Optimized Deep Learning for Intrusion Detection and Threat Analysis in Cloud–Fog Environments
Year: 2026
Type: conference-paper
Abstract: Cloud-fog computing now emerges as a paradigm to help serve latency-various and large-scale applications in domains like the smart infrastructure, healthcare, and industrial systems. But these environments are distributed and decentralized, which increases the security risks greatly and thus efficient intrusion detection and threat analysis is highly essential research topic. In this paper, the authors have described an optimized research-based intrusion detection framework, which applies deep learning in cloud-fog environments. The algorithm is a combination of normalized traffic preprocessing, feature representation optimization and multi-layer deep neural network to learn attack patterns effortlessly and with high precision. A synchronized cloud-fog analysis system allows local detection to be performed at fog nodes at low latency and amplify information about the global network of threats by using the global intelligence located in the cloud. The effectiveness of the approach is proven by experimental testing using benchmark intrusion detection data sets. The given model is detected with the accuracy of 98.6 awareness, which is more than the conventional machine learning algorithms (by about 6.8%), and raw deep learning models (by 4.2). Also, the rate of false positive is minimized almost by 35 percent and less freedom is noted during fog nodes. The findings validate the appropriateness of optimized deep learning to secure and scalable cloud fog infrastructures.
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7203676098