Towards Cloud-Native Agentic Protocol Learning for Conflict-Free 6G: A Case Study on Inter-Slice Resource Allocation

TL;DR

The paper addresses efficient, conflict-free resource sharing across network slices in a 6G edge environment by introducing a cloud-native, Dockerized multi-agent DRL framework with emergent inter-agent communication. It leverages a Kafka-based messaging backbone and real-time Prometheus/Grafana monitoring to coordinate per-slice agents while preserving data privacy. Key contributions include the cloud-native MADRL architecture, emergent communication via learned messages, and demonstrated gains in utilization and latency reduction with a lower incidence of conflicts under synthetic traffic that models eMBB, URLLC, and mMTC. The results indicate the approach is scalable, adaptable to dynamic traffic, and suitable as a foundation for autonomous, self-optimizing inter-slice resource management in future 6G networks.

Abstract

In this paper, we propose a novel cloud-native architecture for collaborative agentic network slicing. Our approach addresses the challenge of managing shared infrastructure, particularly CPU resources, across multiple network slices with heterogeneous requirements. Each network slice is controlled by a dedicated agent operating within a Dockerized environment, ensuring isolation and scalability. The agents dynamically adjust CPU allocations based on real-time traffic demands, optimizing the performance of the overall system. A key innovation of this work is the development of emergent communication among the agents. Through their interactions, the agents autonomously establish a communication protocol that enables them to coordinate more effectively, optimizing resource allocations in response to dynamic traffic demands. Based on synthetic traffic modeled on real-world conditions, accounting for varying load patterns, tests demonstrated the effectiveness of the proposed architecture in handling diverse traffic types, including eMBB, URLLC, and mMTC, by adjusting resource allocations to meet the strict requirements of each slice. Additionally, the cloud-native design enables real-time monitoring and analysis through Prometheus and Grafana, ensuring the system's adaptability and efficiency in dynamic network environments. The agents managed to learn how to maximize the shared infrastructure with a conflict rate of less than 3%.

Figures (5)

• Figure 1: Docker-based protocol learning architecture.
• Figure 2: Average reward and average conflict throughout the training process.
• Figure 3: Conflicts heat map vs. different MA models.
• Figure 4: Latency cumulative distribution function for MA IB
• Figure 5: Utilization vs. episodes