Abstract
Blockchain is becoming one of the main components of the modern world due to its rapid development. Considering its main characteristics as a collaborative distributed database that provides transparency and scalability, blockchain development and application have grown dramatically over the past decade. While the blockchain is a key feature and is used in the industry, it is subject to many risk factors and attacks. Advanced technologies that can successfully detect and mitigate vulnerabilities must be developed for solving these security challenges. This thesis presents the development of a machine learning based approach to discover vulnerabilities in blockchain-based systems. It begins with a comprehensive survey of existing techniques for identifying vulnerabilities within smart contracts, covering both traditional methodologies and modern machine learning approaches. Also, it provides a comparative analysis of key features and limitations of existing mechanisms, emphasizing the need for a more effective solution. Based on these findings, the thesis proposes a machine learning framework employing pruned random forest, neural networks and support vector machines. Additionally, training and testing the model against several smart contract vulnerabilities which are reentrancy, unchecked external calls, timestamp dependency, and integer overflow. The results of the machine learning model demonstrated that Pruned Random Forest is able to deal with complicated patterns which enabled it to perform better than the other two models. The need for better data management including optimization and normalizations methods was highlighted by the fact that the models faced difficulties with unbalanced data and less counted vulnerabilities. The suggested technique demonstrates significant advances in vulnerability detection for blockchain-based applications. By leveraging machine learning techniques, the model effectively identifies critical weaknesses in smart contracts while highlighting the limitations of different models proposed in the past five years. In addition, this thesis provides a solid basis for upcoming improvements that will increase the security and dependability of blockchain technology.
Publication Date
12-2024
Document Type
Thesis
Student Type
Graduate
Degree Name
Computing Security (MS)
Department, Program, or Center
Electrical Engineering
Advisor
Wesam Almobaideen
Advisor/Committee Member
Ali Assi
Advisor/Committee Member
Ioannis Karamitsos
Recommended Citation
Ahli, Alia Mohammad, "An Automated Approach to detect cyber threats with smart contracts in blockchain systems." (2024). Thesis. Rochester Institute of Technology. Accessed from
https://repository.rit.edu/theses/11993
Campus
RIT Dubai
Comments
This thesis has been embargoed. The full-text will be available on or around 4/17/2025.