A Novel Security Analysis for Virtualized Infrastructure using Fuzzy Classification Approach in Cloud Computing
R. Siddthan, Dr.Nagarajan A. · International Journal of Engineering and Technology · 2018
Virtualized infrastructure becomes an attractive goal for cyber attackers for launching advanced attacks in cloud computing.Several existing techniques are utilized for predicting the attacks in the cloud data.It helps to predict the attack effectively and efficiently.But it is difficult to classify the cloud data as normal and attacker's data.Hence a novel security analysis of big data using classification approach is proposed in this work for detecting and classifying the advanced attacks in virtualized infrastructures.Here the logs of the network and user's applications are gathered from the guest virtual machines (VMs).These data are preserved in the Hadoop Distributed File System (HDFS).The process of extracting the features of the attacker is done by using a graph-based event correlation and the possible attack paths are identified based on the Map Reduce parser.After that, the presence of attack can be determined by performing two phase machine learning such as logistic regression and belief propagation.Here the logistic regression can be implemented for calculating the conditional probabilities of an attack regarding the attributes, and belief propagation for calculating the belief in the attack's presence depending upon them.Finally, a fuzzy classification approach is utilized for classifying the normal and attacker's data.The performance of the proposed approach is assessed by utilizing a well-known malware and compared with the prevailing security approaches for virtualized infrastructure.The experimental analysis reveals that our approach performs better in identifying and classifying the attacks with high efficiency and reduced performance overhead Keyword -Cloud Computing, Big Data, Hadoop and MapReduce, Virtual Machine Security, Fuzzy Classification.