The prediction role of hidden Markov model in intrusion detection

Information security is an issue of serious global concern. The development of Internet increases the security risk of information systems greatly. This paper utilizes HMM (hidden Markov model) to realize the forecast ability of IDS (intrusion detection system). In this model, a command sequence or a control information sequence is regarded as a series of state transitions with a certain probability. The performance of several algorithms is compared such as F-BP (forward-back propagation) algorithm, Viterbi learning algorithm, EM (expectation maximization) algorithm, etc. In order to provide a soft boundary to the decision-making, fuzzy math is also introduced to this model. By this means, the intelligence of the IDS is improved and some decision-making abilities and reasoning abilities are offered to IDS. As well this paper reports the results about our project.