Large Language Model Driven Automated Network Protocol Testing
Yunze Wei, Kaiwen Chi, Shibo Du, Xiaohui Xie, Ziyu Geng, Yuwei Han, Zhen Li, Zhanyou Li, Yong Cui · 2025
Traditional network protocol testing methods face significant challenges in adapting to rapid protocol evolution. The challenges stem primarily from protocol specification analysis and customized code development for testing. To address this, we propose NeTestLLM, a Large Language Model (LLM)-powered framework that automates protocol testing through two key components: (1) a hybrid test case generator that extracts protocol specifications and produces high-coverage test cases, and (2) a retrieval-feedback-enhanced engine that translates natural language descriptions into executable code. Preliminary evaluations demonstrate that NeTestLLM achieves 94.1% coverage on protocol specification understanding. A case study with commercial network equipment validates the practical effectiveness of our approach. Our work presents the first LLM-powered framework for automated network protocol testing to keep pace with the rapid evolution of network protocols and standards.