In the realm of Fifth Generation (5 G) and the upcoming Sixth Generation (6 G) networks, the efficient management of network resources becomes increasingly critical, particularly for applications that have strict Quality of Service (QoS) requirements. This paper addresses the complexities associated with Virtual Network Embedding (VNE), a vital process for establishing multiple virtual networks on shared physical infrastructure within the context of network slicing. We introduce the SetpodNet scheduler, a novel orchestration solution that leverages reinforcement learning to enhance the optimization of latency and bandwidth allocation specifically in Kubernetes environments. The SetpodNet scheduler is designed to dynamically adapt to fluctuating slice arrivals and varying resource demands, ensuring that network performance remains consistent and reliable. Through comprehensive experimental evaluations, we demonstrate improvements in slice acceptance ratios and optimizing QoS.

5G and beyond networks aim to support heterogeneous services with strict QoS and isolation requirements. Network slicing addresses these challenges by creating multiple virtual networks over shared infrastructure, each tailored for specific service types. However, despite its potential, the lack of practical, open, and reproducible testbeds for 5G slicing remains a major barrier to experimentation and adoption. In this demo, we present NexSlice, an open-source, Kubernetes-Native testbed that enables the deployment and orchestration of 5G core network slices using open-source components like OpenAirInterface (OAI) and UERANSIM. Our platform supports SST-based slicing, integrates different Radio Access Networks (RANs), enables auto-scaling of user plane functions, and provides real-time monitoring with Prometheus and Grafana. NexSlice aims to bridge the gap between theoretical slicing frameworks and practical, reproducible experimentation, paving the way toward adaptive and automated slice management in future 6G networks.

The Fifth and Sixth Generation (5G/6G) networks aim to support diverse applications with specific QoS and resource needs. Network Slicing has emerged as a key paradigm to meet these demands by creating multiple Virtual Networks (VNs) over shared physical infrastructure. This process, known as Virtual Network Embedding (VNE), maps virtual nodes and links to physical resources. With Kubernetes becoming the dominant orchestration platform, most infrastructures now rely on Kubernetes clusters, which enforce direct pod-to-pod communication, necessitating a direct-link approach to VNE. However, most existing methods focus on path-based link mapping. In this paper, we present DL-ViNE, a Reinforcement Learning(RL)-based algorithm that improves slice acceptance while addressing the specific constraints of Kubernetes-hosted infrastructures.