PROJECT TITLE :

Statistical Learning for Anomaly Detection in Cloud Server Systems: A Multi-Order Markov Chain Framework - 2018

ABSTRACT:

As a significant strategy to make sure the protection of IT infrastructure, anomaly detection plays a additional necessary role in cloud computing platform which hosts the whole applications and knowledge. On high of the classic Markov chain model, we proposed in this Project a possible multi-order Markov chain based framework for anomaly detection. In this approach, each the high-order Markov chain and multivariate time series are adopted to compose a scheme described in algorithms along with the coaching procedure in the form of statistical learning framework. To curb time and house complexity, the algorithms are designed and implemented with non-zero worth table and logarithm values in initial and transition matrices. For validation, the series of system calls and the corresponding come back values are extracted from classic Defense Advanced Analysis Projects Agency (DARPA) intrusion detection analysis knowledge set to form a two-dimensional test input set. The testing results show that the multi-order approach is in a position to produce additional effective indicators: additionally to absolutely the values given by a private single-order model, the changes in ranking positions of outputs from completely different-order ones additionally correlate closely with abnormal behaviours.