On August 27th, 2024, our research group had the honor of presenting the paper titled “Decision Tree Based Inference of Lightning Network Client Implementations” at the 21st International Conference on Modeling Decisions for Artificial Intelligence (MDAI 2024), held in Tokyo, Japan.

This work explores a novel methodology for identifying which specific client implementation a Lightning Network node is using—such as LND, c-lightning (Core Lightning), or Eclair—based solely on observable network behaviors. By leveraging decision tree algorithms, the study demonstrates how client-specific heuristics can be inferred from message timing, routing patterns, and protocol-level fingerprints, all without needing privileged access to the nodes.

The motivation behind this research stems from the increasing importance of understanding software diversity in decentralized networks, where different client implementations may follow subtly different rules or optimizations. Identifying these implementations can support forensic analysis, protocol robustness testing, and interoperability studies—particularly crucial as the Lightning Network continues to evolve as a second-layer solution for Bitcoin scalability.

The conference provided a valuable platform to discuss our findings with international experts in artificial intelligence, data modeling, and cybersecurity. Feedback from the MDAI community helped validate the importance of behavioral inference in financial and privacy-preserving systems, and opened up new directions for future research on metadata leakage and adversarial modeling in blockchain-based environments.

We are proud to contribute to the growing body of work that blends machine learning with decentralized network analysis, and look forward to further collaborations in this area. The full paper is available in the MDAI 2024 proceedings, and we welcome questions or discussions from interested researchers.