Fig: Automatic Fingerprint Generation

Fingerprinting is a widely used technique among the net- working and security communities for identifying differen t implementations of the same piece of networking software running on a remote host. A fingerprint is essentially a set of queries and a classification function that can be applied on the responses to the queries in order to classify the softwar e into classes. So far, identifying fingerprints remains largely an arduous and manual process. This paper proposes a novel approach for automatic fingerprint generation, that automatically explores a set of candidate queries and ap- plies machine learning techniques to identify the set of val id queries and to learn an adequate classification function. Our results show that such an automatic process can gener- ate accurate fingerprints that classify each piece of softwa re into its proper class and that the search space for query ex- ploration remains largely unexploited, with many new such queries awaiting discovery. With a preliminary exploratio n, we are able to identify new queries not previously used for fingerprinting.