Kolmogorov complexity estimates for detection of viruses in biologically inspired security systems: a comparison with traditional approaches

  • Authors:
  • Sanjay Goel;Stephen F. Bush

  • Affiliations:
  • University at Albany, State University of New York;GE Global Research Center, Niskayuna, New York

  • Venue:
  • Complexity - Special issue: Resilient and adaptive defense of computing networks
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

This article presents results in two mutually complementary areas: distributed immunological information assurance and a new signature-matching technique based on Kolmogorov Complexity. This article introduces a distributed model for security based on biological paradigms of epidemiology and immunology. In this model each node in the network has an immune system that identifies and destroys pathogens in the incoming network traffic as well as files resident on the node. The network nodes present a collective defense to the pathogens by working symbiotically and sharing pathogen information with each other. Each node compiles a list of pathogens that are perceived as threats by using information provided from all the nodes in the network. The signatures for these pathogens are incorporated into the detector population of the immune systems to increase the probability of detection. Critical to the success of this system is the detection scheme, which should not only be accurate but also efficient. Three separate schemes for detecting pathogens are examined, namely, contiguous string matching, Hamming distance, and Kolmogorov Complexity. This work provides a model of the system and examines the efficiency of different detection schemes. A simulation model is built to study the sensitivity of model parameters, such as signature length, sampling rate, and network topology, to detection of pathogens.