Online Device Fingerprinting

Device fingerprinting is powerful for network security assess- ment and intrusion detection because its goal is to get the precise name and version of a remote device. This paper is based on device repre- sentations proposed recently: the syntactic structure of a message and the behavior of a device. A comparison function is associated to both of them in order to be applied with recent classification techniques which leverage supervised learning. The approaches are evaluated with the SIP protocol and the evaluation considers the correctness of the identification and also computational complexity for being applied online. Conclusion exhibits the advantages and drawbacks of each method for choosing the more suitable method according to the network environment.