The increasing annual demand necessitates an efficient model for planning large-scale networks, which typically include numerous nodes and edges. Designing an operational pipeline that efficiently utilizes resources presents significant challenges. This study presents a graph-based approach for designing a water pipeline network, with graph edges representing pipelines and nodes representing users (demand) or reservoirs. To address this complexity, we use Graph Embedding approach, a machine-learning technique that compresses large-scale graphs into lower-dimensional vector spaces. This method employ a random walk model to make sequence of nodes with their neighborhoods and a skip-gram model train them to ensure accurate modeling of the shortest paths between them. Then, the nodes are clustered in vector space and optimized using the minimum spanning tree(MST) method. The graph generated still preserves essential network characteristics such as node relationships and spatial proximity. The resulting model produces a graph based on the geographic coordinates of nodes that map the layout of the water pipeline network. In forming large-scale network layouts, the graph embedding approach improves the computation time efficiency. We evaluate the model’s performance using the GFS-Score and validate the network layout with EPANET. This approach offers significant potential in advancing the development of intelligent water network structures for large-scale complex networks.
Message routing protocol is a key factor in delivering messages efficiently in Delay/Disruption-tolerant networks (DTNs). Messages sent by a mobile node are transferred in a Store-Carry-Forward manner among mobile nodes in a DTN. Existing protocols take into account mobile nodes’ behavior, history information, and movement model to optimize their mechanisms for exchanging messages while encountering one another. However, none of them consider the construction of the underlying environment, such as road network construction. We propose CD-PRoPHET, a variant of PRoPHET, to tune the delivery probability by considering the road network density. Our simulation results show that CD-PRoPHET significantly reduces the overhead ratio at the cost of the delivery ratio.
IoT is increasingly being used in different areas of society. IoT is used in society by collecting values acquired by sensor devices through networks and analysing the data. IoT software is needed to use the IoT, but efficient software development methods are desired because software for an increasingly complex society tends to become more complex. This research aims to improve the efficiency of IoT software and develop an intuitive method for developing IoT software.This paper proposes a development methodology that combines a data-flow-based programming approach with a procedural programming language. When the focus is on data, data flow programming is intuitive and efficient. On the other hand, when the focus is on data processing, it is better to use a procedural programming language because data is processed sequentially[1].In the hybrid IoT programming environment proposed in this paper, the code of a procedural programming language is generated from the description by data flow programming, and the program is executed by a micro controller.The products of this research are available as open source software[2], [3].
Employee training is expensive and time-consuming as evidenced by the fact that teaching new employees how to operate factory equipment is usually done in groups to save time and resources. For some industrial facilities, training often requires shutting down operations and putting employees and trainees in potentially dangerous situations. Flour mill operators are typically exposed to work hazards such as dust, excessive noise, explosions, electrocutions, and injuries from improperly guarded machinery. Thus, this paper addressed the issues previously mentioned by developing a virtual reality training software mainly focused on the procedures involved during break release measurement. It is the process wherein a mill operator takes samples from specific roller mills and measures the percentage of flour extracted from a flour milling machine. With this software, the flour mill machine is modeled under simulated conditions giving the user a fully immersive training experience without the fear or danger of experiencing real-life errors and accidents. Twenty-five flour mill operators participated in testing and evaluating the VR trainer software. The results indicated that 92% were satisfied with their training experience, 88% were pleased with the software's functionality, and 98.67% gave positive ratings in the user satisfaction category of the VR training environment.
The proliferation of diverse network technologies has enabled concurrent utilization of multiple network media, such as WiFi and 4G/5G links on modern smartphones. While this advancement facilitates simultaneous downloading of varied network contents, traditional multi-connection management methods, like reception-driven requesting schemes, often struggle to adapt to dynamic network conditions. This paper introduces a novel timer-driven requesting scheme applied to both parallel TCP and parallel MPTCP (Multipath TCP) methods, addressing the limitations of existing approaches. Unlike previous studies that relied solely on simulations, we present an experimental evaluation of this scheme on real machines. Our extensive experimentation reveals significant advantages of the timer-driven requesting scheme, particularly its superior adaptability in fluctuating network environments. These findings not only demonstrate the scheme’s practical viability but also underscore its potential to revolutionize content retrieval in multi-network scenarios, paving the way for more efficient and resilient data transfer mechanisms in increasingly complex network ecosystems.
This position paper presents the preliminary results from a research study focusing on cryptographic agility for consumers of open-source cryptographic libraries. We provide a concise overview of frontiers and recent advancements in cryptographic agility research frontier, examining the utilized approaches and emphasizing the perception of application layer encryption in end-to-end encryption systems. The paper delves into the state of practice of cryptographic libraries and programming interfaces, outlining recognized challenges and knowledge gaps that warrant exploration through new scientific research. Furthermore, we outline the values of cryptographic agility in a manifesto and propose a survey structure to validate our assumptions.
Vulnerability assessment is an important and well-studied subject in software security. Traditional methods use expert knowledge, which is time-consuming. Considering the constantly increasing number of vulnerabilities, automated machine learning (ML)-based solutions have been proposed to assess the severity of vulnerabilities. Existing methods concentrate on predicting the Common Vulnerability Scoring System (CVSS) score or its vector metrics using available vulnerability information. The quality and diversity of the vulnerability description data can greatly affect the accuracy of these predictions. Studies report that less than 60% of such descriptions follow the formal template. On the other hand, the performance of ML-based vulnerability scoring approaches is highly dependent on the quality of the data and the model’s architecture. In this paper, we aim to improve the performance of existing ML-based solutions in vulnerability assessment. We use generative artificial intelligence (AI) and feed the CVSS descriptions to a large-language model. We use GPT3.5Turbo to generate descriptions and propose a fine-tuned BERT-CNN model to predict the CVSS vector metrics. We conduct several experiments to assess the performance of the proposed method against the state-of-the-art. We use both the original dataset (6,370 descriptions) and the descriptions generated by GPT3.5Turbo. Our experiments show that our proposed architecture considerably improves accuracy.
In this paper, we consider three point-prediction methods for the number of software bugs detetced in future with the bug count data experienced in past, where the underlying software bug-detection process is described by a non-homogeneous Poisson process. In general, it is known that the past probability distribution of number of software bugs is not always identical to the future one. Nevertheless, the commonly used technique is to predict the bug counts under a strong assumption that the probability distribution with model parameters estimated from the past observation holds even in the future evolution. Since such a plug-in prediction does not often work well to guarantee the higher prediction accuracy, investigating more accurate prediction of software bug counts is an emerging issue in software reliability engineering. We propose the so-called maximum likelihood predictors to predict the future bug-detection processes and a different model selection scheme from the common information criteria. Through a numerical example with an actual software bug count data set, we compare our new prediction methods with the existing plug-in predictor.
Finding available parking spots in urban areas is a challenging and time-consuming task due to inefficient traditional methods. This paper presents a smart parking solution that leverages You Only Look Once (YOLOv5) for car detection and body type classification, coupled with Simple Online Real-time Tracking (SORT) for continuous tracking of vehicle bounding boxes. This study introduces a straightforward 2D environment modeling to improve parking management. The proposed system demonstrates performance with a top precision of 0.992, a top recall of 1.0, a mean Average Precision (mAP) of 0.995 at IoU threshold 0.50, and a mAP of 0.953 across IoU thresholds 0.50-0.95. The color detection accuracy and parking occupancy detection also achieved a top score of 100% in the 30-day test experiment. Future work involves deploying the system in various sites to evaluate its robustness and effectiveness in diverse real-world conditions.
Regional public transportation services in Japan, including buses and taxis, face challenges such as crew shortages. This is attributed to a combination of factors: a decline in demand due to the Corona disaster and an aging workforce caused by the inability to attract new crews. Despite these challenges, regional public transportation relies heavily on minimal revenue streams like fixed fares. The primary source of funding is not fare revenue but subsidies from local governments overseeing these services. Consequently, these services constantly grapple with financial strain, compelling management to strike a delicate balance. To maintain timely operations amidst personnel shortages, the workload per individual often increases, placing undue pressure on operation managers responsible for crew management as per legal and regulatory guidelines. To alleviate this issue, we developed an crew management support application specifically for these managers. Given that the application manages sensitive personal and driver information, security considerations are paramount. This paper presents the developed support application and delves into its security implications.
Congestion detection in places where people gather is critical for safety monitoring and disaster management. Conventional methods such as crowd counting, typically estimate the number of people based on still images and it is difficult to determine the overall congestion level in situations when only low quality images are available due to occlusion or weather condition. In this paper we propose a congestion prediction method that employs audio data generated by humans such as footsteps and conversations. By combining the deep learning model for crowd counting with audio features, we can obtain more accurate congestion predictions. Finally, we evaluated the proposed method based on real data.
IoT sensors are becoming more essential for remote monitoring of homes, automotives, transportation, healthcare facilities, smart cities, and many other applications. An end-to-end secure data communication from IoT sensor devices to monitoring center is critically important. However, the level of security (number of subsequent encryption layers) requirements and associated costs to implement security for different applications are different. We surveyed and studied the currently published ideas on security for IoT networks and realized that there is a need for mechanism in the network to dynamically decide which application will require what level of security encryption. In this paper we propose and analyze an algorithm that dynamically decides different levels of security for different applications. The proposed solution encrypts data packets one time with an application identifier added in the packet header at IoT sensor device. Based on the application identifier in the data packet sent by IoT sensor, SDN router dynamically makes a collaborative decision with IoT sensor and adds additional levels of security if needed. This mechanism is helpful to securely integrate IoT sensor features and network interfaces in a very cost-effective manner while communicating with billions of possible IoT sensor devices designed for different applications. The idea is also helpful to analyze many other security resolutions for 6G integrated IoT sensors and communications network.
The excessive activation of user dynamics, such as online flaming, causes various social issues, making effective intervention based on early detection desirable. Current early detection methods identify increased user activity based on quantitative changes in time-series data, such as whether the number of social media posts exceeds a threshold. However, from a theoretical standpoint rooted in fundamental principles, it is expected that the precursor to excessive activation of user dynamics due to structural changes in social networks will manifest itself as the emergence of a low-frequency mode in the time series of user dynamics intensity. This research describes a method for the early detection of excessive activation of user dynamics by identifying the emergence of low-frequency modes through frequency spectrum analysis of actual SNS data, a method faster than the quantitative observation of time-series data.
The number of connected devices and the Internet of Things (IoT) continues growing significantly, with global spending expected to exceed $1 trillion by 2026. Despite this growth, IoT and connected devices face security challenges, as millions of devices are reportedly involved in botnets. IoT and connected devices are more vulnerable to medium- and high-severity attacks since more than 91.5% of the IoT’s traffic is unencrypted. Governments have planned or initiated national registries of certified devices and labeling programs to address these challenges. As those registries and labels remain national, multiple governments have started signing mutual recognition between their programs, adding complexity. This motivated us to create a unified and collaborative labeling registry and a rating system that uses 12 criteria to classify IoT devices. Through multiple experiments, 52 users submitted 252 device classifications. Our proposed tool is helping us identify the criteria that define IoT and connected devices’ security.
With the increasing popularity of Internet of Things (IoT) and its connected devices, security has become a major concern. In this paper, we conducted a benchmark to evaluate performance of different deep learning algorithm device based on its network traffic. We developed our own dataset for our pilot study that included three different types of cyberattacks: reverse shell, keylogger, and synflood.The diversity and scope of our research has been enhanced by the incorporation of the CIC IoT dataset, which has been added to our initial work. We conducted a systematic evaluation of the performance of various deep learning models, which included CNNs and LSTM networks. Our benchmarking efforts on the CIC IoT dataset resulted in a significant improvement, with all models achieving an accuracy of over 99% and more than 93% on our custom dataset.
In this paper we propose a predictive model for learning analysis using student teaching data and demonstrate the effectiveness of machine learning methods in predicting retention and dropout. We also examine the impact of online classes due to the COVID-19 pandemic on academic performance, and analyze the correlation between changes in academic performance and prediction of retention and dropout based on recall and precision.
Perimeter defense strategies are inadequate to ensure cybersecurity of infrastructures consisting of heterogeneous and dynamic resources. The Zero Trust security model emerges as the most promising solution to mitigate risks and protect assets, but significant organizational and implementation challenges hinder its adoption. Microsegmentation of networked systems composed by dynamic IT components and mobile devices cause several technological and management concerns. We present a comprehensive analysis of microsegmentation with the goal of identifying the key aspects that distinguish it from traditional perimeter defenses. We then propose a modular architectural design pattern that ensures adherence to the Zero Trust principles and satisfies its security constraints. This design is based on the concept of Security Domain, which represents the fundamental unit of network segmentation. By combining multiple Security Domains and following precise rules that provably preserve network security, it becomes possible to create complex infrastructures from elementary building blocks. We provide also a formal specification of the proposed design by means of the TLA+ modeling language. We leverage this model to verify its correctness and security properties even in the presence of insider threats.
The recent surge in traffic demand has accelerated the development of heterogeneous high-density networks. Consequently, the number of base stations has increased, leading to a rise in the frequency of vertical handovers between heterogeneous base stations, which results in delays and increased power wastage. Earlier studies analyzed the number of vertical handovers in such networks. Unfortunately, while the number of vertical handovers with short sojourn periods is increasing, the sojourn time itself has not been well addressed. This paper prepare to analyze vertical handover with direct consideration of the short cell sojourn time distribution as derived from stochastic geometry. Additionally, by referencing Bertrand’s paradox, we define and simulate randomness in three different scenarios to validate our findings.
Machine Learning (ML) has emerged as one of data science's most transformative and influential domains. However, the widespread adoption of ML introduces privacy-related concerns owing to the increasing number of malicious attacks targeting ML models. To address these concerns, Privacy-Preserving Machine Learning (PPML) methods have been introduced to safeguard the privacy and security of ML models. One such approach is the use of Homomorphic Encryption (HE). However, the significant drawbacks and inefficiencies of traditional HE render it impractical for highly scalable scenarios. Fortunately, a modern cryptographic scheme, Hybrid Homomorphic Encryption (HHE), has recently emerged, combining the strengths of symmetric cryptography and HE to surmount these challenges. Our work seeks to introduce HHE to ML by designing a PPML scheme tailored for end devices. We leverage HHE as the fundamental building block to enable secure learning of classification outcomes over encrypted data, all while preserving the privacy of the input data and ML model. We demonstrate the real-world applicability of our construction by developing and evaluating an HHE-based PPML application for classifying heart disease based on sensitive ECG data. Notably, our evaluations revealed a slight reduction in accuracy compared to inference on plaintext data. Additionally, both the analyst and end devices experience minimal communication and computation costs, underscoring the practical viability of our approach. The successful integration of HHE into PPML provides a glimpse into a more secure and privacy-conscious future for machine learning on relatively constrained end devices.
Due to the increased adoption of edge computing, and growing concern about data privacy, Federated Learning (FL) frameworks have gained significant attention as a promising approach to leverage distributed data while protecting user privacy. However, ensuring the privacy of model updates from multiple edge devices is a significant challenge due to the potential privacy breaches and security vulnerabilities. In this paper, we propose EdgeSA a novel approach that leverages secure aggregation for privacy-preserving federated learning in edge computing. EdgeSA ensures that local model updates are gathered from edge devices in a way that respects privacy and integrity. Our comprehensive security analysis and experiments, demonstrate the effectiveness and efficiency of EdgeSA. The implementation results show that our scheme reduces the computational overhead on edge devices while achieving the same training accuracy as traditional federated learning schemes.
