Multiple negative selection algorithm: Improving detection error rates in IoT intrusion detection systems


The creation of intrusion detection systems for IoT scenarios presents various challenges. One of them being the need for an implementation of unsupervised learning and decision making in the detection system. The algorithm presented in this paper is capable of definitively identifying a large percentage of possible intrusions as true or false without the need of operator input. Our proposal is based on the Negative Selection algorithm and the co-stimulation principles of Immunology. It uses a two-tiered negative selection process to implement a co-stimulation approach aimed at decreasing the number of detection errors without the need of an operator input.


  • Marin Pamukov
  • Vladimir Poulkov


IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS), 2017.



, ,