Sequential modeling for obfuscated network attack action sequences

This work presents a preliminary study of modeling obfuscation techniques for network attack action sequences. The aim is to develop a systematic analytical methodology that shows how intentional obfuscation or unintentional noise may affect the modeling and classification of attack sequences. Discussions on different types of obfuscation techniques will be given, followed by a formal formulation to represent the effect of obfuscation within attack sequences. Preliminary simulation results are presented to demonstrate the impact of attack sequences with and without obfuscated attack actions.