Back to search

Aristotle University Of Thessaloniki (AUTH)

Αναγνώριση επίθεσης σε δικτυακές υπηρεσίες με χρήση του συστήματος αρχείων και τεχνικών μηχανικής μάθησης

Abstract

dc:description

Current Society is highly depended on the Internet. A vast amount of personal information, like user passwords, credit card numbers, e-mails, etc. travels around the world via the Internet backbone in a daily basis through the use of smart phones, tablets, and all different kinds of devices and sensors forming the Internet of Things (IoT). However, for those who are lurking in the cyber-space underground, as the number of on-line devices increases and the use of Internet services grows, so does the attack surface. In the recent years, apart from the cyber-threat protection provided traditionally by antivirus systems and firewalls, a new, second-line of defense has been developed, namely Intrusion Detection Systems (IDSs). The purpose of an IDS is to discover when an Information System is under attack and inform the user or even react when possible. This thesis is focused on how the Operating System's file system can be used in order to aid the intrusion detection process. For this purpose, a feature space is proposed on which raw file system information is projected. Based on this type of information, FI²DS is presented. FI²DS is an anomaly-based intrusion detection system that analyses file system information generated by web daemons to recognise attacks. FI²DS accomplishes high detection rates with low false positive rates, proving that our proposed feature space is highly informative and hence helpful for intrusion detection. Next, we propose a method for analysing sequences of alerts originating from the file system and generated by systems like FI²DS, and through the use of grouping, clustering and anomaly detection algorithms employed in these sequences, a further reduce in FI²DS' false positive rate is achieved. Finally, we propose a new architecture for designing IDSs which is based on behaviours. Behaviours are higher-level information units with respect to the raw information provided by IDS sensors. Based on the idea of behaviours, a prototype IDS is presented (BIDS) that outperforms FI²DS in the context of intrusion detection, indicating that as far as the file system is concerned, higher-level information is more useful than to lower-level information regarding intrusion detection.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2015

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Mamalakis, George
  • Μαμαλάκης, Γεώργιος

Subjects

dc:subject × 14

Rights

dc:rights
Statement dc:rights
  • BY
Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/37846

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Mamalakis, George; Μαμαλάκης, Γεώργιος. Αναγνώριση επίθεσης σε δικτυακές υπηρεσίες με χρήση του συστήματος αρχείων και τεχνικών μηχανικής μάθησης. Aristotle University Of Thessaloniki (AUTH), 2015. http://hdl.handle.net/10442/hedi/37846