Adaptive Graph-LLM Fusion for Context-Aware Risk Assessment in Smart Industrial Networks

The Internet of Things (IoT) boom has revolutionized almost every corner of people’s daily lives: healthcare, environment, transportation, manufacturing, supply chain, and so on. With the recent development of sensor and communication technology, IoT artifacts, including smart wearables, cameras, smartwatches, and autonomous systems can accurately measure and perceive their surrounding environment. Continuous sensing generates massive amounts of data and presents challenges for machine learning. Deep learning models (e.g., convolution neural networks and recurrent neural networks) have been extensively employed in solving IoT tasks by learning patterns from multi-modal sensory data. Graph neural networks (GNNs), an emerging and fast-growing family of neural network models, can capture complex interactions within sensor topology and have been demonstrated to achieve state-of-the-art results in numerous IoT learning tasks. In this survey, we present a comprehensive review of recent advances in the application of GNNs to the IoT field, including a deep dive analysis of GNN design in various IoT sensing environments, an overarching list of public data and source codes from the collected publications, and future research directions. To keep track of newly published works, we collect representative papers and their open-source implementations and create a Github repository at GNN4IoT.

Powerful graph neural network for node classification of the IoT network

Security vulnerabilities of the modern Internet of Things (IoT) systems are unique, mainly due to the complexity and heterogeneity of the technology and data. The risks born out of these IoT systems cannot easily fit into an existing risk framework. There are many cybersecurity risk assessment approaches and frameworks that are under deployment in many governmental and commercial organizations. Extending these existing frameworks to IoT systems alone will not address the new risks that have arisen in the IoT ecosystem. This study has included a review of existing popular cyber risk assessment methodologies and their suitability to IoT systems. National Institute of Standards and Technology, Operationally Critical Threat, Asset, and Vulnerability Evaluation, Threat Assessment & Remediation Analysis, and International Standards Organization are the four main frameworks critically analyzed in this research study. IoT risks are presented and reviewed in terms of the IoT risk category and impacted industries. IoT systems in financial technology and healthcare are dealt with in detail, given their high-risk exposure. Risk vectors for IoT and the Internet of Medical Things (IoMT) are discussed in this study. A unique risk ranking method to rank and quantify IoT risk is introduced in this study. This ranking method initiates a risk assessment approach exclusively for IoT systems by quantifying IoT risk vectors, leading to effective risk mitigation strategies and techniques. A unique computational approach to calculate the cyber risk for IoT systems with IoT-specific impact factors has been designed and explained in the context of IoMT systems.

SummaryIntrusion detection systems (IDSs) are the major component of safe network. Due to the high volume of network data, the false alarm report of intrusion to the network and intrusion detection accuracy is the problem of these security systems. The reliability of Internet of Things (IoT) connected devices based on security model is employed to protect user data and preventing devices from engaging in malicious activity. In this article, intrusion detection framework using auto‐metric graph neural network optimized with hybrid woodpecker mating and capuchin search optimization algorithm in IoT Network (IDF‐AGNN‐HYB‐WMA‐CSOA‐ IoT) is proposed. Initially the attacks affected in the IoT data is taken from the dataset such as CSIC 2010 dataset, ISCXIDS2012 dataset, then these data are preprocessed and the features are extracted to remove the redundant information using improved random forest with local least squares. Then the malicious attacks and the normal attacks are classified using the auto‐metric graph neural network. At last hybrid woodpecker mating and capuchin search optimization algorithm (Hyb‐WMA‐CSOA) is utilized to optimize the weight parameters of AGNN. The performance of ISCXIDS2012 dataset of the proposed method shows higher accuracy 25.37%, 29.57%, and 18.67%, compared with existing methods, such as IDF‐ANN‐IoT, IDF‐BMM‐IoT and IDF‐DNN‐IoT respectively.

In the context of Industry 4.0, the medical industry is horizontally integrating the medical resources of the entire industry through the Internet of Things (IoT) and digital interconnection technologies. Speeding up the establishment of the public retrieval database of diagnosis-related historical data is a common call for the entire industry. Among them, the Magnetic Resonance Imaging (MRI) retrieval system, which is one of the key tools for secure and private the Internet of Medical Things (IoMT), is significant for patients to check their conditions and doctors to make clinical diagnoses securely and privately. Hence, this paper proposes a framework named MRCG that integrates Convolutional Neural Network (CNN) and Graph Neural Network (GNN) by incorporating the relationship between multiple gallery images in the graph structure. First, we adopt a Vgg16-based triplet network jointly trained for similarity learning and classification task. Next, a graph is constructed from the extracted features of triplet CNN where each node feature encodes a query-gallery image pair. The edge weight between nodes represents the similarity between two gallery images. Finally, a GNN with skip connections is adopted to learn on the constructed graph and predict the similarity score of each query-gallery image pair. Besides, Focal loss is also adopted while training GNN to tackle the class imbalance of the nodes. Experimental results on some benchmark datasets, including the CE-MRI dataset and a public MRI dataset from the Kaggle platform, show that the proposed MRCG can achieve 88.64% mAP and 86.59% mAP, respectively. Compared with some other state-of-the-art models, the MRCG can also outperform all the baseline models.

Many real-world networks including the World Wide Web and the Internet of Things are graphs in their abstract forms. Graph neural networks (GNNs) have emerged as the main solution for deep learning on graphs. Recently, tremendous effort has been made to enhance the performance and expressivity of GNNs. In this paper, we review the state-of-the-art graph neural network models and frameworks with a focus on the latest developments in graph representation learning. We propose a new taxonomy which divides general GNNs into recurrent GNNs, spectral GNNs, spatial GNNs and topology-aware GNNs. We will also discuss the inductive biases behind different categories of GNNs.

Large volumes of monitoring and logging data result from the operation of a large scale astrophysical observatory. In the last few years several “Big Data” technologies have been developed to deal with such volumes of data especially in the Internet of Things (IoT) framework. We present the logging, monitoring and alarm system architecture for the ASTRI Mini-Array aimed at supporting the analysis of scientific data and improving the operational activities of the telescope facility . A prototype was designed and built considering the latest software tools and concepts coming from Big Data and IoT and a particular relevance has been given in satisfying quality requirements such as performance, scalability and availability.

In the dynamic environment of the Internet of Things (IoT), edge and cloud computing play critical roles in analysing and storing data from numerous connected devices to produce valuable insights. Efficient resource allocation and workload distribution are vital to ensuring continuous and reliable service in growing IoT ecosystems with increasing data volumes and changing application demands. This study proposes a novel optimisation approach utilising deep learning to tackle these challenges. The integration of Deep Q-Networks (DQN) and Proximal Policy Optimization (PPO) offers a practical approach to addressing the dynamic characteristics of IoT applications. The hybrid algorithm's primary characteristic is its capacity to simultaneously fulfil multiple objectives, including reducing response times, enhancing resource efficiency, and decreasing operational costs. DQN facilitates the formulation of optimal resource allocation strategies in intricate and unpredictable environments. PPO enhances policies in continuous action spaces to guarantee reliable performance in real-time, dynamic IoT settings. This method achieves an optimal equilibrium between policy learning and optimisation, rendering it suitable for contemporary IoT systems. This method improves numerous IoT applications, including smart cities, industrial automation, and healthcare. The hybrid DQN-PPO-GNN-RL model addresses bottlenecks by dynamically managing computing and network resources, allowing for efficient operations in low-latency, high-demand environments such as autonomous systems, sensor networks, and real-time monitoring. The use of Graph Neural Networks (GNNs) improves the accuracy of resource representation, while reinforcement learning-based scheduling allows for seamless adaptation to changing workloads. Simulations using real-world IoT data on the iFogSim platform showed significant improvements: task scheduling time was reduced by 21%, operational costs by 17%, and energy consumption by 22%. The method reliably provided equitable resource distribution, with values between 0.93 and 0.99, guaranteeing efficient allocation throughout the network. This hybrid methodology establishes a novel benchmark for scalable, real-time resource management in extensive, data-centric IoT ecosystems, consequently enhancing system performance and operational efficiency.

The fast developments in micro-computing, mini-hardware manufacturing, and machine-to-machine communications have paved the way for innovative Internet of Things (IoT) solutions that are reshaping a great deal of networking software. The Internet of Things (IoT) has introduced a new branch of IoT that is known as the Internet of Medical Things (IoMT) systems. One of the applications that have been transformed by IoT is the healthcare system. Remote monitoring of patients suffering from chronic conditions is made possible by IoMT devices. As a result, it can deliver rapid diagnostics for patients, which in the event of an emergency may save their life. However, ensuring the safety of these vital systems is one of the primary obstacles standing in the way of their widespread use. The objective of this paper is to detect the replication attack, i.e., DoS (Denial-of-service) attack by using multilayer perceptron (MLP) classification algorithm in graph neural network. The proposed approach achieves better detection accuracy of 98.4% when compared with existing state-of-the-art classification models.

SecEdge: A novel deep learning framework for real-time cybersecurity in mobile IoT environments.

Automatic recognition systems (ARS) have been proposed in scientific and technological research for the care and preservation of endangered species; these systems, consisting of Internet of Things (IoT) devices and object-recognition techniques with artificial intelligence (AI), have emerged as proposed solutions to detect and prevent parasite attacks on Apis mellifera bees. This article presents a pilot ARS for the recognition and analysis of honeybees at the hive entrance using IoT devices and automatic object-recognition techniques, for the early detection of the Varroa mite in test apiaries. Two object-recognition techniques, namely the k-Nearest Neighbor Algorithm (kNN) and Graph Neural Network (GNN), were evaluated with an image dataset of 600 images from a single beehive. The results of the experiments show the viability of using GNN in real environments. GNN has greater accuracy in bee recognition, but with greater processing time, while the kNN classifier requires fewer processing resources but has lower recognition accuracy.

Robust explanations for graph neural network with neuron explanation component

GraphNEI: A GNN-based network entity identification method for IP geolocation

The number of Internet of Things (IoT) devices has exploded in recent years. Due to the simple implementation and difficult-to-patch firmware, IoT devices are vulnerable to malware attacks. Static analysis is a feasible way to understand the behavior of IoT malware for detection and mitigation. However, unlike traditional malware on personal computers or smartphones, the diversity of processor architecture on IoT devices brings a variety of challenges for researchers. Current malware detection methods based on operation code or byte code cannot address the multi-architecture issue well. In this paper, we propose a cross-architecture IoT malware detection method based on graph neural network(GNN). We represent each binary file as a function call graph(FCG), since FCG is a higher-level architecture-independent feature. Natural language processing model is used to extract semantic information from operation code in our method. Enable semantic information as node feature and then we use GNN to extract structural information from FCG. Our method takes both semantic and structural information into account to identify malware. We also create a dataset that covers 5 different processor architectures to evaluate our method. The experiment we conduct over the dataset shows that our method performs better than other methods and is capable to detect unknown malware.

In concert with advances of wireless technologies in facilitating internet connectivity of Internet of Things (IoT) devices, mobile edge computing can provision and distribute computing resources at the cloudlets to efficiently process a high volume of IoT data. Among the IoT applications, multi-party data sharing among IoT devices, wireless access nodes and cloudlets is becoming increasingly critical, not only because the data collected by each single IoT device will often stay unmined, but also because of the security concern. As IoT applications' dependence on the cloud environment grows, the rich resources at cloudlets often become the attack targets, and the IoT data that are stored or processed using the cloud resources will be jeopardized. For the internet of important things, we have investigated how to efficiently and securely share the data among multi-party. In particular, for a group of cooperative IoT devices, by leveraging the cloud resources available at the wireless access points, a secure cache site with fast data uploading rate is chosen for each user. To minimize the overall data downloading time, the multi-party multi-path data delivery scheme is also designed such that each user can efficiently retrieve the data belonging to other parties.