DDoS attacks remain a critical threat to organizations, disrupting
services and inflicting serious financial and reputational damage.
Traditional defenses in the form of rule-based systems and statistical
models often cannot keep up with the sophistication and evolution of
such attacks. This paper introduces a hybrid GenAI framework,
designed to address these challenges by combining the adaptive
capabilities of advanced generative models with the robustness of
traditional rule-based systems. The proposed multi-layered
architecture detects and analyzes anomalous network traffic patterns
indicative of DDoS attacks using Generative Adversarial Networks
(GANs), autoencoders, and transformers. GANs are utilized for
generating realistic attack scenarios for the training and validation of
models in order to enhance the robustness of the detection system.
Autoencoders identify very subtle anomalies in network traffic due to
reconstruction errors, whereas transformers process sequential traffic
data in order to capture long-term dependencies and detect coordinated
attack behaviors. These advanced generative techniques are integrated
with rule-based defenses that apply predefined thresholds, traffic
filtering, and IP blacklisting for immediate response to known attack
vectors. To combine the strengths of both layers, a decision fusion
module is proposed, which integrates insights from generative models
and rule-based systems using weighted scoring and logical decision
trees. This hybrid approach enhances the detection accuracy of DDoS
attacks, minimizes false positives, and ensures prompt response to
threats. Thorough experiments on real-world DDoS datasets
supplemented with synthetic data generated by GANs demonstrate the
superior performance of the proposed framework in detecting and
mitigating a wide range of DDoS attacks. Results show a sharp increase
in detection rates with a reduction in false positives along with
mitigation times that have improved compared to traditional methods.
Moreover, the system demonstrates adaptability to evolutionary attack
patterns, which signifies its feasibility for real-world deployment into
dynamic network environments. Coupling state-of-the-art generative
AI techniques with mature defense mechanisms, this framework
embodies a new benchmark for resilient and scalable, yet intelligent, DDoS mitigation.
K. Muthamil Sudar
Mepco Schlenk Engineering College, India
DDoS, GANs, Autoencoders, GEN AI, Cybersecurity
| January | February | March | April | May | June | July | August | September | October | November | December |
| 3 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Published By :
ICTACT
Published In :
ICTACT Journal on Soft Computing
( Volume: 15 , Issue: 3 , Pages: 3618 - 3624 )
Date of Publication :
January 2025
Page Views :
60
Full Text Views :
5
|