Research Article
Identifying DApps and User Behaviors on Ethereum via Encrypted Traffic
@INPROCEEDINGS{10.1007/978-3-030-63095-9_4, author={Yu Wang and Zhenzhen Li and Gaopeng Gou and Gang Xiong and Chencheng Wang and Zhen Li}, title={Identifying DApps and User Behaviors on Ethereum via Encrypted Traffic}, proceedings={Security and Privacy in Communication Networks. 16th EAI International Conference, SecureComm 2020, Washington, DC, USA, October 21-23, 2020, Proceedings, Part II}, proceedings_a={SECURECOMM PART 2}, year={2020}, month={12}, keywords={DApps and user behaviors Encrypted traffic classification Features extraction Traffic analysis Machine learning}, doi={10.1007/978-3-030-63095-9_4} }- Yu Wang
Zhenzhen Li
Gaopeng Gou
Gang Xiong
Chencheng Wang
Zhen Li
Year: 2020
Identifying DApps and User Behaviors on Ethereum via Encrypted Traffic
SECURECOMM PART 2
Springer
DOI: 10.1007/978-3-030-63095-9_4
Abstract
With the surge in popularity of blockchain, more and more Decentralized Applications (DApps) are deployed on blockchain platforms. DApps bring convenience to people, but cause security and efficiency problems. In this paper, we focus on the security and efficiency problems of DApps on Ethereum. Our research is divided into three application scenarios. In DApps classification, we analyze characteristics of DApps and extract efficient features to recognize 11 representative DApps. In DApps user behaviors classification, we propose behavior-sensitive features and improved time features to recognize 88 DApps user behaviors, which would help to identify malicious behaviors in encrypted traffic. In general user behavior classification, different categories of features are proposed to recognize 15 general user behaviors which represent the performance of DApps. DApps developers can obtain valuable data to improve the quality of service through analyzing the classification results. Experimental results in the three application scenarios achieve excellent performance (99.5% accuracy for DApps classification, 95.65% accuracy for DApps user behaviors classification, 98.58% accuracy for general user behaviors classification) and outperform the state-of-the-art methods.
