Abstract

The shift from monolithic network architectures to dynamic, cloud-native environments was observed to have required a fundamental change in how network security was enforced. Traditional perimeter-based firewalls were determined to be insufficient for the ephemeral, micro-segmented nature of containerized applications and 5G Open RAN (O-Cloud) infrastructures. In this research, the architectural evolution, algorithmic optimization, and autonomous governance of Cloud-Native Firewalls (CNFs) were analyzed. Through a systematic review of open-access studies from IEEE, ACM, MDPI, Wiley, Springer, and Taylor & Francis, the mechanisms of autonomous policy generation were investigated. It was observed that technologies such as Extended Berkeley Packet Filter (eBPF) and hardware-accelerated programmable data planes were utilized to achieve high-performance enforcement. Furthermore, autonomous optimization algorithms employing Deep Learning (DL) were found to significantly mitigate administrative latency and human error.