EPF: An Evolutionary, Protocol-Aware, and Coverage-Guided Network Fuzzing Framework

This paper presents EPF, a network protocol fuzzing framework combining evolutionary algorithms, protocol state modeling, and coverage-guided feedback. The tool integrates Scapy-based protocol definitions with state transition graphs to systematically explore protocol implementations. By employing a simulated annealing-driven mutation strategy and AFL++-style instrumentation, EPF achieves deeper protocol state penetration compared to traditional fuzzers, as demonstrated through case studies on IEC 60870-5-104 implementations in critical infrastructure systems. The framework’s modular architecture enables protocol-specific state machine definitions and mutation operators, while its coverage metrics guide the exploration of edge cases in protocol parsers. Experimental results show EPF’s effectiveness in identifying security vulnerabilities through automated state-aware fuzzing campaigns.