multiplication and tracking down suspected patients of an infectious disease.

Verification of programs at code level has attracted more and more attentions since the cost is high to extract models from source code. Most of approaches available for code level verification are carried out by inserting assertions into programs and then checking whether the assertions are violated. In this way, only safety properties can be verified, however, other temporal properties of programs such as liveness are hard to be verified. To tackle this problem, a novel runtime verification approach, which can verify full regular temporal properties of a program, is proposed in this paper. With this approach, a program to be verified is written in a modeling, simulation and verification language (MSVL) as a program M and a desired property is specified by a propositional projection temporal logic formula P . The negation of the desired property is then translated to an MSVL program M'. Thus, whether M violates P can be checked by evaluating whether there exists an acceptable execution of the new MSVL program “M and M'.” This problem can efficiently be solved with the MSVL compiler where verification cases are generated via dynamic symbolic execution. Further, we adopt parallel mechanism to handle various execution paths of a program for improving the efficiency. The proposed approach has been implemented in a tool called MSV. Experiments show that the performance of MSV outperforms existing tools such as T2, RiTHM, and LTLAutomizer in verifying temporal properties of real-world programs.

It is of great importance to ensure the correctness and reliability of the scheduling protocol of safety-critical systems since the failure will cause serious damage. This paper analyzes a real time scheduling protocol of the safety-critical system and models it using a Modeling, Simulation and Verification Language program. Then the sufficient and necessary conditions for the schedulability are given. Further, the schedulability and other properties are verified using the MSV toolkit.

The paging mechanism is widely used in most modern systems to handle the virtual memory. Many page replacement algorithms have been proposed. Therefore, the cor-rectness and reliability of virtual memory management systems become very important. It is essential to formalize and verify the system in a formal way. In this paper, we model the virtual memory management system with MSVL, which is a parallel programming language used for the modeling, simulation and verification of software and hardware systems. Then we employ the model checking approach based on MSVL to verify the interval related properties and periodic repeated properties of the system.