Calls for papers

 

International Journal of Ad Hoc and Ubiquitous Computing

 

Special Issue on: "Federated Learning for Edge based IoT Anomaly Detection in Big Data Streams"

Guest Editors:
Dr. Uzair Aslam Bhatti, Hainan University, China
Dr. Muhammad Asim Saleem, Chulalongkorn University, Thailand
Dr. Maqbool Khan, Pak-Austria Fachhochschule-Institute of Applied Sciences and Technology, Pakistan
Dr. Yonis Gulzar, King Faisal University, Saudi Arabia

The federated learning (FL) framework that enables decentralised edge gadgets to collaborate in schooling an anomaly detector with a focused mechanism-based algorithm and artificial neural structure strives for an immediate recollection approach. This allows for the implementation of an anomaly detection approach in a supplied ICS system while maintaining excellent precision and protecting data privacy. Despite using a FL-based strategy, the studies don't provide insight into how well such an instructional framework performs when implemented in an edge context. Federated Learning uses local information sampling produced by sources of big data streams to develop a world-wide shared model. The intrinsic variability of IoT contexts offers problems for integral FL related to appropriate development, transmission expenses, model integration quickly, and precision. These issues arise from differences in hardware, network technologies, and electricity supplies.

With the rise of intelligent healthcare apps and gadgets used in eldercare facilities, intelligent houses, and digital medical facilities, the Internet of Medical Things (IoMT) has become commonplace. In order to detect patients' critical bodily characteristics, analyse well-being, and create data with multiple variables for supporting just-in-time healthcare, it makes use of sophisticated medical equipment, cloud computing services, and core Internet of Things (IoT) technology. The vast majority of these big data streams are analysed on centralised servers. Significant reaction time delays with considerable performance overhead are a common concern for anomaly detection (AD) in the centralised medical environment. Transferring patients' private health information to a centralised server also raises intrinsic privacy concerns and introduces many risk factors to the AD architecture, including the potential threat of information. Another challenge that needs to be solved is educating in a dispersed setting while respecting information privacy. Although FL has numerous benefits, it is not acceptable to use FL directly in IoT devices due to many important obstacles. Each device attempts to modify the version it has to operate appropriately based on its big data, obtaining dispersed, non-identical data from the environment. For this reason, every device has a unique regional model. Furthermore, classical FL cannot effectively address these constraints, as limitations on resources only allow IoT devices to train weak models. In a real IoT setting, conventional FL makes it impossible for all component actions to agree on a single model for collaborative development.

In this special issue, anomaly detection and proactive servicing are carried out using a federated learning framework that considers the distributional changes of period sequence datasets. the big data streams to assess how well a number of IoT structures and data sequential anomaly detection methods operate together.

Subject Coverage
Suitable topics include, but are not limited, to the following:

• Modern healthcare using virtual copies and multilevel federated learning based anomaly detection
• Accessible anomaly detection for commercial supervise systems using federated learning
• Federated learning-based intelligent threat identification in IoT edge computing
• Multiple stages federated learning for issues detection based on cloud-edge-client cooperation
• Precise and reliable federated learning-driven edge intelligence for superior video interpretation
• Implementing a hierarchical edge design with federated learning-based AI Approach
• Supported using federated learning, the secure IoT environment protects against data breaches
• An edge-based poisoning attack-resistant partially private federated learning approach
• Internet of Things-based semi-supervised federated learning assault detection technique
• Intelligent anomaly detection based on information based on data in IoT edge: an evaluation
• Federated learning and algorithms for confidentiality big data safety in IoT

Notes for Prospective Authors

Submitted papers should not have been previously published nor be currently under consideration for publication elsewhere. (N.B. Conference papers may only be submitted if the paper has been completely re-written and if appropriate written permissions have been obtained from any copyright holders of the original paper).

All papers are refereed through a peer review process.

All papers must be submitted online. To submit a paper, please read our Submitting articles page.

Important Dates

Manuscripts due by: 30 August, 2024

Notification to authors: 30 December, 2024

Final versions due by: 30 March, 2025