An Efficient Approach for Classifying Intrusion using Fusion based HMM & Clustering [ READ ]
Chandrima Dutta, Prof. Amit Saxena, Dr. Manish Manoria
Intrusion detection system is a method of identifying unnecessary packets that may be creates some damage in the network; hence various Intrusion detection based methods are implemented to provide security in the network traffic flow. Here in this paper an efficient technique of identifying intrusions is implemented using the concept of hidden markov model and then classification of these intrusions is done. The methodology implemented here is applied on KDDCup 99 dataset where the data to be detected is first group some by using clustering approach so that the similar packets are grouped into one and the dissimilar packets are grouped into another. Now some of the important attributes are selected from the dataset and defined as the states of Hidden Markov Model and the probability is calculated from each of the state to other state and finally these probabilities are fused to find the overall probability of the dataset and hence on the basis of threshold probability packets can be classified as low and medium and high intrusions.
Chandrima Dutta, Prof. Amit Saxena, Dr. Manish Manoria
Intrusion detection system is a method of identifying unnecessary packets that may be creates some damage in the network; hence various Intrusion detection based methods are implemented to provide security in the network traffic flow. Here in this paper an efficient technique of identifying intrusions is implemented using the concept of hidden markov model and then classification of these intrusions is done. The methodology implemented here is applied on KDDCup 99 dataset where the data to be detected is first group some by using clustering approach so that the similar packets are grouped into one and the dissimilar packets are grouped into another. Now some of the important attributes are selected from the dataset and defined as the states of Hidden Markov Model and the probability is calculated from each of the state to other state and finally these probabilities are fused to find the overall probability of the dataset and hence on the basis of threshold probability packets can be classified as low and medium and high intrusions.
