Analyzing audit trails in a distributed and hybrid intrusion detection platform

• Autor(a) principal: Alves, Pedro Miguel de Freitas
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/18328
• topic: Intrusion Detection Systems (IDS); Distributed and hybrid IDS; Analysis of audit-trails; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Efforts have been made over the last decades in order to design and perfect Intrusion Detection Systems (IDS). In addition to the widespread use of Intrusion Prevention Systems (IPS) as perimeter defense devices in systems and networks, various IDS solutions are used together as elements of holistic approaches to cyber security incident detection and prevention, including Network-Intrusion Detection Systems (NIDS) and Host-Intrusion Detection Systems (HIDS). Nevertheless, specific IDS and IPS technology face several effectiveness challenges to respond to the increasing scale and complexity of information systems and sophistication of attacks. The use of isolated IDS components, focused on one-dimensional approaches, strongly limits a common analysis based on evidence correlation. Today, most organizations’ cyber-security operations centers still rely on conventional SIEM (Security Information and Event Management) technology. However, SIEM platforms also have significant drawbacks in dealing with heterogeneous and specialized security event-sources, lacking the support for flexible and uniform multi-level analysis of security audit-trails involving distributed and heterogeneous systems. In this thesis, we propose an auditing solution that leverages on different intrusion detection components and synergistically combines them in a Distributed and Hybrid IDS (DHIDS) platform, taking advantage of their benefits while overcoming the effectiveness drawbacks of each one. In this approach, security events are detected by multiple probes forming a pervasive, heterogeneous and distributed monitoring environment spread over the network, integrating NIDS, HIDS and specialized Honeypot probing systems. Events from those heterogeneous sources are converted to a canonical representation format, and then conveyed through a Publish-Subscribe middleware to a dedicated logging and auditing system, built on top of an elastic and scalable document-oriented storage system. The aggregated events can then be queried and matched against suspicious attack signature patterns, by means of a proposed declarative query-language that provides event-correlation semantics.