Deliverable D2.1 Closed loop fuzzing algorithms

Laurent Andrey (1) , Humberto Abdelnur (1) , Jorge Lucangeli Obes (1) , Olivier Festor (1) , Radu State (2)

1 MADYNES - Management of dynamic networks and services
2 Uni.lu - Université du Luxembourg

The techniques and tools described in this report propose a way to measure the impact of a fuzzer on a running system. The work focuses on protocols fuzzing. So tested systems are protocol entities and inputs protocol messages. Therefore the elementary measure assesses the impact of a crafted protocol message injected into the running system under test. From this point several interesting uses can be derived: • The overall impact, the coverage of a sequence generated by a the fuzzer can be calculated. • Then two fuzzers can be compared. • One or several sequences can be optimized: only messages introducing the best coverage can be selected to limit the cost (duration) to apply the test. • The process, the strategy that generates sequences can itself take advantage of this impact measurement to directly produce new optimized sequences. The Session Initiation Protocol (SIP) is the target protocol of the study.

Domaines: Réseaux et télécommunications [cs.NI]

Laurent Andrey, Humberto Abdelnur, Jorge Lucangeli Obes, Olivier Festor, Radu State. Deliverable D2.1 Closed loop fuzzing algorithms. 2010. ⟨inria-00546964⟩