Risk Assessment Model for Cloud Computing using Machine Learning Techniques

Abstract

Cloud computing has recently emerged as a new paradigm for hosting and delivering services to the customers over the Internet. A cloud computing system is a set of resources designed to be allocated ad hoc to run applications, rather than be assigned a static set of applications as is the case in client/server computing. Cloud Computing is being introduced and marketed with many attractive promises that are enticing to many companies and managers around the world, such as reduced costs, and relief from managing complex IT infrastructure. Virtualization technologies enable the abstraction and pooling of resources to be shared across the organization, data centers are designed around virtual machines, which are the new atomic units of computing. Traditionally, it is believed that any connectivity to systems or organizations outside of an organization provides an opening for unauthorized entities to gain access or tamper with information resources. Cloud computing moves computing and data away from desktop and portable PC’s into large data center distributed around the world. As a result, this will create a need for a considerable risk assessment approach to manage the various types of risks. Risk assessment is a concept that has developed to the point where it has the potential to address current limitations in cloud computing assessment methodologies. A risk assessment model for estimating the risk of cloud computing resources provides a solution to the risk problem, and would increase the chances of cloud computing adoption, as well as help in building trust in the cloud computing services. This thesis presents a new practical model of risk assessment to assess risk factors associated with cloud computing environment. In order to build a comprehensive risk assessment methodology, an extensive literature review was conducted to identify all risk factors that may affect cloud computing adoption. In this context 18 risk factors were identified. After the identification of risk factors, feature selection methods used to select the most effective features. The novelty of this thesis comes from the use of machine learning technique as a novel and efficient technique to assess risk in cloud computing environment. To build the model; first data mining algorithms are applied, then the ensemble method is used to combine the outputs of the data mining algorithms. The results of this research demonstrate the strengths of the use of data mining algorithms to assess risks, and it indicates that the methodology of using ensemble of machine learning algorithm represent a valuable alternative to existing methodologies.