Multi-layer multi-technology firewall optimisation in beyond 5G networks using machine learning classifiers

Enhancing the security of Beyond 5G (B5G) and Pre-6G networks poses significant challenges, particularly in effectively implementing firewalls. Within a wide range of technologies aimed at implementing mitigation mechanisms, achieving optimal technology selection and rule set configuration within these diverse technologies is immensely complex. In addition, these rules are usually based on pre-configured template and lack of optimisation with information of real-time network status, often resulting in sub-optimal configurations. In this paper, an architecture that enables the optimisation of multi-layer multi-technology firewalls integrated in a B5G network testbed is presented. Our proposed framework supports network control monitoring and automatic deployment of firewall rules in three different virtual function implementations: iptables, Open vSwitch and Linux traffic control. After performing a comparison among four popular machine learning (ML) models for the optimal selection, our results show that Random Forest is the best algorithm for the proposed solution with a F1-score of 0.9083.