An evaluation of an artificial immune system based approach to anomaly detection in computer programs

Abstract

The growing sophistication of cyber threats and the limitations of traditional intrusion detection systems (IDS) have led researchers to explore biologically inspired models. One promising approach involves the application of artificial immune systems (AIS), which mimic the self/nonself discrimination mechanism of biological immune systems. In this paper, we propose an IDS based on the negative selection algorithm (NSA), enhanced by a novel modification involving intercellular receptors (ICRs). This dual-receptor architecture improves detection accuracy by targeting both standard and intercellular anomalies in program code. We present a mathematical model of the system, describe its implementation, and evaluate its performance across three key metrics: detection rate, memory efficiency, and processing speed. Experimental results demonstrate that the modified NSA with ICRs matches or outperforms existing methods, achieving an average detection accuracy improvement of 8.1%.