A Vulnerability based Attack Detection and Mitigation in Cloud SaaS Framework

Cloud computing provides various cost-effective on-demand services to the user and so it is rising up as a real trend in the IT service model. More particularly, the security is the ultimate question in the mind of each cloud user. A huge interest is being paid by the research community to detect the attack exist in the network. This research work intends to introduce a novel framework that combines the attack detection and mitigation, which gives the SaaS provider to decide relevant fields to be extracted, when huge traffic is available. Deep learning-based attack detection is carried out and based on the vulnerability in the network, the model switches to mitigation process. Initially, the feature selection is carried out by Opposition Based Crow Search Algorithm (OCSA). The selected features are subjected to attack detection process via Deep Belief Network (DBN) model, where the presence of attacks is determined. Subsequently, the ‘vulnerability assessment’ is carried out by evaluating the risk level via correntropy variation features. This phase decides ‘how vulnerable the network is’ with the presence of attack. This decision is based on fixing a threshold on risk level (RL). Moreover, the decision tells whether to execute the ‘mitigation process’ or not. In the attack mitigation phase, bait-based mitigation process is carried out. The proposed vulnerability-based attack detection and mitigation system beat the traditional methods with a packet loss ratio of 16% and a throughput of 92% © 2022 School of Science, IHU. All rights reserved