A multiprocessor system is classified as t/s-diagnosable if it is possible to pinpoint all malfunctioning processors within a group of no more than s processors, assuming the total number of faults does not surpass t. The notion of t/(t + 1)-diagnosability represents the highest value of k for which the system is considered k/(k + 1)-diagnosable. This metric has received considerable attention in prior research concerning various multiprocessor systems. This paper's significant contributions include introducing a necessary and sufficient criterion for t/(t + 1)-diagnosability of multiprocessor systems under the PMC model. Moreover, we assess the t/(t + 1)-diagnosability for an n-dimensional alternating group graph AGn within the same model. Specifically, we find that the diagnosability of AG4 is 5, AG5 is 13, and for n >= 6, the diagnosability of AGn is 6n-16.
This paper studies secure multi-party computation (secure MPC) protocols constructed by secret-sharing techniques. One of well-known secret-sharing-based secure MPC protocols is the CCD protocol previously developed by Chaum, Crepeau and Da & ring;mgard which is of high efficiency. However, the CCD-based MPC protocol has a limitation that the number of participants is required to be larger than 2. Until now, it is still unknown whether there a CCD-based secure two-party computation (secure 2PC) protocol exists or not. In this work, we address this open problem by first defining the feasibility of CCD-based secure 2PC protocol construction. Then, based on the feasibility definition, we give an impossibility proof for constructing a CCD-based secure 2PC protocol.
Visual testing for cross-browser testing is an essential part of web automated regression testing. However, there is an inevitable challenge in visual testing, which is "dynamic content." Dynamic content, such as advertisements and news, constantly changes its appearance over and overAtime. AExisting visual testing tools fail to handleAdynamic content, and some even treat it as cross-browser incompatibilities (XBIs), Athen generate false positives warning to developer. Currently there are two types of approaches to identifying dynamic content. First, manually marking the positions of dynamic content does not signifi-cantly alleviate the developers' workload. ASecond, repeatedly executing the website on the same browser and regarding different pixels as dynamic content. However, this method can lead to false judgments. In this paper, we propose a Siamese network-based comparison approach to detect visual disparities by comparing the visual structuresAofAa webpage rendered by two browsers. This approach can identify cross-browser inconsistencies and accommodate minor element shifts, achieving a high F1-score of 0.9749. A comparison with the usage of structural similarity (SSIM) and multi-scale structural similarity (MS-SSIM) is also carried out and discussed.
The vision of the Internet of Things (IoT) involves devices seamlessly communicating and collaborating over a network. However, IoT devices often face significant limitations due to limited computing power and energy consumption, which require dedicated networks tailored to specific use cases. This, in turn, complicates inter-device communication across different networks. Although the Web of Things (WoT) and 6LoWPAN have addressed some of these challenges, implementing 6LoWPAN over non-IEEE 802.15.4 networks, such as BLE, has proven to be complex and inefficient. This paper proposes a novel RESTful programming model, called WeBLE (Web of BLE devices), that enables access to devices in BLE networks as if they were WoT endpoints (Web resources) on an IP network, without modifying the existing protocol stack. We address three key challenges in achieving transparency - addressability, accessibility, and findability - and design solutions to address these issues. These mechanisms not only enhance device interoperability but also lay the foundation for more robust and scalable IoT applications.
The integration of dual-message encoding in a single quick response (QR) code has been a challenge because of standard reader limitations. In this paper, we present a novel two-level QR code system that enables the encoding of two distinct messages, discernible through standard QR code readers with and without a colored cellophane mask. Our method enhances QR code interactivity by leveraging color module design and optical filtering. Thus, it offers high feasibility and practical applications in diverse settings. Our experiments demonstrate a high scanning success rate, particularly with the red and green color combination, showcasing the robustness of our approach. Our code and experimental data are available at https://github.com/yjhuang-tw/color-masked-qr-code/.
Project issues often capture the challenges and solutions encountered during software development, making them valuable learning resources for avoiding redundant problem-solving efforts. However, the volume and complexity of issue data present a significant challenge for effective learning. This study proposes an AI-assisted approach to issue-based learning. Leveraging a large language model (LLM) for issue classification and filtering and drawing inspiration from the generative component of Generative Adversarial Networks (GANs) (while excluding the adversarial aspect), we generate plausible yet incorrect answer options to enhance question difficulty. This design supports a gradual, layered learning process. Using open-source project issues from GitHub as the data source, we developed a learning platform called SPILSG to generate and deliver learning materials. Experimental results demonstrate the following: (1) The GPT-4-0613 model achieved the highest compliance rates at 37.4% and 50.6%; (2) In terms of reading effectiveness, the LLM-based SPILSG system significantly improved novices' understanding of software project issues; and (3) Usability evaluations indicated a generally positive user experience. In summary, AI-assisted learning offers a time-efficient way for learners to comprehend project contexts, familiarize themselves with codebases, and enhance their programming proficiency.
Cloud computing is an emerging technology paradigm that provides on-demand accessibility of network resources for data storage and processing. However, cyberthreats approach in the form of phishing to steal the confidential user credentials. In cloud computing, the detection of phishing attacks is challenging which is one of the severe threats used against cloud service users. However, the security and privacy concerns such as inadequate access control and weak authentication procedures are the primary obstacles to the wide adoption of clouds. To successfully deal with these challenges, a novel Deep Learning (DL) Based Phishing AttaCK Detection in URL Using CNN (PACK-NET) is proposed for appropriate detection and Phishing attack mitigation. The user submits the URL to the User Safeguard Tool (UST) to get access to the cloud resources. The Case-Based Reasoning (CBR) system pre-processes the URL and the CNN network categorizes the URL into Phishing or Non-Phishing Link. The PACK-NET technique is simulated using a cloud simulator (Cloudsim) and assessed using ADFA-LD and ISCX-URL2016 datasets. The efficacy of the PACK-NET framework has been analyzed in terms of accuracy, specificity, precision, F1-score, recall, detection rate, processing time, inference time, and resource utilization. In comparison, the PACK-NET methodology achieves an overall accuracy of 6.7%, 5.39%, and 3.06% than ANN, LSTM, and RNN techniques respectively.
Image dehazing is an essential process for enhancing the visibility and clarity of images degraded by atmospheric haze, a common issue in fields such as remote sensing, satellite imagery, and aerial surveillance. Haze diminishes the quality of images by scattering light, leading to reduced contrast and color fidelity, which makes accurate image interpretation challenging. In this paper, we propose a novel Generative Adversarial Network (GAN)-based model for image dehazing. Our approach leverages a deep learning framework to remove haze and restore the fine details in hazy images, offering improvements over existing dehazing techniques. The model consists of a generator that learns to produce dehazed images from hazy inputs and a discriminator that evaluates their authenticity. We evaluate our model on the Haze1k dataset, which includes images with varying levels of haze, and compare its performance to state-of-the-art methods using standard image quality metrics such as Structural Similarity Index (SSIM) and Peak Signal-to-Noise Ratio (PSNR). The experimental results demonstrate that our model outperforms existing methods in terms of both SSIM and PSNR across different haze intensities. Our model provides a significant contribution to the field of image restoration, particularly for remote sensing and other applications requiring clear visual data. Future work will focus on extending the model to handle more complex real-world haze scenarios and improving computational efficiency for large-scale datasets.
Behavior-driven development (BDD) is a new method that has increasingly attracted attention from the Internet of Things (IoT) research and development community. To contribute to the increasing adoption of BDD for IoT development, we propose a new method and a Gherkin language enhancement for building executable system behavioral specifications by reusing and composing the component scenarios already available. With the objectives of achieving maintainability and diagnosability, the new method applies the problem frames approach to obtain a problem diagram, which captures the part-whole relationship between the system and its components and the essential elements needed. A system scenario is modeled as a directed path in the problem diagram that readily translates into a script of the enhanced Gherkin. The system scenario acts as a co-routine alongside the given component scenarios at two levels. Executing a system scenario concurrently sets its component scenarios in motion, driving the components to produce the required behaviors that coalesce into the overall system behavior described by the system scenario. The proposed method is qualitatively demonstrated to balance modifiability with diagnosability better than the one-level scripting of the current Gherkin. The two-level scripting is supported by concurrentSpec, an embedded Gherkin tool in Python that supports the composition, execution, and reporting of the two-level executable specifications.
This work proposes a framework that integrates stacked long short-term memory (S-LSTM) with large-scale multiple-input multiple-output (L-MIMO) and non-orthogonal multiple access (NOMA) to mitigate temporal fading impairments that are prevalent in high-mobility vehicle-to-vehicle (V2V) wireless communication environments. The fading channel is modeled as Nakagami-m distributed fading under imperfect channel state information (CSI). The proposed S-LSTM framework leverages deep recurrent neural layers to suppress Doppler-induced distortions, improve robustness to outdated CSI, and enhance temporal feature learning and memory depth for accurate CSI estimation and adaptability to rapidly varying channels. Simulation studies demonstrate that the S-LSTM achieves a 4-9 dB performance gain in reducing outage probability (OP) and symbol error rate (SER) compared to conventional long short-term memory (LSTM), minimum mean square error (MMSE), and least squares error (LSE) estimators across the 10-30 dB signal-to-noise ratio (SNR) range. By leveraging long-term temporal patterns, the S-LSTM achieves up to a tenfold reduction in SER and allows for reliable communication at speeds higher than 300 mph. The signal-to-interference-plus-noise ratio (SINR) is assumed to be the performance metric used to evaluate system robustness. Although both SER and OP decrease as SNR increases, a persistent error floor remains due to residual interference, mobility effects, and CSI imperfections. The model is a viable con-tender for robust channel estimation in next-generation wireless networks (NGWNs) because of its deeper S-LSTM architecture, which further increases training accuracy, and its smaller batch size, which allows for faster convergence and dynamic adaptability.
Local diagnosability is an effective approach to assess system diagnosability by evaluating each processor individually. According to the basic definition of diagnosis, the underlying topology of existing local comparison diagnosis models can be modeled as directed graphs with bidirectional edges, so it is difficult to apply them in classical digraphs. In this paper, we introduce a (p + q)-directed local diagnosis structure D(u, p, q) and present a local diagnosis algorithm for digraphs. This algorithm identifies faults or fault-free status of each processor under the comparison model. Our results indicate that any vertex in a digraph with the structure D(u, p, q) is locally (p +q)-diagnosable, with a time complexity of O(p+q). We apply the (p + q)-directed local diagnosis structure to unidirectional hyper-cubes, determining the local diagnosability for any vertex. Simulation results show that our algorithm maintains high performance even with a 40% fault probability for each vertex, achieving ACC >= 0.836, PPV >= 0.827, and NPV >= 0.816.
This study proposed a web-based system that utilizes an AI model trained on candlestick patterns to display buy and sell signals, and it uses Direct Preference Optimization (DPO) to dynamically adjust parameters, ensuring that the model can adapt to market changes. Traditional models focus on individual stocks or indices, which are often timeconsuming and complex. In contrast, this approach effectively identifies trend initiation points using candlestick patterns. Experimental results show strong performance in recent markets, highlighting the potential of advanced AI techniques for robust stock market analysis and decision-making in financial applications.
The first requirement faced during the transformation to automation in the traditional manufacturing industry is the conversion of process instructions, originally written in natural language, into a machine-readable format. Many large-scale factories, such as automotive assembly factories, have achieved full automation. However, there are still many traditional assembly industries, especially small and medium-sized enterprises, that cannot transition to automated factories due to equipment costs and environmental limitations. Therefore, there is a need to implement supervision and guidance systems to detect errors and ensure correct assembly operations. This study presents an automated modeling tool based on state diagrams for small and medium-sized assembly factories. This tool converts natural language instructions into a state diagram representation, simplifying the general rules for natural language modeling and enhancing their applicability to assembly processes. This can be used to supervise and guide operators in manual assembly. In addition, a simple verification method is proposed to verify whether the state diagram accurately reflects the original assembly manual content.
A surgery scheduling problem has been characterized by the integer linear program in the previous work. The objective function is to maximize the total weighted operating time of the week, which equivalently minimizes the idle time on the operating rooms (ORs). The constraints ensure that each operation can only be scheduled once and guarantee that the total time spent on all operations for all departments does not exceed the working hour of a day. We propose two randomized linear program rounding algorithms to solve the problem approximately and in this paper finally show their performances in terms of approximation ratios with a bounded probability for infeasibility and with a feasibility guarantee, respectively. We also supplement the analytical result with numericlal result from experiments.
This study investigates the effectiveness of abstractive text summarization in the context of scientific documents using 40 diverse Large Language Models (LLMs). Unlike traditional extractive approaches that often produce fragmented and less coherent summaries, our work focuses on enhancing semantic fidelity, coherence, and comprehensive content coverage. Through a recall-oriented evaluation supported by BERT and METEOR metrics, our experimental results show that models such as Claude v2.1, Qwen-14B, Zephyr-7B, and Phi-3 emerged as top performers, achieving outstanding F1 scores above 0.93 and METEOR scores as high as 1.00. These models demonstrated a strong ability to retain critical information while producing fluent, human-like summaries. Our findings provide valuable benchmarks for selecting high-performing LLMs in summarization tasks and offer a foundation for future advancements, including domain adaptation, fact-checking integration, and multimodal summarization approaches in real-world Natural Language Processing applications.
Despite rapid growth in the number of programming learners, the continuous influx of new information and a lack of human guidance has left many learners sifting through online resources for reliable content. Moreover, many learners struggle to frame questions that accurately capture their specific concerns. This study addressed the growing need for efficient learning resources by developing a microservice-based Chatbot to synthesize information from diverse sources for programming learners. The proposed MPAbot system utilizes keyword extraction and cross-referencing across multiple sourced posts to help users solve questions more effectively. MPAbot incorporates word embeddings, sentence similarity, Latent Dirichlet Allocation (LDA) topic modeling, and multi-criteria decision analysis to filter out redundant information, thereby reducing browsing time and enhancing learning efficiency. Besides, MPAbot also integrates generative AIto improve the accuracy of its answers to users' questions. Experimental results show that our GPT-ranked approach achieved a 52.43% higher mean rank score compared to purely GPT-generated responses in user evaluation tests, demonstrating the effectiveness of the proposed multi-source integration approach.
Given an ultrametric space ({1,2,. ..,n}, d) (where n is given as input and d is provided as an oracle), Chen and Chang's [1] deterministic O(F +n log n)-time algorithm enumerates all furthest pairs (without repetitions) in ({1,2,. ..,n}, d), where F denotes the number of furthest pairs. We show that their algorithm runs in O(F) time.
Mobile edge computing (MEC)-enabled optical interconnected edge networks are composed of multiple MEC servers deployed at base station (BS) locations or independently which are interconnected by optical networks, and terminal devices (TDs) access MEC servers by connecting to BSs through a wireless network. By assuming that TD's task offloading needs to be supported by specific services, this paper investigates task offloading in the MEC-enabled optical interconnected edge networks, where services for offloading tasks are deployed at different MEC servers, and the user's cost budget of task offloading is also taken into account. In the proposed task offloading scheme, for a task that needs to be offloaded, we first select the appropriate BS to access the optical interconnected edge networks, then the MEC server and optical transmission path are determined to carry out task offloading. We first formulate the task offloading problem as an Integer Linear Program (ILP) with the objective of minimizing the overall offloading cost of all tasks. To achieve the real-time solution, a heuristic is then proposed which divides the offloading operations into two procedures i.e., BS selection, and MEC server and path selection. The former is achieved by a genetic algorithm, and a strategy for properly selecting the MEC servers and the corresponding transmission paths is designed in the latter. Extensive numerical results are provided to validate the proposed task offloading scheme.
In today's dynamic networking landscape, integrating Software-Defined Networking (SDN) with Traffic-Expert Virtual Networks (TEVN) presents a promising avenue for optimizing network performance. This research investigates the implementation of TEVN Embedding within SDN frameworks, utilizing the Ryu controller to address inefficiencies in traditional virtual network embedding algorithms. Methodologically, the study proposes a framework for TEVN and evaluates its performance against benchmark methods using various parameters such as throughput, bandwidth, packet loss, and Round-Trip Time (RTT). The evaluation is conducted through extensive experimentation in simulated SDN environments, with results analyzed and compared comprehensively. The findings reveal that TEVN significantly improves network efficiency, achieving higher throughput, lower latency, and reduced packet loss compared to default embedding algorithms. These results underscore the potential of TEVN to revolutionize network management practices, offering a promising solution for addressing the evolving challenges of modern network infrastructures. This research contributes to advancing SDN technologies and gives insights into enhancing network efficiency in dynamic environments.
The tremendous growth of the population globally is leading to a requirement for a revolution in food production to meet the demands. Hence, the agriculture sector is facing challenges like an increase in production via sustainable methods quantitatively as well as qualitatively. IoT has proved to be a promising technology in providing reliable and efficient solutions in this field. These methods have enabled the monitoring of agricultural farms distantly. However, the latency of wireless networks has turned out to be a significant challenge in such farming technologies. 5G has emerged as a cutting-edge technology that has revolutionized the sector of farming by making farming more productive, sustainable, and effective. This research paper delves into the progress domain of integrated 5GIoT technology in Smart Agriculture. This paper provides an extensive survey of this emerging technology. The applications and open challenges have been discussed in depth. This study will be useful to researchers keen to add to the realm of Smart Agriculture.
