Compositional Learning for Modular Multi-Agent Self-Organizing Networks

2025-06-03

AI Summary

To address the challenges of strong parameter coupling and conflicting multi-objective optimization in self-organizing networks (SONs), this paper proposes a modular two-tier multi-agent architecture enabling cooperative decision-making at heterogeneous granularities—namely, cell-level and cell-pair-level. We introduce two novel learning paradigms: Compositional Deep Reinforcement Learning (CDRL) and Compositional Predictive Decision Modeling (CPDM), which jointly enhance training efficiency and scalability while guaranteeing operational safety. Experimental results demonstrate that the proposed approach significantly reduces handover failure rates, improves network throughput, and decreases end-to-end latency. Moreover, it achieves faster convergence, higher sample efficiency, and superior training stability and deployment adaptability in large-scale network scenarios.

Technology Category

• Search and Optimization: Learning to Search
• Multiagent Systems: Multiagent Learning
• Planning, Routing, and Scheduling: Planning/Scheduling and Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Self-organizing networks face challenges from complex parameter interdependencies and conflicting objectives. This study introduces two compositional learning approaches-Compositional Deep Reinforcement Learning (CDRL) and Compositional Predictive Decision-Making (CPDM)-and evaluates their performance under training time and safety constraints in multi-agent systems. We propose a modular, two-tier framework with cell-level and cell-pair-level agents to manage heterogeneous agent granularities while reducing model complexity. Numerical simulations reveal a significant reduction in handover failures, along with improved throughput and latency, outperforming conventional multi-agent deep reinforcement learning approaches. The approach also demonstrates superior scalability, faster convergence, higher sample efficiency, and safer training in large-scale self-organizing networks.

Problem

• Address complex parameter interdependencies in self-organizing networks
• Manage conflicting objectives in multi-agent systems
• Reduce model complexity for heterogeneous agent granularities

Innovation

• Compositional Deep Reinforcement Learning approach
• Modular two-tier framework for agents
• Reduces handover failures and improves throughput