The development of vehicular networking technology continuously enhances the internet connectivity of modern vehicles. However, for in-vehicle networks, constant communication with the outside world dramatically incre...
详细信息
The rapid deployment of IoT networks in different industrial services has caused the emanation of a huge volume of data from sensors and monitors. The efficient analysis and compact representation of the big data gene...
详细信息
Regenerating codes are a class of distributed storage codes that allow for efficient repair of failed nodes, as compared to traditional erasure codes, which enables it to achieve high data reliability, security, and c...
详细信息
ISBN:
(纸本)9781665457194
Regenerating codes are a class of distributed storage codes that allow for efficient repair of failed nodes, as compared to traditional erasure codes, which enables it to achieve high data reliability, security, and cost-efficiency and a critical infrastructure of the computing system. Existing data storage largely depends on a centralized cloud, which is not only costly but also vulnerable to single points of failure and other types of security attacks. To provide data security, data encryption has to be used, which requires extensive computing power and cumbersome key management. distributed storage system (DSS) is being widely viewed as a natural solution to future online data storage due to improved access time and lower storage cost. However, the existing DSS also has the limitations of low storage efficiency and lack of data security. In this paper, we investigate multi-layer code-based distributed data storage systems that can achieve inherit content confidentiality and optimal storage efficiency. Our comprehensive shows that the optimal code can improve the reliable data storage by nearly 50% comparing to the existing state-of-art research.
With the rapid advancement of smart grids and energy innovation strategies, the widespread promotion of active distribution network control technology will become the key driving force behind the intelligent developme...
详细信息
Forest fires pose significant threats to ecosystems and communities globally. This paper explores the development of an automated system for forest fire monitoring and prevention, particularly focusing on regions arou...
详细信息
ISBN:
(纸本)9798350357899;9798350357882
Forest fires pose significant threats to ecosystems and communities globally. This paper explores the development of an automated system for forest fire monitoring and prevention, particularly focusing on regions around Northern Thailand. Despite recent applications of Unmanned Aerial Vehicles (UAVs) for fire detection, existing technologies face challenges in detecting early onset of fires in dense forests due to limited resolution and sensitivity of thermal cameras. To address this, this paper proposes a novel approach combining UAV-based surveying with ground-based Internet of Things (IoT) sensors to enable early detection of forest fires, even when obscured by tree canopies. The low-cost IoT sensors measure temperature, humidity and air quality at forest ground level. To overcome limitations in 4G communication for the IoT sensors, our system leverages UAVs as communication hubs to collect data from IoT sensors and survey the area for smoke and fire. The proposed system, part of the FireFly Project in collaboration with Chiang Mai University and the University of Glasgow, aims to overcome limitations of existing technologies and provide effective forest fire monitoring and prevention solutions. Experimental results presented in this paper demonstrate the performance of the distributed UAV-IoT system in detecting and communicating potential forest fires, paving the way for enhanced wildfire management strategies in fire-prone areas.
In-Network Computing has brought computing and communication together at every communication node in the digital world, this work lays the foundations for doing the same in the quantum world, improving communication p...
详细信息
ISBN:
(纸本)9798350348439;9798350384611
In-Network Computing has brought computing and communication together at every communication node in the digital world, this work lays the foundations for doing the same in the quantum world, improving communication properties in the process - a combination of quantum computing and quantum communication. Full network softwarization, in-network intelligence, and massive connectivity will create an unprecedented demand for computing resources in the digital world. Accordingly, the scientific and industrial communities have begun to explore technologies such as quantum computing and have made significant efforts to demonstrate an algorithmic advantage for various problems. However, practical quantum computers are resource-inefficient and difficult to build. This article introduces a new communication paradigm leading to the concept of quantum in-network computing in the context of entanglement-assisted communication and computing for inherent distributed resilience and sensing. We review the fundamentals of digital hardware as characterized by Turing's computability theory and demonstrate their relevance to mathematically rigorous characterizations of gate-based quantum computing. We then provide such a characterization using methods from effective analysis, leading to significant results that reveal the inherent theoretical limitations of universal gate-based quantum computers. These results support our assessment that gate-based quantum in-network computing is only possible through specialized, non-universal solutions that are seamlessly integrated with high-performance digital computing.
The proceedings contain 115 papers. The topics discussed include: receding horizon and optimization-based control for UAV path planning with collision avoidance;application of machine learning in technological forecas...
The proceedings contain 115 papers. The topics discussed include: receding horizon and optimization-based control for UAV path planning with collision avoidance;application of machine learning in technological forecasting;unbiased text categorization in IoT-based digital content using a word-to-graph model;multimodal sentiment analysis based on video and audio inputs;asynchronous inter-class project collaboration adaptation method to agile practices in university context;resource consumption analysis of distributed machine learning for the security of future networks;modeling Wazuh rules with weighted timed automata;evaluating safe region sizes for accuracy in approximate continuous nearest neighbor queries;and intrusion detection systems in IoT based on machine learning: a state of the art.
Malwares is designed to infiltrate devices and in most cases the underlying operating system and are packaged in multiple file formats. Considering that Windows operating system is one of the most opted operating syst...
详细信息
This paper evaluates performance of multitier heterogeneous cellular networks (HCNs) with distributed macro-base stations (MBS) and pico-base stations (PBS) using orthogonal multiple access (OMA) to support sixth-gene...
详细信息
Graph neural networks (GNNs) operate on data represented as graphs, and are useful for a wide variety of tasks from chemical reaction and protein structure prediction to content recommendation systems. However, traini...
详细信息
ISBN:
(纸本)9781665497473
Graph neural networks (GNNs) operate on data represented as graphs, and are useful for a wide variety of tasks from chemical reaction and protein structure prediction to content recommendation systems. However, training for large graphs and improving training performance remain significant challenges. Existing distributed training systems partition a graph among all compute nodes to train for large graphs;however, this results in a communication overhead to degrade training performance. In this study, to solve these two problems, we propose a scalable data-parallel distributed GNN training system designed to partition a graph redundantly. It is implemented using remote direct memory access (RDMA) and nonblocking active messages to efficiently utilize network performance and hide communication overhead by overlapping with the training computation. Experimental results are presented to show the strong scalability of the proposed approach, which achieved parallel efficiencies of 0.93 using eight compute nodes for the ogbn-products dataset in the Open Graph Benchmark (OGB) and 0.95 based on two compute nodes using 32 compute nodes for the ogbn-papersl00M dataset. The proposed system exhibited a training performance 18.9% better than the state-of-the-art DistDGL, even with only a single compute node. The results demonstrate that the proposed approach may be considered a promising method to achieve scalable training performance for large graphs.
暂无评论