University of Ioannina
Μαθηματικά μοντέλα ανάλυσης εισβολών σε δίκτυα υπολογιστών με χρήση των χαρακτηριστικών αποδοτικότητας των δικτύων
Abstract
dc:descriptionA computer virus is a malicious, self-propagating piece of code that is able to spread fast in computer networks. Although a growing number of researchers focus their efforts on devising new techniques for detecting and eliminating worms, there seems to be less intense activity towards the development and evaluation of theoretical models able to account of how worms exploit vulnerabilities of computer networks and propagate, accordingly. In our work, we propose and analyze mathematical models for the co-evolution of the populations of virus and antivirus software agents across an infected computer network. We first model a two populations system comprised of worm and antivirus agents, and analyze it theoretically showing that its prediction are in harmony with observations from propagation of real worms. We extend this model, adding the traps population, using a system of three non-linear differential equations that describe their interactions. Then we view the attack propagation/elimination problem in networks from another perspective that avoids the use of non-linear evolution dynamics. This model takes into account the traffic and server characteristics of the network computers. We propose a kind of interaction between virus and antivirus agents that result in a product form steady state distribution of the agent numbers for each network node, much like the product form solution for the distribution of network tasks for Jackson open networks of queues. Afterwards we study DNS worms, these worms generate random strings, as possible network domain names, and then query Domain Name Servers in order to discover the IP address. We present models for capturing the dynamics of the co-evolution of worm agents in the presence of anti-worm agents that move in the network in order to locate and stop worm propagation. We further enhance the model with ‘‘honeypot’’ domain name servers that attempt to lure worm agents to issue queries, introducing only a delay and providing no answer. Finally we propose a general predator-prey model for studying the joint evolution of a prey population that coexists with a number of predator populations. The model allows the definition of different behaviors between individuals from these populations as well as different interaction patterns between each of these populations and the prey population. Moreover, we model the behaviour between individuals of the two predator populations using as simple game, called Hawks and Doves game, according to which one predator population is aggressive (Hawks) and one predator population is submissive (Doves).
Degree
thesis:*- Grantor dc:publisher
- University of Ioannina
- Year dc:date
- 2010
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Kammas, Pantelis
- Καμμάς, Παντελής
Subjects
dc:subject × 12Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/19872