Abstract
In the last decade, smart contract security issues lead to tremendous losses, which has attracted increasing public attention both in industry and in academia. Researchers have embarked on efforts with logic rules, symbolic analysis, and formal analysis to achieve encouraging results in smart contract vulnerability detection tasks. However, the existing detection tools are far from satisfactory. In this paper, we attempt to utilize the deep learning-based approach, namely bidirectional long-short term memory with attention mechanism (BLSTM-ATT), aiming to precisely detect reentrancy bugs. Furthermore, we propose contract snippet representations for smart contracts, which contributes to capturing essential semantic information and control flow dependencies. Our extensive experimental studies on over 42,000 real-world smart contracts show that our proposed model and contract snippet representations significantly outperform state-of-the-art methods. In addition, this work proves that it is practical to apply deep learning-based technology on smart contract vulnerability detection, which is able to promote future research towards this area.
Highlights
Software or program carrying security flaws can potentially allow attackers to compromise systems and applications
We show that our deep learning-based approach outperforms state-of-the-art smart contract vulnerability detection tools
In order to highlight the importance of some output results for vulnerability detection, we introduce the attention mechanism [33] making the final sequential model obtained (i.e., BLSTM-ATT), which authentically improves the effect of experiments
Summary
Software or program carrying security flaws can potentially allow attackers to compromise systems and applications. P. Qian et al.: Towards Automated Reentrancy Detection for Smart Contracts Based on Sequential Models especially with the expansion transactions reliance on smart contracts in almost blockchain platforms. More and more practices are based on deep learning, there is still a lack of effort exploring to use a deep learning-based method for smart contract vulnerability detection due to the novelty and complexity of smart contracts. Towards this target, we introduce sequential models for reentrancy detection at a contract source code level. We show that our deep learning-based approach (i.e., sequential models) outperforms state-of-the-art smart contract vulnerability detection tools.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have