Usable Agent Discovery for Decentralized AI Systems

2026-04-24

AI Summary

This study addresses the challenge of efficient and reliable discovery in distributed AI systems subject to dual churn—simultaneous dynamics at both node and agent levels. It presents the first systematic analysis of how this dual churn impacts decentralized agent discovery mechanisms. The authors comparatively evaluate structured overlay networks (exemplified by Kademlia) against gossip-based approaches (Cyclon+Vicinity) across four scenarios: stable, node-only churn, agent-only churn ("cooling"), and mixed churn. Performance is assessed in terms of routing efficiency, robustness, and service readiness. The findings reveal that structured overlays excel in efficiency and robustness under stable or node-dynamic conditions, whereas gossip-based protocols offer significant advantages in service readiness when rapid availability is critical. This work delineates clear applicability boundaries for each overlay type in highly dynamic environments.

Technology Category

• Search and Optimization: Distributed Search
• Multiagent Systems: Other Foundations of Multi Agent Systems
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Web Mining and Content Analysis: Content-based information diffusion

Abstract

Large-scale agentic systems run on distributed infrastructures where many software agents share physical hosts and are discovered via peer-to-peer mechanisms. Discovery must handle node-level churn from failures and host departures and agent-level churn from demand-driven activation, deactivation, and state changes. Their interaction reshapes classic trade-offs between structured and unstructured overlays. We study decentralized agent discovery under this two-level churn, assuming nodes host multiple agents, overlays are structured or gossip-based, and agents switch between warm and cold states. Using Kademlia as a structured and Cyclon+Vicinity as a gossip baseline, we compare stable, node-churn-only, agent-cooling-only, and combined regimes to see when routing efficiency, resilience, and service readiness align or favor different designs. Structured overlays are more robust and efficient in stable and node-churn regimes, while gossip-based overlays remain competitive and can be faster when readiness dominates.

Problem

• decentralized AI systems
• agent discovery
• node churn
• agent churn
• overlay networks

Innovation

• decentralized agent discovery
• two-level churn
• structured overlay
• gossip-based overlay
• service readiness