Data-driven network layer security detection model and simulation for the Internet of Things based on an artificial immune system

Abstract

Traditional network security detection models are trained offline using attack samples of known types. Although such models have high detection rates for known attack types, they cannot identify new attack types in the network layer. At present, these detection systems have the disadvantages of slow system construction and a high cost of model updating. Facing the increasing expansion of networks and endless attacks, these detection systems lack self-adaptability and expansibility, so it is difficult to detect complex and changeable attack events in networks. In this paper, the integrated use of immunology theory, complex adaptive system theory, and computational experiment technology is proposed to develop an Internet network layer security detection model based on an artificial immune system as an improvement over existing models of Internet network layer security. On the basis of testing knowledge, when a new type of attack is encountered, the online detection and learning process enables the dynamic extension of the network security detection model. Experimental calculations and an example analysis are presented to verify the scientific validity and feasibility of the model.