Worm infection dynamics on a local area network

Abstract

The aim of this thesis is to investigate the spreading behaviour of worms on a Local Area Network that is supervised by a central administrator who is authorized to enforce the updates on individual computers, but leaves their activation to the users. For this purpose, the processes involved in a specific worm attack have been modelled with a Markovian approach, and simulated in the Matlab programming en vironment. Most parameters in the model are set to realistic estimates of the average values of the corresponding variables, while some other parameters are varied in rea sonable ranges to investigate their impact on performance of the computers on the LAN. The performance and its deterioration are evaluated in terms of the number of computers that are infected and have to be repaired, as well as in terms of the average percentage of active computers on the LAN. The proposed model can be used to develop security policies to reduce the average number of infected computers on a LAN. Here, the effects of some parameters related to infection and security on the percentage of infected computers have been investigated.