Comparing performance of binary-coded detectors and constraint-based detectors

Artificial immune systems can be used to detect intrusion by classifying network activities as normal or abnormal. High detection rates and low false positive rates are two necessary features of successful AIS. Strong detectors are the basis of creating a successful AIS. Some preliminary experiments showed its promise to encode detectors in the form of data triples. Currently, there are two types of detectors: binary-coded and constraint-based. This paper compares the two types of detectors using simulated network traffic data. The results show that constraint-based detectors perform better than binary-coded detectors.