Finding Influential Nodes in Ethereum Using Machine Learning

Muhammad Fawad; Khalid; Muhammad Asfandyar; Zia Ullah; Abid Ullah; Muhammad Naeem Ullah

Authors

Muhammad Fawad Lecturer, Department: CS & IT, Sarhad University of ScienceandTechnology, Peshawar, Pakistan
Khalid Graduate Scholar, Department: CS & IT, Sarhad Universityof Science and technology, Peshawar, Pakistan
Muhammad Asfandyar Lecturer, Department: CS & IT, Sarhad University of ScienceandTechnology, Peshawar, Pakistan
Zia Ullah Graduate Scholar, Department: CS & IT, Sarhad Universityof Science and technology, Peshawar, Pakistan
Abid Ullah Graduate Scholar, Department, Department of ComputingAbasynUniversity, Peshawar
Muhammad Naeem Ullah MS Scholar, Department: Department of Computing AbasynUniversity, Peshawar

Abstract

Ethereum blockchain is the market leading platformfordecentralized applications and smart contracts that have poweredthe new age of financial ecosystem. In order to improve securityand performance, identify influential nodes, and understand network dynamics on Ethereum it is critical to identify influential nodes in Ethereum. In this thesis work, we explore machinelearning techniques for discovery of these nodes usinggraphbased algorithms, centrality measures and clustering methods. Itstudies the impact of a node in terms of frequency of usage, connectivity and computational power for a node. Finally, wecompare performance of our proposed methodology combiningsupervised learning and graph neural networks to their traditional counterparts and demonstrate our approach outperforms existingmethods. We demonstrate that highly influential nodes engageinunique patterns of behavior, which are detectableandcategorizable. We contribute to understanding of the networkstructure of Ethereum, along with a scalable approachtomonitoring and optimising blockchain ecosystems. Moreover wediscuss the implications for network robustness, frauddetectionand protocol enhancements, and demonstrate the promiseof
machine learning for blockchain analytics.

Keywords: Ethereum, Blockchain, Machine Learning, Influential Nodes, Graph Neural Networks, Decentralized Networks