Intrusion Aware System-on-a-Chip Design with Uncertainty Classification

In this paper, we have proposed a System-on-a-Chip (SoC) architectural design to avoid potential intrusion or attacks from external devices. Either using misuse detection or anomaly detection techniques to design intrusion detection systems, a large amount of traffic data is needed to be collected in advance for analysis. However, it is not feasible in the limited resources available in SoC systems. We propose to incorporate fuzzy clustering technique along with Dempster-Shafer theory into our intrusion detection design to solve uncertainty problems caused by ambiguous and limited information. Also, the k-NN technique is applied to speed up the detection process. We compare the results of our proposed approach with those of k-NN classifier, fuzzy k-NN classifier and evidence-theoretic k-NN classifier. It indicates that our approach is able to achieve higher detection rates than those from the other three classifiers, thus is more useful in the implementation of intrusion aware mechanism in SoC design.