2. Literature survey
[12] HammiBadis, Guillaume Doyen and RidaKhatoun had presented a Collaborative Approach for a Source based Detection of Botclouds. Each tenant is treated separately and for each one a tree is constructed. The objective of using a tree-based hierarchical architecture is to reduce the complexity of the time and cycles of execution of the detection algorithm. Indeed, the proposed approach is based on the idea of (1) applying the PCA-based detection algorithm, small sets of VMs after their data has been standardized, and (2) aggregating the result set to decide on the state of the tenant (attacker or not). However, it doesn’t prevent against attacks in CSPs by cloud infrastructure.
[13] RémiCogranne, Guillaume Doyen, NisrineGhadban and BadisHammi had submitted a Decentralized and Robust Detection Method for Multi-Tenant Virtualized Environments This technique focused on developing and validating a solution to detect malicious operations by virtual hosts infected with botnets in the public cloud context. Although, this method doesn’t prevent from data leakage problem thus leads to major security issues in cloud.
[14] GauravSomani, Manoj Singh Gaur Suggested a fresh ”Inside-out Scale” approach that reduces the ”Resource Utilization Factor” to a minimum value for fast assault absorption. The proposed strategy sacrifices victim service resources in relation to other co-located facilities and provides mitigation facilities with those resources to determine their accessibility during the assault. Eventhough, this method does not deter data collection assaults in the cloud processing.
[15]WesamBhaya, Mehdi EbadyManaa A combination of unsupervised data mining techniques was introduced with the introduction of an intrusion detection system. The entropy idea in terms of windowing incoming packets is implemented using information mining method using Clustering Using Representative (CURE) as a cluster analysis to detect the network flow DDoS attack. However, only the information theory was regarded to transform nominal data to numerical data using entropy windows is considered. It doesn’t prevent from the attacks in frequency of packets during network flow.
[16]AmjadAlsirhani, SrinivasSampalli and Peter Bodorik the DDoS detection scheme provided those advantages from the resources of cloud computing. It entails of three notions: algorithms of classification, similarcalculating, besides a system of fuzzy logic. The notion of parallelism is used to effectively speed up the implementation of the classification algorithms used. It assessed the classification algorithm and the DDoS detection parallel processing by configuring a test-bed that consists of one master and three slaves. Although, this method only detect the attacks not prevent the system from the detection.
[17] Salman Iqbal et al. In terms of their cloud models, cloud-based attacks and vulnerabilities are collected and classified. We also present taxonomy of cloud security assaults and prospective mitigation strategies in order to provide a thorough knowledge of safety needs in the cloud setting. However, it fails to ensure integrity and confidentiality.
[18] Bo Li et al. New cluster-based intrusion detection system to track network traffic in the cloud environment to support intra-VM networks traffic monitor. A traffic deduction framework is also designed to eliminate redundant network traffic and reduce the burden of IDS clusters. It does not involve physical switch support and uses intra-VM network traffic to control virtual network traffic instead of virtualized IDS. Although in real cloud environments, this system is not efficiently suited.
[19] Mohamed Idhammad et al. had presented a distributed intrusion detection system for cloud settings based on machine learning. This system had been intended to be inserted side by side in the cloud with the cloud provider’s edge network parts. It is used to intercept incoming network traffic to the physical layer’s edge network routers. Even though, it is not well suitable for real world’s IDS deployment.
Thus from the above discussion, it is revealed that, in [12] doesn’t prevent against attacks in CSPs, [13] doesn’t prevent from data leakage problem, [14] doesn’t prevent from the attacks in frequency of packets, [15] does not deter data collection assaults, [16] only detect the attacks not prevent the system from the detection, [17] fails to ensure integrity and confidentiality, [18] not efficiently suitable in real cloud environments, [19] not well suitable for real world’s IDS deployment. Hence in order to tackle such issues, there is a need to innovate a new strategy in the field of cloud environment.