Computing最新文献

In recent years, containerization is becoming more and more popular for deploying applications and services and it has significantly contributed to the expansion of edge computing. The demand for effective and scalable container image management, however, increases as the number of containers deployed grows. One solution is to use a localized Docker registry at the edge, where the images are stored closer to the deployment site. This approach can considerably reduce the latency and bandwidth required to download images from a central registry. In addition, it acts as a proactive caching mechanism by optimizing the download delays and the network traffic. In this paper, we introduce an edge-enabled storage framework that incorporates a localized Docker registry. This framework aims to streamline the storage and distribution of container images, providing improved control, scalability, and optimized capabilities for edge deployment. Four demanding XR applications are employed as use cases to experiment with the proposed solution.

Wireless Body Area Networks (WBANs) are wireless sensor networks that monitor the physiological and contextual data of the human body. Nodes in a WBAN communicate using short-range and low-power transmissions to minimize any impact on the human body’s health and mobility. These transmissions thus become subject to failures caused by radiofrequency interference or body mobility. Additionally, WBAN applications typically have timing constraints and carry dynamic traffic, which can change depending on the physiological conditions of the human body. Several approaches for the Medium Access Control (MAC) sublayer have been proposed to improve the reliability and efficiency of the WBANs. This paper proposes and uses a systematic literature review (SLR) method to identify, classify, and statistically analyze the published works with MAC approaches for WBAN efficiency and reliability under dynamic network traffic, radiofrequency interference, and body mobility. In particular, we extend a traditional SLR method by adding a new step to select publications based on qualitative parameters. As a result, we identify the challenges and proposed solutions, highlight advantages and disadvantages, and suggest future works.

Computer system resilience refers to the ability of a computer system to continue functioning even in the face of unexpected events or disruptions. These disruptions can be caused by a variety of factors, such as hardware failures, software glitches, cyber attacks, or even natural disasters. Modern computational environments need applications that can recover quickly from major disruptions while also being environmentally sustainable. Balancing system resilience with energy efficiency is challenging, as efforts to improve one can harm the other. This paper presents a method to enhance disaster survivability in microservice architectures, particularly those using Kubernetes in cloud-based environments, focusing on optimizing electrical energy use. Aiming to save energy, our work adopt the consolidation strategy that means grouping multiple microservices on a single host. Our aproach uses a widely adopted analytical model, the Generalized Stochastic Petri Net (GSPN). GSPN are a powerful modeling technique that is widely used in various fields, including engineering, computer science, and operations research. One of the primary advantages of GSPN is its ability to model complex systems with a high degree of accuracy. Additionally, GSPN allows for the modeling of both logical and stochastic behavior, making it ideal for systems that involve a combination of both. Our GSPN models compute a number of metrics such as: recovery time, system availability, reliability, Mean Time to Failure, and the configuration of cloud-based microservices. We compared our approach against others focusing on survivability or efficiency. Our approach aligns with Recovery Time Objectives during sudden disasters and offers the fastest recovery, requiring 9% less warning time to fully recover in cases of disaster with alert when compared to strategies with similar electrical consumption. It also saves about 27% energy compared to low consolidation strategies and 5% against high consolidation under static conditions.

We present a simple and quick method to approximate network centrality indexes. Our approach, called QuickCent, is inspired by so-called fast and frugal heuristics, which are heuristics initially proposed to model some human decision and inference processes. The centrality index that we estimate is the harmonic centrality, which is a measure based on shortest-path distances, so infeasible to compute on large networks. We compare QuickCent with known machine learning algorithms on synthetic network datasets, and some empirical networks. Our experiments show that QuickCent can make estimates that are competitive in accuracy with the best alternative methods tested, either on synthetic scale-free networks or empirical networks. QuickCent has the feature of achieving low error variance estimates, even with a small training set. Moreover, QuickCent is comparable in efficiency—accuracy and time cost—to more complex methods. We discuss and provide some insight into how QuickCent exploits the fact that in some networks, such as those generated by preferential attachment, local density measures such as the in-degree, can be a good proxy for the size of the network region to which a node has access, opening up the possibility of approximating expensive indices based on size such as the harmonic centrality. This same fact may explain some evidence we provide that QuickCent would have a superior performance on empirical information networks, such as citations or the internet. Our initial results show that simple heuristics are a promising line of research in the context of network measure estimations.

Software requirements play a vital role in ensuring a software product’s success. However, it remains a challenging task to implement all of the user requirements, especially in a resource-constrained development environment. To deal with this situation, a requirements prioritization (RP) process can help determine the sequence for the user requirements to be implemented. However, existing RP techniques are suffered from some major challenges such as lack of automation, excessive effort, and reliance on stakeholders’ involvement to initiate the process. This study intends to propose an automated requirements prioritization approach called association rule mining-oriented (ARMO) to address these challenges. The automation process of the ARMO approach incorporates activities to first pre-process the requirements description and extract features. The features are then examined and analyzed through the applied rule mining technique to prioritize the requirements automatically and efficiently without the involvement of stakeholders. In this work, an evaluation model was further developed to assess the effectiveness of the proposed ARMO approach. To validate the efficacy of ARMO approach, a case study was conducted on real-world software projects grounded on the accuracy, precision, recall, and f1-score measures. The promising experimental results demonstrate the ability of the proposed approach to prioritize user requirements. The proposed approach can successfully prioritize user requirements automatically without requiring a significant amount of effort and stakeholders’ involvement to initiate the RP process.

Code smell identification is crucial in software maintenance. The existing literature mostly focuses on single code smell identification. However, in practice, a software artefact typically exhibits multiple code smells simultaneously where their diffuseness has been assessed, suggesting that 59% of smelly classes are affected by more than one smell. So to meet this complexity found in real-world projects, we propose a multi-label learning-based approach to identify eight code smells at the class-level, i.e. the most sever software artefacts that need to be prioritized in the refactoring process. In our experiments, we have used 12 algorithms from different multi-label learning methods across 30 open-source Java projects, where significant findings have been presented. We have explored co-occurrences between class code smells and examined the impact of correlations on prediction results. Additionally, we assess multi-label learning methods to compare data adaptation versus algorithm adaptation. Our findings highlight the effectiveness of the Ensemble of Classifier Chains and Binary Relevance in achieving high-performance results.

The advancement of technology allows for easy adaptability with IoT devices. Internet of Things (IoT) devices can interact without human intervention, which leads to the creation of smart cities. Nevertheless, security concerns persist within IoT networks. To address this, Software Defined Networking (SDN) has been introduced as a centrally controlled network that can solve security issues in IoT devices. Although there is a security concern with integrating SDN and IoT, it specifically targets Distributed Denial of Service (DDoS) attacks. These attacks focus on the network controller since it is centrally controlled. Real-time, high-performance, and precise solutions are necessary to tackle this issue effectively. In recent years, there has been a growing interest in using intelligent deep learning techniques in Network Intrusion Detection Systems (NIDS) through a Software-Defined IoT network (SDN-IoT). The concept of a Wireless Network Intrusion Detection System (WNIDS) aims to create an SDN controller that efficiently monitors and manages smart IoT devices. The proposed WNIDS method analyzes the CSE-CIC-IDS2018 and SDN-IoT datasets to detect and categorize intrusions or attacks in the SDN-IoT network. Implementing a deep learning method called Bidirectional LSTM (BiLSTM)--based WNIDS model effectively detects intrusions in the SDN-IoT network. This model has achieved impressive accuracy rates of 99.97% and 99.96% for binary and multi-class classification using the CSE-CIC-IDS2018 dataset. Similarly, with the SDN-IoT dataset, the model has achieved 95.13% accuracy for binary classification and 92.90% accuracy for multi-class classification, showing superior performance in both datasets.

Context-aware data management becomes the focus of several research efforts, which can be placed at the intersection between the Internet of Things (IoT) and Edge Computing (EC). Huge volumes of data captured by IoT devices are processed in EC environments. Even if edge nodes undertake the responsibility of data management tasks, they are characterized by limited storage and computational resources compared to Cloud. Apparently, this mobilises the introduction of intelligent data selection methods capable of deciding which of the collected data should be kept locally based on end users/applications requests. In this paper, we devise a mechanism where edge nodes learn their own data selection filters, and decide the distributed allocation of newly collected data to their peers and/or Cloud once these data are not conformed with the local data filters. Our mechanism intents to postpone final decisions on data transfer to Cloud (e.g., data centers) to pervasively keep relevant data as close and as long to end users/applications as possible. The proposed mechanism derives a data-selection map across edge nodes by learning specific data sub-spaces, which facilitate the placement of processing tasks (e.g., analytics queries). This is very critical when we target to support near real time decision making and would like to minimize all parts of the tasks allocation procedure. We evaluate and compare our approach against baselines and schemes found in the literature showcasing its applicability in pervasive edge computing environments.

In Differentially Private Federated Learning (DPFL), gradient clipping and random noise addition disproportionately affect statistically heterogeneous data. As a consequence, DPFL has a disparate impact: the accuracy of models trained with DPFL tends to decrease more on these data. If the accuracy of the original model decreases on heterogeneous data, DPFL may degrade the accuracy performance more. In this work, we study the utility loss inequality due to differential privacy and compare the convergence of the private and non-private models. Specifically, we analyze the gradient differences caused by statistically heterogeneous data and explain how statistical heterogeneity relates to the effect of privacy on model convergence. In addition, we propose an improved DPFL algorithm, called R-DPFL, to achieve differential privacy at the same cost but with good utility. R-DPFL adjusts the gradient clipping value and the number of selected users at beginning according to the degree of statistical heterogeneity of the data, and weakens the direct proportional relationship between the differential privacy and the gradient difference, thereby reducing the impact of differential privacy on the model trained on heterogeneous data. Our experimental evaluation shows the effectiveness of our elimination algorithm in achieving the same cost of differential privacy with satisfactory utility. Our code is publicly available at https://github.com/chengshuyan/R-DPFL.

The smart healthcare system advancements have introduced the Internet of Things, enabling technologies to improve the quality of medical services. The main idea of these healthcare systems is to provide data security, interaction between entities, efficient data transfer, and sustainability. However, privacy concerning patient information is a fundamental problem in smart healthcare systems. Many authentications and critical management protocols exist in the literature for healthcare systems, but ensuring security still needs to be improved. Even if security is achieved, it still requires fast communication and computations. In this paper, we have introduced a new secure privacy-enhanced fast authentication key management scheme that effectively applies to lightweight resource-constrained devices in healthcare systems to overcome the issue. The proposed framework is applicable for quick authentication, efficient key management between the entities, and minimising computation and communication overheads. We verified our proposed framework with formal and informal verification using BAN logic, Scyther simulation, and the Drozer tool. The simulation and tool verification shows that the proposed system is free from well-known attacks, reducing communication and computation costs compared to the existing healthcare systems.