The existing network security infrastructure is not ready for future protocols such as Multipath TCP (MPTCP). The outcome is that middleboxes are configured to block such protocols. This paper studies the security risk that arises if future protocols are used over unaware infrastructures. In particular, the practicality and severity of cross-path fragmentation attacks utilizing MPTCP against the signature-matching capability of the Snort intrusion detection system (IDS) is investigated. Results reveal that the attack is realistic and opens the possibility to evade any signature-based IDS. To mitigate the attack, a solution is also proposed in the form of the MPTCP Linker tool. The work outlines the importance of MPTCP support in future network security middleboxes.

Multipath TCP (MPTCP) is a proposed extension to TCP that enables a number of performance advantages that have not been offered before. While the protocol specification is close to being finalized, there still remain some unaddressed challenges regarding the deployment and security implications of the protocol. This work attempts to tackle some of these concerns by proposing and implementing MPTCP aware security services and deploying them inside a proof of concept MPTCP proxy. The aim is to enable hosts, even those without native MPTCP support, to securely benefit from the MPTCP performance advantages. Our evaluations show that the security services that are implemented enable proper intrusion detection and prevention to thwart potential attacks as well as threshold rules to prevent denial of service (DoS) attacks.

Levenshtein distance is well known for its use in comparing two strings for similarity. However, the set of considered edit operations used when comparing can be reduced in a number of situations. In such cases, the application of the generic Levenshtein distance can result in degraded detection and computational performance. Other metrics in the literature enable limiting the considered edit operations to a smaller subset. However, the possibility where a difference can only result from deleted bytes is not yet explored. To this end, we propose an insert-only variation of the Levenshtein distance to enable comparison of two strings for the case in which differences occur only because of missing bytes. The proposed distance metric is named slice distance and is formally presented and its computational complexity is discussed. We also provide a discussion of the potential security applications of the slice distance.

Traditional security mechanisms such as signature basedintrusion detection systems (IDSs) attempt to find a perfect match of aset of signatures in network traffic. Such IDSs depend on the availabilityof a complete application data stream. With emerging protocols such asMultipath TCP (MPTCP), this precondition cannot be ensured, result-ing in false negatives and IDS evasion. On the other hand, if approximatesignature matching is used instead in an IDS, a potentially high numberof false positives make the detection impractical. In this paper, we showthat, by using a specially tailored partial signature matcher and knowl-edge about MPTCP semantics, the Snort3 IDS can be empowered withpartial signature detection. Additionally, we uncover the type of Snort3rules suitable for the task of partial matching. Experimental results withthese rules show a low false positive rate for benign traffic and highdetection coverage for attack traffic.

Encryption on the Internet is as pervasive as ever. This hasprotected communications and enhanced the privacy of users. Unfortu-nately, at the same time malware is also increasingly using encryptionto hide its operation. The detection of such encrypted malware is cru-cial, but the traditional detection solutions assume access to payloaddata. To overcome this limitation, such solutions employ traffic decryp-tion strategies that have severe drawbacks. This paper studies the usageof encryption for malicious and benign purposes using large datasets andproposes a machine learning based solution to detect malware using con-nection and TLS metadata without any decryption. The classification isshown to be highly accurate with high precision and recall rates by usinga small number of features. Furthermore, we consider the deployment as-pects of the solution and discuss different strategies to reduce the falsepositive rate.

Signature-based intrusion detection systems (IDSs) are commonly utilized in enterprise networks to detect and possibly block a wide variety of attacks. Their application in industrial control systems (ICSs) is also growing rapidly as modem ICSs increasingly use open standard protocols instead of proprietary. Due to an ever changing threat landscape, the rulesets used by these IDSs have grown large and there is no way to verify their precision or accuracy. Such broad and non-optimized rulesets lead to false positives and an unnecessary burden on the IDS, resulting in possible degradation of the security. This work proposes a methodology consisting of a set of tools to help optimize the IDS rulesets and make rule management easier. The work also provides attack traffic data that is expected to benefit the task of IDS assessment.