Securing internet of things using machine and deep learning methods: a survey

Cluster Computing, Journal Year: 2024, Volume and Issue: 27(7), P. 9065 - 9089

Published: April 16, 2024

Abstract The Internet of Things (IoT) is a vast network devices with sensors or actuators connected through wired wireless networks. It has transformative effect on integrating technology into people’s daily lives. IoT covers essential areas such as smart cities, homes, and health-based industries. However, security privacy challenges arise the rapid growth applications. Vulnerabilities node spoofing, unauthorized access to data, cyberattacks denial service (DoS), eavesdropping, intrusion detection have emerged significant concerns. Recently, machine learning (ML) deep (DL) methods significantly progressed are robust solutions address these issues in devices. This paper comprehensively reviews research focusing ML/DL approaches. also categorizes recent studies based highlights their opportunities, advantages, limitations. These insights provide potential directions for future challenges.

Language: Английский

---

Current research on Internet of Things (IoT) security protocols: A survey

Computers & Security, Journal Year: 2025, Volume and Issue: unknown, P. 104310 - 104310

Published: Jan. 1, 2025

Language: Английский

---

AWE-DPFL: Adaptive weighting and dynamic privacy budget federated learning for heterogeneous data in IoT

Computers & Electrical Engineering, Journal Year: 2025, Volume and Issue: 123, P. 110070 - 110070

Published: Jan. 22, 2025

Language: Английский

---

Future Outdoor Safety Monitoring: Integrating Human Activity Recognition with the Internet of Physical–Virtual Things

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(7), P. 3434 - 3434

Published: March 21, 2025

The convergence of the Internet Physical–Virtual Things (IoPVT) and Metaverse presents a transformative opportunity for safety health monitoring in outdoor environments. This concept paper explores how integrating human activity recognition (HAR) with IoPVT within can revolutionize public safety, particularly urban settings challenging climates architectures. By seamlessly blending physical sensor networks immersive virtual environments, highlights future where real-time data collection, digital twin modeling, advanced analytics, predictive planning proactively enhance well-being. Specifically, three dimensions humans, technology, environment interact toward measuring health, climate. Three cultural scenarios showcase to utilize HAR–IoPVT sensors external staircases, rural climate, coastal infrastructure. Advanced algorithms analytics would identify potential hazards, enabling timely interventions reducing accidents. also societal benefits, such as proactive monitoring, enhanced emergency response, contributions smart city initiatives. Additionally, we address challenges research directions necessary realize this future, emphasizing AI technical scalability, ethical considerations, importance interdisciplinary collaboration designs policies. articulating an AI-driven HAR vision along required advancements edge-based fusion, responsiveness fog computing, social through cloud aim inspire academic community, industry stakeholders, policymakers collaborate shaping technology profoundly improves enhances enriches quality life.

Language: Английский

---

Enhancing IoT Healthcare with Federated Learning and Variational Autoencoder

Sensors, Journal Year: 2024, Volume and Issue: 24(11), P. 3632 - 3632

Published: June 4, 2024

The growth of IoT healthcare is aimed at providing efficient services to patients by utilizing data from local hospitals. However, privacy concerns can impede sharing among third parties. Federated learning offers a solution enabling the training neural networks while maintaining data. To integrate federated into healthcare, hospitals must be part network jointly train global central model on server. Local using their patient datasets and send trained localized models These are then aggregated enhance process. aggregation dramatically influences performance training, mainly due heterogeneous nature Existing solutions address this issue iterative, slow, susceptible convergence. We propose two novel approaches that form groups efficiently assign weightage considering essential parameters vital for training. Specifically, our method utilizes an autoencoder extract features learn divergence between latent representations groups, facilitating more handling heterogeneity. Additionally, we another process several factors, including extracted data, maximize further. Our proposed group formation weighting outperform existing conventional methods. Notably, significant results obtained, one which shows achieves 20.8% higher accuracy 7% lower loss reduction compared

Language: Английский

---

Efficient Distributed Denial of Service Attack Detection in Internet of Vehicles Using Gini Index Feature Selection and Federated Learning

Future Internet, Journal Year: 2025, Volume and Issue: 17(1), P. 9 - 9

Published: Jan. 1, 2025

Considering that smart vehicles are becoming interconnected through the Internet of Vehicles, cybersecurity threats like Distributed Denial Service (DDoS) attacks pose a great challenge. Detection methods currently face challenges due to complex and enormous amounts data inherent in IoV systems. This paper presents new approach toward improving DDoS attack detection by using Gini index feature selection Federated Learning during model training. The assists filtering out important features, hence simplifying models for higher accuracy. FL enables decentralized training across many devices while preserving privacy allowing scalability. results show case this is detecting attacks, bringing confidentiality, reducing computational load. As noted paper, average accuracy 91%. Moreover, different types were identified employing our proposed technique. Precisions achieved as follows: DrDoS_DNS: 28.65%, DrDoS_SNMP: 28.94%, DrDoS_UDP: 9.20%, NetBIOS: 20.61%. In research, we foresee potential harvesting from integrating advanced with so systems can meet modern requirements. It also provides robust efficient solution future automotive industry. By carefully selecting only most features decentralizing devices, reduce both time memory usage. makes system much faster lighter on resources, making it perfect real-time applications. Our effective environments.

Language: Английский

---

Privacy-Preservation for Federated Learning: Survey and Future Directions

Communications in computer and information science, Journal Year: 2025, Volume and Issue: unknown, P. 311 - 325

Published: Jan. 1, 2025

Language: Английский

---

Heterogeneity Challenges of Federated Learning for Future Wireless Communication Networks

Journal of Sensor and Actuator Networks, Journal Year: 2025, Volume and Issue: 14(2), P. 37 - 37

Published: April 1, 2025

Two technologies of great interest in recent years—Artificial Intelligence (AI) and massive wireless communication networks—have found a significant point convergence through Federated Learning (FL). is Machine (ML) technique that enables multiple participants to collaboratively train model while keeping their data local. Several studies indicate improving performance metrics—such as accuracy, loss reduction, or computation time—is primary goal, achieving this real-world scenarios remains challenging. This difficulty arises due various heterogeneity characteristics inherent the devices participating Federation. Heterogeneity when contribute differently, leading challenges training process. may appear architecture, statistics, behavior. System from differences device capabilities, including processing power, transmission speeds, availability, energy constraints, network limitations, among others. Statistical occurs non-independent non-identically distributed (non-IID) data. situation can harm global instead it, especially are poor quality too scarce. The third type, behavioral heterogeneity, refers cases where unwilling engage expect rewards despite minimal effort. Given growing research area, we present summary provide broader perspective on emerging technology. We also outline key challenges, opportunities, future directions for Learning. Finally, conduct simulation using LEAF framework illustrate impact

Language: Английский

---

A Security-Enhanced Federated Learning Scheme Based on Homomorphic Encryption and Secret Sharing

Mathematics, Journal Year: 2024, Volume and Issue: 12(13), P. 1993 - 1993

Published: June 27, 2024

Although federated learning is gaining prevalence in smart sensor networks, substantial risks to data privacy and security persist. An improper application of techniques can lead critical breaches. Practical effective privacy-enhanced (PEPFL) a widely used framework characterized by low communication overhead efficient encryption decryption processes. Initially, our analysis scrutinized vulnerabilities within the PEPFL identified an attack strategy. This strategy enables server derive private keys from content uploaded participants, achieving 100% success rate extracting participants’ information. Moreover, when number participants does not exceed 300, time surpass 3.72 s. Secondly, this paper proposes model that integrates homomorphic secret sharing. By using sharing among instead secure multi-party computation, amount information available servers reduced, thereby effectively preventing inferring gradients. Finally, scheme was validated through experiments, it found significantly reduce inherent collusion unique scenario. even if some are unavailable, reconstructable nature ensures process continue uninterrupted, allowing remaining users proceed with further training. Importantly, proposed exerts negligible impact on accuracy

Language: Английский

---

Asynchronous Privacy-Preservation Federated Learning Method for Mobile Edge Network in Industrial Internet of Things Ecosystem

Electronics, Journal Year: 2024, Volume and Issue: 13(9), P. 1610 - 1610

Published: April 23, 2024

The typical industrial Internet of Things (IIoT) network system relies on a real-time data upload for timely processing. However, the incidence device heterogeneity, high latency, or malicious central server during transmission has propensity privacy leakage loss model accuracy. Federated learning comes in handy, as edge requires less time and enables local processing to reduce delay upload. It allows neighboring nodes share while maintaining confidentiality. this can be challenged by disruption making sensors go offline experience an alteration process, thereby exposing already transmitted that eavesdrops channel, intercepts transit, gleans information, evading within network. To mitigate effect, paper proposes asynchronous privacy-preservation federated mobile networks IIoT ecosystem (APPFL-MEN) incorporates iteration design update strategy (IMDUS) scheme, enabling more updates with online sharing nodes, without node hack. In addition, it adopts double-weight modification communication between gateway enhanced training process. Furthermore, convergence boosting resulting error-prone, secured global model. performance evaluation numerical results shows good accuracy, efficiency, lower bandwidth usage APPFL-MEN preserving compared state-of-the-art methods.

Language: Английский

---