Danger! Computer Security meets Operational Research

Artificial Immune Systems (AIS) are a collection of algorithms inspired by aspects of the human immune system. The arguably most obvious application of AIS was to detect intrusions in computer networks. If we can fight viruses with our immune systems, then surely we can fight computer viruses with a computer immune system? Early AIS approaches used a technique called Negative Selection, which did not provide a sufficient level of protection against intrusions. Problems were found with scalability and large numbers of false positives (false alarms) were generated. In 2003, Dr Uwe Aickelin and his colleagues proposed that the reason for the poor performance of the AIS is that Negative Selection is based on outdated concepts in immunology. It was proposed that the incorporation of the Danger Theory, a modern principle of immunology, could improve the performance of AIS when applied to intrusion detection. The result of the proposal is the Danger Project (EPSRC Adventure Fund) – an interdisciplinary collaboration between computer security experts, computer scientists, and 'wetlab' immunologists. Recently, Uwe has been awarded an Advanced ESPRC Fellowship to investigate how OR models such as set covering problems can help direct future research. The work performed to achieve these goals is the focus of this part of the seminar. Instead of trying to improve on existing heuristics, Julie Greensmith spent one year trying to understand the intricate methods the human immune system uses to detect invaders when the body is under attack. A novel idea in immunology is that the immune system releases danger signals in response to damage caused by infection. One cell in particular is involved in the collection and processing of danger signals. These cells are called Dendritic Cells (DCs). DCs are a major control unit cell of the human immune system, but were previously ignored by the AIS