A Novel Immune Detection Approach Enhanced by Attack Graph Based Correlation

Artificial immune systems (AIS) are computational intelligence inspired by the human biological immune system. The AIS four main algorithms are negative selection, clonal selection, immune network, and danger theory. This paper incorporates the AIS approach to develop an agent-based detection method to analyze network traffic. The system works with an attack graph based correlation technique. This technique can improve detection performance by decreasing false alerts. This work was tested for denial of service (DoS), remote to local (R2L), user to root (U2R) and probe attack classes. Results have shown the addition of the correlation technique can aid to the detection performance of AIS detection systems.