5
$\begingroup$

I'm a physics graduate who ended up doing infosec so most of the statistics I ever learned is useful for thermodynamics. I'm currently trying to think of a model for working out how many of a population of computers are infected with viruses, though I assume the maths works out the same way for real-world diseases so references in or answers relevant to that field would be welcome too.

Here's what I've come up with so far:

  • assume I know the total population of computers, N.
  • I know the fraction D of computers that have virus-detection software (i.e. the amount of the population that is being screened)
  • I know the fraction I of computers that have detection software that has reported an infection
  • I don't know, but can find out or estimate, the probability of Type I and II errors in the detection software.
  • I don't (yet) care about the time evolution of the population.

So where do I go from here? Would you model infection as a binomial distribution with probability like (I given D), or as a Poisson? Or is the distribution different?

$\endgroup$
5
$\begingroup$

Computer virus propagation is structurally similar to infectious diseases propagation (vaccinations = anti-virus software, virus via email = getting a virus from someone etc).

Use the following links http://en.wikipedia.org/wiki/Mathematical_modelling_in_epidemiology and http://en.wikipedia.org/wiki/Compartmental_models_in_epidemiology as jumping points for your model.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy