Darragh O'Brien

Machine Learning for Automatic Defence Against Distributed Denial of Service Attacks

Distributed denial of service attacks pose a serious threat to many businesses which rely on constant availability of their network services. Companies like Google, Yahoo and Amazon are completely reliant on the Internet for their business. It is very hard to defend against these attacks because of the many different ways in which hackers may strike. Distinguishing between legitimate and malicious traffic is a complex task. Setting up filtering by hand is often impossible due to the large number of hosts involved in the attack. The goal of this paper is to explore the effectiveness of machine learning techniques in developing automatic defences against DDoS attacks. As a first step, a data collection and traffic filtering framework is developed. This foundation is then used to explore the potential of artificial neural networks in the defence against DDoS attacks.

Teaching Operating Systems Concepts with SystemTap

The study of operating systems is a fundamental component of all undergraduate computer science degree programmes. Making operating system concepts concrete typically entails large programming projects. Such projects traditionally involve enhancing an existing module in a real-world operating system or extending a pedagogical operating system. The latter programming projects represent the gold standard in the teaching of operating systems and their value is undoubted. However, there is room in introductory operating systems courses for supplementary approaches and tools that support the demonstration of operating system concepts in the context of a live, real-world operating system. This paper describes an approach where the Linux monitoring tool SystemTap is used to capture kernel-level events in order to illustrate, with concrete examples, operating system concepts in the areas of scheduling, file system implementation and memory management. For instructors and students (where often for the latter seeing is believing) this approach offers an additional simple and valuable resource for solidifying understanding of concepts that might otherwise remain purely theoretical.

Spectral Dynamics as a Source of Discontinuity in Concatenative Speech Synthesis

The quality of concatenative speech synthesis depends on the cost function employed for unit selection. Effective cost functions for spectral continuity have proven difficult to define and standard measures do not accurately reflect human perception of spectral discontinuity in concatenated speech. Previous studies on spectral join costs have focused predominantly on static spectral measures extracted from the unit boundary. In this paper spectral dynamic behaviour is investigated as a source of discontinuity in concatenated speech. A number of measures representing spectral dynamics are tested for the task of detecting discontinuities. The spectral dynamic measures tested contain information correlating with human perception of discontinuities, suggesting that spectral dynamics are a source of discontinuity in concatenated speech. A strategy to effectively combine dynamic and static measures is proposed using principal component analysis (PCA).