DyPBP: Dynamic Peer Beneficialness Prediction for Cryptocurrency P2P Networking

TL;DR

This paper tackles the problem that traditional beneficialness scoring in Bitcoin's P2P network relies on block/transaction arrivals and lengthy connection durations, which often fail to converge due to frequent disconnects. It proposes DyPBP, a dynamic, behavior-driven framework that predicts peer beneficialness before new blocks arrive by leveraging real-time networking features and a remembrance mechanism that preserves state across disconnects. Using supervised learning on Mainnet-collected data, it shows that Linear Regression and Random Forest can achieve dramatic improvements in prediction accuracy (lower MAE) when equipped with remembrance, outperforming KNN. The work demonstrates potential for active P2P control and malicious-peer detection, offering practical benefits for secure, efficient block and transaction propagation in cryptocurrency networks.

Abstract

Distributed peer-to-peer (P2P) networking delivers the new blocks and transactions and is critical for the cryptocurrency blockchain system operations. Having poor P2P connectivity reduces the financial rewards from the mining consensus protocol. Previous research defines beneficalness of each Bitcoin peer connection and estimates the beneficialness based on the observations of the blocks and transactions delivery, which are after they are delivered. However, due to the infrequent block arrivals and the sporadic and unstable peer connections, the peers do not stay connected long enough to have the beneficialness score to converge to its expected beneficialness. We design and build Dynamic Peer Beneficialness Prediction (DyPBP) which predicts a peer's beneficialness by using networking behavior observations beyond just the block and transaction arrivals. DyPBP advances the previous research by estimating the beneficialness of a peer connection before it delivers new blocks and transactions. To achieve such goal, DyPBP introduces a new feature for remembrance to address the dynamic connectivity issue, as Bitcoin's peers using distributed networking often disconnect and re-connect. We implement DyPBP on an active Bitcoin node connected to the Mainnet and use machine learning for the beneficialness prediction. Our experimental results validate and evaluate the effectiveness of DyPBP; for example, the error performance improves by 2 to 13 orders of magnitude depending on the machine-learning model selection. DyPBP's use of the remembrance feature also informs our model selection. DyPBP enables the P2P connection's beneficialness estimation from the connection start before a new block arrives.

Figures (6)

• Figure 1: CDF of Connection Durations for Different Peer Groups.
• Figure 2: Our Implementation from Networking from Bitcoin Mainnet to Sensing to Machine-Learning Processing.
• Figure 3: Top-10 Features based on Mutual Information.
• Figure 4: Model Performance Across Machine Learning Algorithms when Fixing $w_B=0.5$ and With vs. Without Remembrance.
• Figure 5: Model Performance of Linear Regression and Random Forest when Varying Weights and if Remembrance is on.