Network Protocol Greybox Fuzzing Based on Semantic Fragments

Network intelligent devices communicate with each other effectively and reliably based on protocols, but attackers can also exploit network protocols for remote network attacks, making protocol security a critical concern. Fuzzing, due to its convenience and effectiveness, is widely used in security testing. However, mainstream fuzzing tools for network protocols often heavily depend on documentation, are insensitive to protocol fields, and generate low-quality data. Therefore, this paper proposes a fuzzing method based on protocol semantic fragments. It segments protocol messages into fields based on device response information and mutates messages according to the semantics of protocol fields. Experiments were conducted using the existing protocol fuzzing benchmark dataset, Profuzzbench. The results show that compared to tools like AFLNET, the proposed method achieves significant improvements, demonstrating its effectiveness.