Prevention of Runtime Malware Injection Attack in Cloud Using Unsupervised Learning

Cloud computing utilizes various Internet-based technologies to enhance the Internet user experience. Cloud systems are on the rise, as this technology has completely revolutionized the digital industry. Currently, many users rely on cloud-based solutions to acquire business information and knowledge. As a result, cloud computing services such as SaaS and PaaS store a warehouse of sensitive and valuable information, which has turned the cloud systems into the obvious target for many malware creators and hackers. These malicious attackers attempt to gain illegal access to a myriad of valuable information such as user personal information, password, credit/debit card numbers, etc., from systems as the unsecured e-learning ones. As an important part of cloud services, security is needed to protect business customers and users from unauthorized threats. This paper aims to identify malware that attacks cloud-based software solutions using an unsupervised learning model with fixed-weight Hamming and Mexiannet. Different types of attack methodologies and various ways of malicious instructions targeting unknown files in cloud services are investigated. The result and analysis in this study provide an evolution of the unsupervised learning detection algorithm with an accuracy of 94.05%.