Emulator-Based Configuration on QoS Measurement: A Case Study of DRL Actions on Mininet and RYU Testbed

Deep reinforcement learning (DRL) has found widespread applications in the realm of communication and networking, addressing a multitude of service scenarios. In this study, we offer a detailed case study involving DRL action configuration on the Mininet and RYU testbeds. The working flow guides the dynamic adjustment of software-defined networking (SDN) flows, resource allocation, and NFV service scaling to enhance network elasticity and virtualization. Our experiments are conducted in an emulator testbed by leveraging the networking tools such as Mininet, Mini-NFV, and the RYU SDN controller. The results are captured to illustrate that our proposed approach significantly enhances Quality of Service (QoS) satisfaction, covering key aspects such as latency, throughput, packet loss, and resource utilization. Our study also includes a comparison with established baseline configurations, namely Fixed QoS and Reactive QoS, further illustrating the dynamic provisioning through DRL. The testbed configuration holds the applicability for responsive and reliable network management in core control domains.