No abstract

Modern network intrusion detection systems rely on machine learning techniques to detect trac anomalies and thus intruders. However, the ability to learn the network behaviour in real-time comes at a cost: malicious software can interfere with the learning process, and teach the intrusion detection system to accept dangerous trac. This paper presents an intrusion detection system (IDS) that is able to detect common network attacks including but not limited to, denial-of-service, bot nets, intrusions, and network scans. With the help of the proposed example IDS, we show to what extent the training attack (and more sophisticated variants of it) has an impact on machine-learning based detection schemes, and how it can be detected.

Recently, the password-authenticated key exchange protocol J-PAKE of Hao and Ryan (Workshop on Security Protocols 2008) was formally proven secure in the algebraic adversary model by Abdalla et al. (IEEE S&P 2015). In this paper, we propose and examine two variants of J-PAKE -which we call RO-J-PAKE and CRS-J-PAKE -that each makes the use of two less zero-knowledge proofs than the original protocol. We show that they are provably secure following a similar strategy to that of Abdalla et al. We also study their efficiency as compared to J-PAKE's, also taking into account how the groups are chosen. Namely, we treat the cases of subgroups of finite fields and elliptic curves. Our work reveals that, for subgroups of finite fields, CRS-J-PAKE is indeed more efficient than J-PAKE, while RO-J-PAKE is much less efficient. On the other hand, when instantiated with elliptic curves, both RO-J-PAKE and CRS-J-PAKE are more efficient than J-PAKE, with CRS-J-PAKE being the best of the three. We illustrate this experimentally, making use of recent research by Brier et al. (CRYPTO 2010). Regardless of implementation, we note that RO-J-PAKE enjoys a looser security reduction than both J-PAKE and CRS-J-PAKE. CRS-J-PAKE has the tightest security proof, but relies on an additional trust assumption at setup time. We believe our results can be useful to anyone interested in implementing J-PAKE, as perhaps either of these two new protocols may also be options, depending on the deployment context.

Abstract. Dragonfly Dragonfly is a password-authenticated key exchange protocol that was proposed by Harkins [11] in 2008. It is currently a candidate for standardization by the Internet Engineering Task Force, and would greatly benefit from a security proof. In this paper, we prove the security of a very close variant of Dragonfly Dragonfly in the random oracle model. It shows in particular that Dragonfly Dragonfly's main flows -a kind of Diffie-Hellman variation with a password-derived base -are sound. We employ the standard Bellare et al.[2] security model, which incorporates forward secrecy.

No abstract