Developing and Evaluating Generative AI Models for Detection and Mitigation of Security Threats in 5G Networks

Fuente: Zenodo
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Bansal, Mukesh Kumar, Gupta, Mukesh Kumar, Tiwari, Amit
Format: Recurso digital
Sprache:Englisch
Veröffentlicht: Zenodo 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866901635343581184
author Bansal, Mukesh Kumar
Gupta, Mukesh Kumar
Tiwari, Amit
author_facet Bansal, Mukesh Kumar
Gupta, Mukesh Kumar
Tiwari, Amit
contents <p>The rapid advancement of technology has enhanced the connectivity and data exchange but has also introduced challenges of security threats and vulnerabilities. This study explores the development of Generative Artificial Intelligence (GAI) models to detect and mitigate 5G networks threats. The proposed framework integrates Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Large Language Models (LLMs), leveraging their unique strengths for cybersecurity. The hybrid framework achieves the superior performance with an accuracy of 97.5% and detects both known and unknown threats. Metrics such as detection accuracy, false positive rates (FPRs), computational efficiency, and robustness against the adversarial attacks are used to evaluate the system. The framework also demonstrates flexibility to adversarial threats, continuously learning, and improving threats detection and mitigation. The proposed framework of hybrid approach provides an adaptive approach to address new security challenges to the growing field of AI-driven cybersecurity. </p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_17875369
institution Zenodo
language eng
publishDate 2025
publisher Zenodo
record_format zenodo
spellingShingle Developing and Evaluating Generative AI Models for Detection and Mitigation of Security Threats in 5G Networks
Bansal, Mukesh Kumar
Gupta, Mukesh Kumar
Tiwari, Amit
Generative AI
Cybersecurity
Threat Detection
Hybrid Approach
Metrics
<p>The rapid advancement of technology has enhanced the connectivity and data exchange but has also introduced challenges of security threats and vulnerabilities. This study explores the development of Generative Artificial Intelligence (GAI) models to detect and mitigate 5G networks threats. The proposed framework integrates Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Large Language Models (LLMs), leveraging their unique strengths for cybersecurity. The hybrid framework achieves the superior performance with an accuracy of 97.5% and detects both known and unknown threats. Metrics such as detection accuracy, false positive rates (FPRs), computational efficiency, and robustness against the adversarial attacks are used to evaluate the system. The framework also demonstrates flexibility to adversarial threats, continuously learning, and improving threats detection and mitigation. The proposed framework of hybrid approach provides an adaptive approach to address new security challenges to the growing field of AI-driven cybersecurity. </p>
title Developing and Evaluating Generative AI Models for Detection and Mitigation of Security Threats in 5G Networks
topic Generative AI
Cybersecurity
Threat Detection
Hybrid Approach
Metrics
url https://doi.org/10.5281/zenodo.17875369