Zero-Day Attack Path Identification using Probabilistic and Graph Approach based Back Propagation Neural Network in Cloud

Abstract

In the current environment, Networks are generally installed and employed by fundamental security defense procedures like firewalls, Intrusion Detection Systems. It is generally not stress-free for adversaries' to break down the machine. Rather than targets, it usually depends on an attack events chain to flourish threats. A zero-day attack is defined as unknown threats in software for which either patch is not issued or developers are not aware of it. Among many other attacks, this attack is considered as most susceptible one. The number of these exploits discovered remains rising at an increasing rate in the current situation. When these exploits happen in an attack path, the path remains a zero-day attack. The proposed work is developed to identify the Zero-Day Attack path using Probabilistic and Graph Approach based Back Propagation-Neural Network. If specific attack actions avoid system calls, proposed instance graphs capture the complete zero-day attack paths. An approach based on Back Propagation Neural Network outperforms the existing Accuracy, Correctness, and Misclassification parameters. The experimental result shows the effectiveness of the proposed work Back Propagation-Neural Network for zero-day attack path identification, which achieves a better result than the existing work.