A Hybrid Deep Learning Model with Consensus-Based Feature Selection for DDoS Attacks Detection in SDN

Procedia Computer Science, Journal Year: 2025, Volume and Issue: 252, P. 643 - 652

Published: Jan. 1, 2025

Language: Английский

---

Deep Reinforcement Learning Based Flow Aware‐QoS Provisioning in SD‐IoT for Precision Agriculture

Computational Intelligence, Journal Year: 2025, Volume and Issue: 41(1)

Published: Feb. 1, 2025

ABSTRACT To meet the demands of modern technologies such as 5G, big data, edge computing, precision, and sustainable agriculture, combination Internet‐of‐Things (IoT) with software‐defined networking (SDN) known SD‐IoT is suggested to automate network by leveraging programmable centralized SDN interfaces. The previous literature has quality‐of‐service (QoS) aware flow processing using manual strategies or heuristic algorithms, however, these schemes proposed white‐box approaches do not provide effective results scales dynamic changes are happening. This article proposes a novel QoS provision strategy deep reinforcement learning (DRL) calculate optimal routes autonomously for traffic. satisfy different flows in divided into two types. Hence, based on their service demand generated them per request. scenario explained precision agriculture compared benchmark strategies. A real internet topology used evaluation results. indicated that method gives improvements delay, throughput, packet loss rate, jitter models.

Language: Английский

---

High-speed threat detection in 5G SDN with particle swarm optimizer integrated GRU-driven generative adversarial network

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 23, 2025

Abstract Detecting attacks in 5G software-defined network (SDN) environments requires a comprehensive approach that leverages traditional security measures, such as firewalls, intrusion prevention systems, and specialized techniques personalized to the unique characteristics of network. The attack detection SDN involves Machine learning (ML) Deep (DL) algorithms analyze large volumes data identify patterns indicative attacks. study’s main objective is develop an efficient DL model improve performance respond breaches effectively environment. integrates Particle Swarm Optimizer-Gated Recurrent Unit Layer-Generative Adversarial Network-Intrusion Detection System classifier (PSO-GRUGAN-IDS). PSO optimizes weight GAN backpropagation while generating synthetic (attack data) generator using GRU. discriminator uses PSO-optimized produce real forecast attack. Finally, deep classification (IDS) trained GRU with model-produced classify whether traffic malicious or normal. Moreover, this evaluated InSDN dataset compared existing model-based approaches results demonstrate significantly higher accuracy rate 98.4%, precision 98%, recall 98.5%, less time 2.464 s, lesser Log loss 1.0 more metrics instilling confidence effectiveness proposed method.

Language: Английский

---

Generative Adversarial Network Models for Anomaly Detection in Software-Defined Networks

Journal of Network and Systems Management, Journal Year: 2024, Volume and Issue: 32(4)

Published: Sept. 12, 2024

Language: Английский

---

A Comprehensive Survey on Generative AI Solutions in IoT Security

Electronics, Journal Year: 2024, Volume and Issue: 13(24), P. 4965 - 4965

Published: Dec. 17, 2024

The influence of Artificial Intelligence in our society is becoming important due to the possibility carrying out analysis large amount data that increasing number interconnected devices capture and send as well making autonomous instant decisions from information machines are now able extract, saving time efforts some determined tasks, specially cyberspace. One key issues concerns security this cyberspace controlled by machines, so system can run properly. A particular situation, given heterogeneous special nature environment, case IoT. limited resources components such a network distributed topology make these types environments vulnerable many different attacks leakages. capability Generative generate contents autonomously learn predict situations be very useful for automatically instantly, significantly enhancing IoT systems. Our aim work provide an overview Intelligence-based existing solutions diverse set try anticipate future research lines field delve deeper.

Language: Английский

---