With the emergence of intelligent services and applications powered by artificial intelligence (AI), the Internet of Things (IoT) affects many aspects of our daily lives. Traditional approaches to machine learning (ML) relied on centralized data collection and processing, where data was collected and analyzed in one place. However, with the development of Deep Federated Learning (DFL), models can now be trained on decentralized data, reducing the need for centralized data storage and processing. In this work, we provide a detailed analysis of DFL and its benefits, followed by an extensive survey of the use of DFL in various IoT services and applications. We have studied the impact of DFL and how to preserve security and privacy by ensuring compliance in machine learning-enabled IoT systems. In addition, we present a generic architecture for a GDPR-compliant DFL-based framework. Finally, we discuss the existing obstacles and possible future research directions for DFL in IoT.

Deep Federated Learning (DFL) revolutionizes machine learning (ML) by enabling collaborative model training across diverse, decentralized data sources without direct data sharing, emphasizing user privacy and data sovereignty. Despite its potential, DFL’s application in sensitive sectors is hindered by challenges in meeting rigorous standards like the GDPR, with traditional setups struggling to ensure compliance and maintain trust. Addressing these issues, our research introduces an innovative Zero Trust-based DFL architecture designed for GDPR compliant systems, integrating advanced security and privacy mechanisms to ensure safe and transparent cross-node data processing. Our base paper proposed the basic GDPR-Compliant DFL Architecture. Now we validate the previously proposed architecture by formally verifying it using High-Level Petri Nets (HLPNs). This Zero Trust-based framework facilitates secure, decentralized model training without direct data sharing. Furthermore, we have also implemented a case study using the MNIST and CIFAR-10 datasets to evaluate the existing approach with the proposed Zero Trust-based DFL methodology. Our experiments confirmed its effectiveness in enhancing trust, complying with GDPR, and promoting DFL adoption in privacy-sensitive areas, achieving secure, ethical Artificial Intelligence (AI) with transparent and efficient data processing.

A Systematic Literature Review (SLR) was conducted using tailored searches based on our study topic. We completed all SLR processes, including periodic reviews as SLR. Researchers may find out about the justification, the review procedure, and the research question by using search keywords. This paper describes the trial approach to elaborate the search keywords, resources, restrictions, and validations that were, and explores search strategies made. The reviews are carried out by assessing the publication's quality, devising a data extraction approach, and synthesizing the results. All four research questions were used to analyze the papers concerning the findings. Finally, reports on the categorization of computer malware were analyzed for their detection methods, factors, and how they infiltrate computer systems have been published. SLR identifies the element, characteristics, and detection techniques that are explained in this research paper. Computer malware infects the computer system. This comprehensive literature review's is mainly based on recommendations by earlier studies.

Federated Learning (FL) can be defined as an effective solution for using the benefits of machine learning (ML) in distributed systems, in which the data of the clients remain protected. However overlaid challenges affecting today’s FL systems are centered on time optimization, scalability, and security. To these challenges, this paper responds with a new solution comprising the dynamic cohort creation via smart contracts and a hierarchical blockchain approach. Firstly, the research utilizes smart contracts for the dynamic tracking of cohorts in real time and addresses a two-tier blockchain structure for safety and efficiency in storage. In addition, Gaussian Differential Privacy (GDP) is used as a privacy-preserving mechanism that adds controlled noise to the data or model updates to protect individual data points from being inferred by adversaries. The method we are proposing includes four practical steps that include efficient update validation and aggregation; this will enhance training time, and increase model accuracy as well as data confidentiality. The standard dataset is used to show enhanced performance and scalability which validates this method. Based on the above investigations, it could be concluded that the proposed approach improves FL efficiency and creates a new direction in the development of secure, accurate, and scalable ML. The present study indicates that the implementation of blockchain with FL fortified by GDP will establish a novel innovation between intelligent and safe Artificial Intelligence (AI) architecture for safeguarding the privacy of ML system.