Obfuscation-Resilient Purpose Identification for WebAssembly Functions
ACM Transactions on Internet Technology · 28.02.2027 · 27 (1) · 1-21
DOI: 10.1145/3837761 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM Transactions on Internet Technology, online erschienen am 14.11.2026. Verfasst von Yang Zhang und 3 weiteren Autor:innen. Die Arbeit nennt 62 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Zhejiang Province.
Abstract (Originalsprache)
WebAssembly (Wasm) has emerged as a widely adopted binary instruction format that enables applications to run across diverse platforms with near-native performance. However, its compact and low-level structure makes it difficult to interpret. It is essential to identify the purpose of functions within WebAssembly binaries for both functionality understanding and security analysis. Developers often need to understand the functionality of obfuscated third-party WebAssembly modules, and attackers may employ obfuscation to conceal malicious applications. Identifying the intent of WebAssembly functions becomes more challenging when the binaries are obfuscated, as obfuscation techniques transform the code structure and logic. To address this challenge, we propose WASMOP, an obfuscation-resilient WebAssembly function classification framework, where functions are classified into semantic label categories serving as abstractions for function purpose. WASMOP utilizes Instruction Transition Matrices (ITMs) to extract features that are robust to obfuscation, which captures the contextual relationships between instructions. By analyzing WebAssembly files before and after obfuscation, it extract those semantic-related instruction transition patterns as features. We construct a dataset of 27,751 WebAssembly files sourced from real-world web deployments and open-source repositories, and apply multiple binary-level obfuscation techniques, categorizing the functions into 685 semantic labels. Our experiments show that WASMOP outperforms the state-of-the-art method by 15.4% on original binaries and by up to 29.8% under obfuscation, illustrating that our approach is effective at identifying function purposes in obfuscated WebAssembly binaries.
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Zhejiang Province
Zitieren
APA
Zhang, Y., Jin, L., Gao, Y., & Dong, W. (2027). Obfuscation-Resilient Purpose Identification for WebAssembly Functions. ACM Transactions on Internet Technology, 27(1), 1-21. https://doi.org/10.1145/3837761
Vancouver
Zhang Y, Jin L, Gao Y, Dong W. Obfuscation-Resilient Purpose Identification for WebAssembly Functions. ACM Transactions on Internet Technology. 2027;27(1):1-21. doi:10.1145/3837761
BibTeX
@article{Zhang2027,
author = {Zhang, Yang and Jin, Lewei and Gao, Yi and Dong, Wei},
title = {Obfuscation-Resilient Purpose Identification for WebAssembly Functions},
journal = {ACM Transactions on Internet Technology},
year = {2027},
volume = {27},
number = {1},
pages = {1--21},
doi = {10.1145/3837761}
}
EUVIMED