Computer viruses Computer viruses are not merely an irritating and de-structive feature of personal computing, they also mimic the behaviour of biological pests. Computer viruses hi-jack the facilities provided by the host computer, lie dormant, cross-infect other hosts, and when activated cause various forms of damage, from obvious destruc-tion of host data to more subtle changes that are much harder to detect. The predominantly human medical metaphors em-ployed when discussing computer viruses are mislead-ing because of ethical connotations. The word \virus itself is a rich metaphor, and terms like \infection, \viral attack, \disinfectant, \viral damage are fre-quently used of computer viruses. While these terms lack any precise meaning in relation to computers, we know roughly what they mean. Furthermore the debate about the computer virus threat is predominantly expe-riential: awareness of viruses is biased towards frequent infections caused by successful viruses; debate centres around newly identi¯ed virus cases. In order to begin to tackle viruses e®ectively we should de¯ne the essential, general features exhibited by viruses without appealing to metaphor or limited experience to ¯ll in the gaps in our understanding. Computer viruses seem to be a simple phenomenon (hackers manage to write them in quantity, and there are virus construction kits that anyone can use to de-velop sophisticated viruses) yet they do pose a real haz-ard. The science ¯ction anticipation of viruses was thought to be obvious [7]. It is of interest, then, to explore the fundamental limits of detection and pre-vention and other issues. On the face of it, one would look to standard models of computation, such as Tur-ing Machines, as a starting point for this investigation. However a closer analysis requires concepts not usu-ally considered in the standard models of computation. We agree with Wegner [48] that Turing equivalent for-malisms are not su±ciently expressive for systems that interact. A virus has to enter a system to infect it, and this is such an interaction. And as we shall ar-gue, the biological metaphor is inadequate too (though it can still inspire computational ideas, both good and bad). The computational process undertaken by a virus does not terminate; moreover, when the (possibly) Tur-ing process of an infected program terminates, the virus has already infected another program that will subse-quently execute | a virus that terminated would not be infective. Viruses should not be dismissed as a trivial problem for users of computers that do not run secure operat-ing systems. Together with Trojan Horses, they are a major menace for any sort of system, and are a particu-lar concern for computer systems connected to networks As well as being a practical problem, this paper will show that viruses pose theoretical problems also. Indeed, this paper, by laying out some of these prob-lems, begs many questions that raise many further re-search questions. Some of these research questions will be pointed out explicitly throughout the paper. In summary, this paper highlights the limitations of applying the conventional computational models to computer virus phenomena, and it years a new frame-work. We motivate our paper by de¯ning and nam-ing the phenomena under discussion. We then exam-ine a variety of related phenomena, such as: compiler viruses, Cohens proof about virus detection, the possi-ble structures of virally resistant systems, and, ¯nally, Kochs Postulates and other biological issues. The rea-son why people write viruses is beyond the scope of the present paper;
Posted on: Sun, 27 Jul 2014 08:40:31 +0000
Trending Topics
Recently Viewed Topics
© 2015