Research Of Cloud Computing System Security-state Prediction Based On Semi-quantitative Information

Cloud computing as a business symbol of utility computing in current information technology,has a significantly influence on the whole operation patterns of the IT industry.To establish a stable and reliable cloud computing platform can not only guarantee the objectives of existing business,but even provides strong technical support for the development and innovation of cloud-driven business model.The security-state monitoring technology of cloud platform is an important branch of quality of cloud service,which can reflect the current condition of the cloud computing system according to the characteristic indicators.Moreover,the system managers can make the judgment more quickly and accurately through the results of the cloud computing security-state assessment,and then make early decision on the follow-up work to ensure the normal system operation of cloud service.Cloud computing system is a complex network cluster service system.To achieve a comprehensive and accurate cloud environment situation,it is necessary to effectively modeling using various organizations of information and technical knowledge in the system.To construct the nonlinear complex system,it is needed to understand the structure of the system and then extract the characteristic indicators to make a reasonable identification.Compared to the linear system identification,nonlinear system identification is more difficult to be settled in current academic research.Meanwhile,to date there is no mature theoretical system for the nonlinear identification of complex systems.As such,the content of this research reflects the importance of theoretical research and practical application.To solve the problems mentioned above,this article focuses on improving the service quality of complex and ensure the reliability and security of the cloud system.A modeling method for cloud computing system is proposed using semi-quantitative information?information containing both quantitative data and qualitative knowledge?,where the Evidential Reasoning?ER?algorithm is used for indicators fusion and Belief Rule Base?BRB?model is regarded as the inference tool.BRB-based modeling can deal with semi--quantitative information,while can also describe the knowledge uncertainty of ambiguity and probability.Using ER algorithm to integrate the attributes and belief rules of system,it makes the reasoning process more reasonable and easy to be explained.Therefore,this method can deeply excavate the characteristic information in each link of the cloud system,so as to obtain a more realistic state assessment,and provide technical support for the managers to guarantee the quality of operation and maintenance.The main contents of this article are summarized with the following four parts.To analyze the reliability and security environment of cloud computing system,the Service Level Agreement?SLA?of cloud service is presented and regarded as background.Based on the Quality of Service,this chapter introduces the characteristic indicators of reliability and security in cloud system.SLA is a formal agreement to describe the guarantee provided by the cloud provider according to user requirement,which purpose is to maintain the normal operation of cloud system while satisfy the service quality.Focusing on the reliability and security of the cloud system,a framework of security assessment for cloud computing platform is established in this chapter.This framework can be comprehensively described multiple aspects of cloud system to ensure the reliability and security.To solve the problem of cloud computing security-state assessment,security indicators are adequately analyzed with the consideration of observational data and professional qualitative knowledge.This chapter proposed an assessment method of cloud computing security-state based on semi-quantitative information using ER algorithm.ER algorithm is employed to integrate all kinds of parameters with uncertain information included in cloud computing platform.This method can be used to describe the security state of cloud environment more objectively and accurately.A case study was presented to verify the effectiveness via the actual cloud computing platform.To solve the problem of ambient noise impacted on cloud environment,with the analysis of each security factor in the cloud system,a new cloud computing security-state assessment method based on Hidden BRB?HBRB?model is proposed in this chapter.The assessment model can be divided into two parts,the first denoted as BRB₁ is employed to capture the relationship between the impact of ambient noise of cloud environment and the hidden behavior,and the other denoted as BRB₂ is employed to construct the assessment model with hidden behavior in accordance with the observable data of internal cloud system.The ER algorithm is used to integrate the security logs which are captured from the internal cloud system.Moreover,due to the absence of prior knowledge,the accuracy of the initial attributes may not be accurate,which results in inaccurate assessment.In this chapter,the Maximum Likelihood?ML?algorithm is used to optimize the initial parameters of the model.A simulation experiment based on the dual factors of external attack and internal security incident was analyzed and the more accurate assessment strategy is validated.The results of the assessment are closer to the actual situation.To solve the problem of cloud security-state prediction,according to the research results above,a method of security-state prediction based on BRB model with the consideration of monitoring reliability is proposed in this chapter.In the actual cloud computing platform,complete system structure makes the malicious attacks occurred infrequently in a system.However,the vulnerability in the cloud system is indeed an objective existence that managers cannot ignore this highly disruptive destabilizing factor.Meanwhile,simulations in the actual system will cost much more time and financial costs for certain operators.Therefore,it is necessary to grab the observational data for security-state prediction.In the actual monitoring conditions,due to the interference influence on the monitoring accuracy and ambient noise,resulting in the low reliability index of data obtained may be reduced the accuracy.A concept of reliability in the prior attribute of the perceptive model is introduced in this chapter,where makes modeling more realistic.Moreover,considering the initial parameters are assigned may not be accurate,The Covariance Matrix Adaptation Evolution Strategy?CMA-ES?algorithm is employed to train the parameters.This optimization algorithm is highly efficient,especially in addressing the problem of cloud system,whose types of indicators are various as well as quantities of samples are deficient.A simulation experiment was analyzed,and the comparison results pointed out the optimized BRB prediction model can better predict the security state of the cloud system with higher prediction accuracy.