Impregnable Defence Architecture using Dynamic Correlation-based Graded Intrusion Detection System for Cloud

Abstract

Data security and privacy are perennial concerns related to cloud migration, whether it is about applications, business or customers. In this paper, novel security architecture for the cloud environment designed with intrusion detection and prevention system (IDPS) components as a graded multi-tier defense framework. It is a defensive formation of collaborative IDPS components with dynamically revolving alert data placed in multiple tiers of virtual local area networks (VLANs). The model has two significant contributions for impregnable protection, one is to reduce alert generation delay by dynamic correlation and the second is to support the supervised learning of malware detection through system call analysis. The defence formation facilitates malware detection with linear support vector machine- stochastic gradient descent (SVM-SGD) statistical algorithm. It requires little computational effort to counter the distributed, co-ordinated attacks efficiently. The framework design, then, takes distributed port scan attack as an example for assessing the efficiency in terms of reduction in alert generation delay, the number of false positives and learning time through comparison with existing techniques is discussed.