Active inference of protocol state machines from incomplete message domains

• Introduces a novel active inference method for protocol state machines using the minimally adequate teacher (MAT) framework, addressing incomplete message domains. • Achieves a comprehensive protocol state machine by broadening the input space via session completion and deterministic mutation techniques. • Optimizes the L+ M algorithm with traffic deduplication, expanded prefix tree acceptor construction, response-based query optimization, and random counterexample generation, reducing execution time by ~40.7%. • Demonstrates significant efficiency gains on RTSP and SMTP protocols, cutting connections by ~28.6% and interactions by ~46.6% compared to AALpy.