Learning and Predicting Method of Security State of Cloud Platform Based on Improved Hidden Markov Model

摘要

With the development and application of cloud computing platform, the learning and predicting method of security state of cloud platform provides security guarantee for cloud platform. However, due to the dynamic and uncertain nature of the cloud platform environment, the performance of learning and predicting of cloud platform security state will be affected. Therefore, the method of this paper combined the internal security state and observable state of the cloud platform to construct the cloud platform security state transition model, and established a linear regression AdaBoost learning and predicting algorithm for the observation state of cloud platform. Then, based on the observable state learning and predicting results, the hidden Markov model was used to learn and predict the security state of the cloud platform, and calculated the probability trend of the internal security state of the cloud platform in the future. The experimental results showed that compared with HMM, the proposed method could predict the hidden security state probability in the next two time periods in advance, and the convergence time and job response time increased by about 16% and 1% respectively.