Efficient digital forensics in the IoT environment: a hybrid framework using deep-federated learning
Year: 2026
Type: article
Abstract: In an era of interconnected devices, robust cybersecurity is essential. This research presents a deep learning-based forensics framework for investigating and identifying cyber-attacks in IoT ecosystems. At its core, a hybrid CNN-LSTM model, enhanced by particle swarm optimisation (PSO), dynamically optimises parameters for peak performance. Integrating federated learning (FL), the framework ensures effective generalisation across diverse IoT datasets while preserving data privacy. This lightweight yet highly accurate solution outperforms existing models in accuracy and efficiency. The proposed framework achieves 97.66% accuracy and improves time efficiency by 76.82%, detecting various cyber-attacks across IoT applications such as vehicle networks, smart homes, and smart cities. This advancement strengthens IoT security and provides an efficient method for tracing malicious activities.
Authors: Waad Almadud, Asma Abdulghani Al Shargabi
Institution: Qassim University
Language: English
Topic: Digital and Cyber Forensics
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7119528865