Journal of Information and Communications Technology: Algorithms, Systems And Applications
A single-blind peer-reviewed, quarterly, open-access journal committed to advancing cutting-edge research across the full spectrum of ICT.
Intrusion Detection in Transition: A Survey of Deep Learning, Federated Learning, Adversarial, Lightweight, and Explainable Approaches
J. Inf. Commun. Technol. Algorithms Syst. Appl., 2025, 1(3), 25314 https://doi.org/10.64189/ict.25314
Received: 27 October 2025 | Revised: 17 December 2025 | Accepted: 19 December 2025
Cite article
K. M. Dhule, R. Bansode, Intrusion detection in transition: a survey of deep learning, federated learning, adversarial, lightweight, and explainable approaches, Journal of Information and Communications Technology: Algorithms, Systems and Applications, 2025, 1(3), 25314, doi: . https://doi.org/10.64189/ict.25314
Rights and permissions
(c) The Author(s) 2025.
Open Access
This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits the non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as appropriate credit is given and changes are indicated. https://creativecommons.org/licenses/by-nc/4.0/
Abstract
Intrusion detection systems (IDS) are still among the central systems in protecting networked settings in contemporary cybersecurity studies. The threats to cyberspace are growing in scale and in severity, and therefore, the challenges to cybersecurity are growing more complex, in a way that necessitates an ongoing evolution of protective systems and the requirement to adjust to the new threats that appear constantly. This review analyzes the new developments in the field of IDS by reviewing new methodological strategies, datasets used for analysis and solution development. It outlines five major trends: the adoption of deep-learning models; federated learning in IDS architectures; resistance to adversarial perturbations; lightweight variants for IoT devices; and Explainable Artificial Intelligence (XAI) approaches to make model behavior interpretable. Datasets in the survey include CIC-IDS2017, UNSW-NB15, and IoT traffic logs, and dwells on problems such as performance evaluation, reproducibility, and benchmarking. These innovations placed in the context of existing challenges provide a holistic description of the evolution of IDS technology and provide important information to researchers as well as industry players.
Graphical Abstract
Novelty Statement
This survey presents a holistic analysis of modern intrusion detection systems, identifying emerging challenges and providing future research directions.