Scholarlink

The transition to 5G networks opens up a great deal of opportunities for digital services but at the same time, it can only be implemented with the help of more advanced great solutions to cope with the intricacies and security concerns of net- work traffic management. The concept of verticality is very essential as it allows several logical networks to coexist on a common physical infrastructure with each optimized to resolve the various requirements of the different applications and services in a given telecommunications environment. Allocating the ideal slice to a user dynamically as well as addressing the traffic growth and security issues continues to present visible challenges. In this paper, we propose a deep learning based network slicing management approach that focuses on revenue enhancement via slice allocation optimization while risk is constantly managed. In our approach, we employ advanced neural network architectures who are designed to classify the network traffic for efficiency and use ensemble classifiers to enhance classification performance. Classification accuracy, precision, recall and F1 score of our model built on HAI Security Dataset, considerably higher than other conventional machine learning approaches. In addition, performance of our model under the conditions of network faults is quite promising, where 97.8% of the cases requiring traffic reallocation when main network slice becomes unavailable were handled successfully. These findings illustrate the importance of management processes supported by AI network slicing in the efficient functioning of the future 5G, primarily in terms of risks mitigation and NP service continuity assurance.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimizing 5G Network Slicing for Threat Mitigation and Failure Resilience Using Deep Learning Techniques

  • 章节

  • 发表日期：

    2025年

查看全文
  • Vinoth Kumar Kolluru,
  • Likhit Sai Eswar Kalla,
  • Yagnesh Challagundla,
  • Advaitha Naidu Chintakunta,
  • Manik Chand Patnaik

摘要

The transition to 5G networks opens up a great deal of opportunities for digital services but at the same time, it can only be implemented with the help of more advanced great solutions to cope with the intricacies and security concerns of net- work traffic management. The concept of verticality is very essential as it allows several logical networks to coexist on a common physical infrastructure with each optimized to resolve the various requirements of the different applications and services in a given telecommunications environment. Allocating the ideal slice to a user dynamically as well as addressing the traffic growth and security issues continues to present visible challenges. In this paper, we propose a deep learning based network slicing management approach that focuses on revenue enhancement via slice allocation optimization while risk is constantly managed. In our approach, we employ advanced neural network architectures who are designed to classify the network traffic for efficiency and use ensemble classifiers to enhance classification performance. Classification accuracy, precision, recall and F1 score of our model built on HAI Security Dataset, considerably higher than other conventional machine learning approaches. In addition, performance of our model under the conditions of network faults is quite promising, where 97.8% of the cases requiring traffic reallocation when main network slice becomes unavailable were handled successfully. These findings illustrate the importance of management processes supported by AI network slicing in the efficient functioning of the future 5G, primarily in terms of risks mitigation and NP service continuity assurance.