Vulnerability Detection in Blockchain Smart Contracts Using Cloud-Based Automated Machine Learning

As software code, blockchain smart contracts often have vulnerabilities that are susceptible to cyberattacks. The use of smart contracts to store and transfer billions of dollars in value, coupled with the increasing number of known vulnerabilities, makes them an attractive target for hackers. Rapidly and accurately detecting vulnerabilities in smart contracts would assist developers with addressing the flaws and prevent vigilant parties from participating in transactions associated with vulnerable smart contracts.Prior research focuses on the use of static and dynamic code analysis tools, and traditional machine learning built models to detect smart contract vulnerabilities. These approaches to vulnerability detection, however, present several challenges and limitations. This praxis examines the use of cloud-based automated machine learning to build smart contract vulnerability detection models using native smart contract source code, without an intermediary representation, to detect insecure arithmetic, block values as a proxy for time, reentrancy, locked Ether, and transaction order dependency flaws. The built binary classification-based vulnerability detection models were then evaluated for effectiveness through statistical performance metrics such as accuracy and precision. The research concludes that cloud-based automated machine learning and native smart contract source code can be used to build models that detect several types of vulnerabilities with a high degree of accuracy and precision. The research results also demonstrated that cloud-based automated machine learning can be used to build effective vulnerability detection models in a matter of hours, and the built models can evaluate a smart contract for vulnerabilities within a fraction of a second - a substantial improvement over traditional static code analysis and dynamic code analysis tools, and comparable to traditional machine learning based detection models.

Author

• Appulingam, Arthiraiyan

Language

• en

Keyword

• Blockchain Smart Contracts
• Source Code Vulnerabilities
• Ethereum
• Vulnerability Detection
• Cybersecurity
• Machine Learning

Date created

• 2023

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Cybersecurity Analytics

Degree

• D.Engr.

Advisor

• Sahlin, John

Committee Member(s)

• Sahlin, John
• Fossaceca, John
• Sarkani, Shahryar

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/jh343t31x