As intellectual property rights, the copyright protection of deep models is becoming increasingly important. Existing work has made many attempts at model watermarking and fingerprinting, but they have ignored homologous models trained with similar structures or training datasets. We highlight challenges in efficiently querying black-box piracy models to protect model copyrights without misidentifying homologous models. To address these challenges, we propose a novel method called DeepCore, which discovers that the classification confidence of the model is positively correlated with the distance of the predicted sample from the model decision boundary and piracy models behave more similarly at high-confidence classified sample points. Then DeepCore constructs core points far away from the decision boundary by optimizing the predicted confidence of a few sample points and leverages behavioral discrepancies between piracy and homologous models to identify piracy models. Finally, we design different model identification methods, including two similarity-based methods and a clustering-based method, to identify piracy models using the models’ predictions of core points. Extensive experiments show the effectiveness of DeepCore in identifying various piracy models, achieving lower missed and false identification rates, and outperforming state-of-the-art methods. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Flexible job-shop scheduling problem (FJSP) is a well-known and complex variant of the classical Job-shop scheduling Problem (JSP), which finds significant applications in manufacturing industries where flexibility in the use of machines is crucial for enhancing production efficiency and reducing operational costs. Currently, many popular hybrid meta-heuristic algorithms which involve combining two or more metaheuristic strategies such as genetic algorithm (GA) and tabu search (TS) procedures, often leads to improved performance in terms of solution quality and computational efficiency. In this paper, we design an improved hybrid genetic algorithm with parallel acceleration (HFGA) for dynamic FJSP with new job insertion, which can further speed up the execution of each operator including selection, crossover, mutation and local search. Experimental results show that our proposed hybrid algorithm with parallel acceleration demonstrates quite superior performance. © 2024 Copyright held by the owner/author(s).

Internet of Things (IoT) device identification has become an indispensable prerequisite for secure network management and security policy implementation. However, existing passive device identification methods work under a "closed-world" assumption, failing to take into account the emergence of new and unfamiliar devices in open scenarios. To combat the open-world challenge, we propose a novel evolutionary model which can continuously learn with new device traffic. Our model employs a decoupled architecture suitable for evolutionary learning, which consists of device feature representation and device inference. For device feature representation, an auto-encoder based on metric learning is innovatively introduced to mine latent feature representation of device traffic and form independent compact clusters for each device. For device inference, the nearest class mean (NCM) classification strategy is adopted on the feature representation. In addition, to alleviate the forgetting of old devices during evolutionary learning with new devices, we develop a less-forgetting constraint based on spatial knowledge distillation and impose control on the distribution distance between clusters to reduce inter-class interference. We evaluate our method on the union of three public IoT traffic datasets, in which the accuracy is as high as 87.9% after multi-stage evolutionary learning, outperforming all state-of-the-art methods under diverse experimental settings. © 2023, ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering.

In modern society, smart home has become a part of people's daily life. The smart home will generate a large amount of data to be shared inside and outside the home which is sensitive and private. Therefore, access control plays a critical role in the many issues that need to be addressed in the smart home. However, there is still a lack of an access control scheme suitable for the smart home to ensure the security of users' data. Attribute-based access control(ABAC) is a flexible authorization model. It controls whether there is permission to operate objects through the attributes of entities, operation types, and related environments, but it cannot resist malicious attacks within the environment. Therefore, in this paper, we propose a trust-based ABAC model, which is extended on the basis of the ABAC model and applies the trust model to the smart home ABAC model to achieve fine-grained and effective access control. Our model can not only use the trust model to monitor the behavior deviation of devices in smart homes, and establish trust relationships between devices, but also control the permissions of each device. Finally, we test the trust model on the ns-3 platform and simulate the access control model on the small real Internet of Things(IoT). The results show that the trust model has good accuracy, convergence, and anti-attack, and the ABAC model has good applicability in smart home scenarios. © 2023 IEEE.

It is essential to identify potentially vulnerable code in our software systems. Deep neural network techniques have been used for vulnerability detection. However, existing methods usually ignore the feature representation of vulnerable datasets, resulting in unsatisfactory model performance. Such vulnerability detection techniques should achieve high accuracy, relatively high true-positive rate, and low false-negative rate. At the same time, it needs to be able to complete the vulnerability detection of actual projects and does not require additional expert knowledge or tedious configuration. In this article, we propose and implement VDDRL (A Vulnerability Detection Method Based On Deep Representation Learning). This deep representation learning-based vulnerability detection method combines feature extraction and ensemble learning. VDDRL uses the word2vec model to convert the source code into a vector representation. Deep representations of vulnerable code are learned from vulnerable code token sequences using LSTM models and then trained for classification using traditional machine learning algorithms. The training dataset we use is derived from actual projects and contains seven different types of vulnerabilities. Through comparative experiments on datasets, VDDRL achieves an Accuracy of 95.6%–98.7%, a Precision of 91.6%–99.0%, a Recall of 84.7%–99.5%, and an F1 of 88.1%–99.2%. Both perform better than the baseline method. Our experimental results show that VDDRL is a generic, lightweight, and extensible vulnerability detection method. Compared with other methods, it has better performance and robustness. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

In edge computing, computing, storage, and network services are provided at the edge of the network. If the edge server is configured with the appropriate virtualization services, end-device requests can be processed at the edge to save network bandwidth and reduce response time. It is necessary for edge devices to run for a long period of time in order to receive user requests, however, users do not always send requests, which may result in some edge devices being idle for a long time and causing a significant amount of idle energy consumption.In this paper, we design a sleep-mode based edge computing architecture that uses containers as resource units. Also, based on the proximal policy optimization algorithm (PPO), we propose a task scheduling algorithm using this edge computing architecture, namely Proximal Policy Optimization Task Scheduling based on Sleep Mode (PPO-TSSM). Our goal is to minimize the total utility of the task; where the total utility is the tradeoff between completion time, energy consumption and deadline. The experimental results show that our proposed algorithm PPO-TSSM reduces the total cost by at least 3.33% (up to 62.92%) compared with the baseline algorithm, and reduces the minimum completion time by 5.37% to 157.7%, and the energy consumption by 2.95% to 8.97% compared with other baseline algorithms. Meanwhile, we set up PPO-TS-NS without sleep mode to compare with the environment using sleep mode, and find that PPO-TSSM reduces the energy consumption by 74.76% compared with PPO-TS-NS, while the minimum completion time only increases by 5.33%. Therefore, introducing sleep mode can effectively reduce the cost of the whole edge computing system. © 2023 IEEE.

In smart transportation system, the traffic flow is one of the important metrics to mature road traffic. Along with increasing urbanization, road traffic becomes a thorny issue to limit the development of a city. As a result of technological advancements, the transportation system is enriched with smarter devices that help it perform more efficiently and effectively. One efficient approach is leveraging existing data to predict the road traffic flow, in order to optimize road usage efficiency. There are many studies to leverage existing data to predict road traffic flow. However, the limitation is the generic traffic model cannot represent real dynamic traffic flow precisely. In this study, we analyze macroscopic traffic flow and traffic density, as well as microscopic speed, acceleration, and distance for each individual vehicle in the smart transportation context. Based on the analysis, we propose a macroscopic traffic flow model. Focus on the limitation of existing research, we also involve lane switching in the model and propose the lane switching decision making strategy. After we optimize the traffic model, we adopt BP neural network to predict the road traffic flow. The evaluation results show that our multi-feature traffic model can increase the prediction accuracy on real road traffic datasets. © 2023 IEEE.

We report a 30-month nationwide deployment and operation study of an indoor arrival detection system based on Bluetooth Low Energy called VALID in 364 Chinese cities. VALID is pilot-studied, deployed, and operated in the wild to infer real-time indoor arrival status of couriers, and improve their status reporting behavior based on the detection. During its full nationwide operation (2018/12- 2021/01), VALID consists of virtual devices at 3 million shops and restaurants, where 530,859 of them are in multi-story malls and markets to infer and influence 1 million couriers' behavior, and assist the scheduling of 3.9 billion orders for 186 million customers. Although indoor arrival detection is straightforward in controlled environments, the scale of our platform makes the cost prohibitively high. In this work, we explore to use merchants' smartphones under their consent as a virtual infrastructure to design, build, deploy, and operate VALID from in-lab conception to nationwide operation in three phases for 30 months. We consider metrics including system evolution, reliability, utility, participation, energy, privacy, monetary benefits, along with couriers' behavior changes. We share three lessons and their implications for similar wireless sensing or communication systems with large geospatial operations. © 2021 ACM.

We report a 3-year city-wide study of an operational indoor sensing system based on Bluetooth Low Energy (BLE) called aBeacon (short for alibaba Beacon). aBeacon is pilot-studied, A/B tested, deployed, and operated in Shanghai, China to infer the indoor status of Alibaba couriers, e.g., arrival and departure at the merchants participating in the Alibaba Local Services platform. In its full operation stage (2018/01-2020/04), aBeacon consists of customized BLE devices at 12,109 merchants, interacting with 109,378 couriers to infer their status to assist the scheduling of 64 million delivery orders for 7.3 million customers with a total amount of $\$ $ 600 million order values. Although in an academic setting, using BLE devices to detect arrival and departure looks straightforward, it is non-trivial to design, build, deploy, and operate aBeacon from its conception to its retirement at city scale in a metric-based approach by considering the tradeoffs between various practical factors (e.g., cost and performance) during long-term system evolution. We report our study in two phases, i.e., an 8-month pilot study and a 28-month deployment and operation in the wild. We focus on an in-depth reporting on the five lessons learned and provide their implications in other systems with long-term operation and broad geospatial coverage, e.g., Edge Computing. © 1993-2012 IEEE.

Container technology has been widely deployed in edge computing environments. Using the OS-level resource isolation and management, it can achieve incomparable high efficiency in contrast to the traditional virtual machine approach. However, recent studies have shown that the container environment is vulnerable to various security attacks. Furthermore, the highly customizable and dynamic change nature of the container exacerbated its vulnerability. In this paper, we propose a new anomaly detection framework combining cluster algorithm to improve anomaly detection efficiency in the edge computing environment that contains a large number of containers. It utilizes cluster algorithm to automatically identify containers that running the same application, and builds an anomaly detection model for each category separately. We investigated 8 real-world vulnerabilities from several frequently used applications and evaluated our framework on them. Experiment results show that our proposed framework can effectively identify containers built on the same application image without any manual labeling, reduce the FPR of anomaly detection from 0.61% to 0.09%, and increase TPR from 90.3% to 96.2% compared to the traditional method. © 2022 IEEE.