Klm-based Profiling and Preventing Security Attacks for Cloud Computing: A Comparative Study

Abstract

Cloud computing is a digital era technology which uses the Internet to maintain data as well as applications in cloud data centers. However, this technology still meet numerous challenges and suffers from several attacks. For this reason, we proposed recently a new scheme called “klm-based profiling and preventing security attacks (klm-PPSA)” to detect both known and unknown attacks. In this study, we exhibit a comparative study of the klm-PPSA model using separately two accurate and interpretable machine learning algorithms: regularized class association rules (RCAR) and classification based on associations (CBA). Moreover, considering an interesting data set, three case studies of the proposal with three different implementations of the $klm$ security factors are given ( $k$ -PPSA, km-PPSA and klm-PPSA models). The experiments for each case study with run-time measurement were done. The obtained results show that: compared to $k$ -PPSA and km-PPSA models, the klm-PPSA model gives the highest performances in terms of sensitivity with both CBA and RCAR but with a processing time seven times more than CBA. However, RCAR gives an accuracy and specificity better than the CBA for all the models. Eventually, klm-PPSA system is able to detect and prevent several types of known and unknown attacks.