Article Preview
TopIntroduction
Since the dawning of the computer networking, IDS have played an essential role in assuring of the network's safety for the users, but this role has changed in recent decades (Keegan et al., 2016). Previous studies have not capable to efficiently detect the attacks instantly. therefore, they have been commonly utilized as forensic means to examine security evidence in ex-post facto. (Keegan et al., 2016). As network traffics are increased and attacks became more and more obvious with the prevalence of using the Internet since the 1990s, the demand for a speedier intrusion detection system to diagnosis and block attacks became essentially. To address this issue, researchers worked on understanding patterns of network traffic, which resulted in the development of behavior-based and signature-based detection approaches. In Signature-based techniques, the network traffic is compared with the pattern of previous known attack signatures. whereas behavior-based techniques detect intrusions by differentiation it from legitimate network traffic behavior (Keegan et al., 2016). Cloud computing is a modern technology that facilitates and allows the dynamic and varied of using computational resources and services. In spite of the advantages of cloud computing, the security and privacy of cloud services are still the critsl issues.The adoption of cloud services have influenced the security situation of organizations infrastructures; so, the emerging threats and risks that carried out via this model should be mitigated and classified (Kandias, Virvilis, & Gritzalis, 2011).
To utilize intrusion detection for cloud computing, new computational models are necessary to be constructed using the paradigm of parallel processing. As an alternative, current computational models require modified for the cloud computing environment. Numerous computational algorithms unable to use used in the cloud computing structure directly (Zhao, Ma, & He, 2009). To address this shortcoming, researchers have begun implement and designing several machine learning applications that have the ability to work in environment of cloud computing. Additionally, many researchers have highlighted the importance of utilizing machine learning algorithms to cloud computing in order to intrusion detection evaluations (Hu & Shen, 2012; Suthaharan, 2014).
Popular attacks form that target cloud computing is distributed denial of service (DDoS) attacks (Badve & Gupta, 2016; Ficco & Rak, 2015; Gupta & Badve, 2017; Osanaiye, Choo, & Dlodlo, 2016). Interested readers may refer to (Gou, Yamaguchi, & Gupta, 2016; A. N. Khan, Kiah, Madani, Ali, & Shamshirband, 2014) for additional cloud-related privacy and security issues. In DDoS attacks, the attacker attempts to infect and take control of enormous numbers vulnerable computers, to carry out and coordinated attacks. Current defence techniques that are able of managing quantities of data, generally suffer from handling data include irrelevant or redundant features, which rise the training and classification time (Peng, Choo, & Ashman, 2016).