Adversarial Robust Reinforcement Learning for Secure and Regulation-Aware Internet of Things Systems

Mohammed Farsi; Elsayed Atlam

doi:10.2478/ias-2026-0005

.blurhash-client-img { display: none !important; }

Adversarial Robust Reinforcement Learning for Secure and Regulation-Aware Internet of Things Systems

Journal of Information Assurance and Security

Volume 21 (2026): Issue 2 (July 2026)

By: Mohammed Farsi and Elsayed Atlam

Open Access

|Jul 2026

Figures & Tables

Bird’s-eye view of the proposed adversarially robust and regulation-aware reinforcement learning framework.

Overview of adversarially robust and regulation-aware reinforcement learning policy optimization. The framework integrates robust optimization, uncertainty-aware reward shaping, and dynamic regulation-aware constraints into a unified policy optimization process.

Decentralized and Federated Reinforcement Learning Framework for Robust and Regulation-Aware IoT Intrusion Response

Policy Stability and Security Degradation Under Adversarial Learning-Loop Attacks

Policy Stability Under Decentralized and Federated Learning with Malicious Agents

Regulation Compliance and Robust Performance Trade-off

Dataset	Method	Compliance Violations (%) ↓	Avg. Constraint Slack ↓	Security Performance (%) ↑
NSL-KDD	Static Constraints	18.6	0.041	84.3
	Proposed	3.2	0.009	88.7
UNSW-NB15	Static Constraints	24.1	0.053	81.6
	Proposed	4.8	0.012	86.9
BoT-IoT	Static Constraints	21.7	0.047	79.2
	Proposed	5.5	0.015	83.8

Stability in Decentralized and Federated Learning

Dataset	Method	Policy Divergence (\|\| θ – θ^\* \|\|) ↓	Malicious Agents (%)	Global Performance Drop (%) ↓
NSL-KDD	Federated RL	0.217	20	26.5
	Proposed	0.082	20	9.4
UNSW-NB15	Federated RL	0.284	25	33.8
	Proposed	0.096	25	11.8
BoT-IoT	Federated RL	0.312	30	37.4
	Proposed	0.124	30	14.6

Policy Stability and Security Degradation under Learning-Loop_

Dataset	Method	Policy Instability (Var[J]) ↓	Convergence Steps ↓	Security Degradation (%) ↓
NSL-KDD	Classical RL	0.182	4,200	31.4
	Proposed	0.061	2,750	11.2
UNSW-NB15	Classical RL	0.247	5,100	38.9
	Proposed	0.073	3,100	13.6
BoT-IoT	Classical RL	0.301	5,800	42.7
	Proposed	0.098	3,950	17.9

References

Authors

Metrics

Articles in this issue

DOI: https://doi.org/10.2478/ias-2026-0005 | Journal eISSN: 1554-1029 | Journal ISSN: 1554-1010

Journal RSS Feed

Language: English

Page range: 89 - 106

Published on: Jul 8, 2026

Published by: Cerebration Science Publishing Co., Limited

In partnership with: Paradigm Publishing Services

Publication frequency: 6 issues per year

Keywords:

Reinforcement learning,

Secure learning,

IOT,

Robust adversarial-aware

Related subjects:

Computer sciences,

Fundamentals of computer sciences,

Theoretical computer sciences,

IT-security and cryptology

© 2026 Mohammed Farsi, Elsayed Atlam, published by Cerebration Science Publishing Co., Limited
This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License.

Volume 21 (2026): Issue 2 (July 2026)

Adversarial Robust Reinforcement Learning for Secure and Regulation-Aware Internet of Things Systems

Figures & Tables

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Regulation Compliance and Robust Performance Trade-off

Stability in Decentralized and Federated Learning

Policy Stability and Security Degradation under Learning-Loop_

Paradigm

My account