Research on Behavior Trust Based on Bayesian Inference in Trusted Computing Networks

Recent years have seen explosive growth in information security problem. Trusted computing is increasingly used in the fields of information security. The basic idea of trusted computing is different from traditional security method. In other words, system abstracts normal behavior and compare with real-time behavior. Most of the current research focuses on static integrity metrics. However, it is not enough for dynamic measurement in trusted computing environment. In this paper, we present an efficient identification methodology of malicious insider user behavior. We focus on modeling insider user normal behavior in trusted computing networks. Our research methodology provides an approach to construct, assess, and optimize insider user normal behavior model. Our model framework derives from a dynamic Bayesian perspective. The experiment shows efficiency and robustness of our prototype model.