Trust-Based access control in Cloud Computing using machine learning

Abstract

Cloud computing is a distributed environment, who hosts very dedicated computing resources accessed anytime from anywhere. This brings many advantages such as flexibility of data access, data omnipresence, and elasticity. As there is no control of data owner over the data so, this brings security threats. In order to securing the cloud environment from the malicious user is a complex task. Over the time various control access model shave been proposed for secure access in the cloud environment such as cryptographic-based access model, identity-based access control model and trust-based access control model. The users and cloud resources should be trusted before accessing the cloud. The existing access control models mainly overlook the user behavior and scalability of the trust management system. We have considered the trust-based approach, which provides access to the user in the cloud by their past behavior trust value. We consider some parameters such as: user behavior, bogus request, unauthorized request, forbidden request, Method not allowed and range not satisfied. We proposed a trust evaluation strategy based on The machine learning approach predicting the trust values of user and resources. The machine learning used such as: K-Nearest neighbor, decision tree, logistic regression, naive Bayes are considered in the work thesis. We implemented our proposed machine learning method in jupyter notebook simulator tool. We found the better result in terms of efficiency, prediction time and error rate.