FUZZY MODELLING OF A NETWORK DENIAL OF SERVICE (DOS) ATTACK PHENOMENON.

摘要

This paper presents the fuzzy modeling of a network Denial of Service (DoS) attack characteristics. Conventional methods for achieving same purpose were showcased. Also, the need for a fuzzy logic approach as well as an improved mechanism for generating the fuzzy inference rules were outlined. Further, the paper discusses the basic concept of fuzzy logic, fuzzy systems and reasons for their adoption in modern control operations. DoS attack detection, using some basic traffic characteristics: bitrates, entropy and Hurst experiment as it?s inputs, was x-rayed. Here, an attack trace file collected at the edge router of the Computer Science Department, University of California Los Angeles was used to develop the fuzzy inference model. The data set was partitioned into a training dataset and a testing dataset. The fuzzy concept learning system (FCLS) algorithm was used for constructing the fuzzy decision tree, using the trace files. Inference rules were then generated from the constructed decision tree. The simulation and evaluation of the fuzzy model was performed with the testing dataset.