A Novel Approach for Detecting Online Malware Detection LSTMRNN and GRU Based Recurrent Neural Network in Cloud Environment

Abstract

Now a day’s most of the organizations depend on cloud infrastructure for application, storage and real time access perspective. Cloud based application save as a backbone to organization in terms of maintainability, scalability and management underlying infrastructure. To meet the demand of such organization cloud service providers provide Infra-Structure as a Service (IaaS) through Amazon EC2, Microsoft and Azure Virtual Machine. Because of higher utilization of cloud platform, it becomes more targeted to attackers. In security view of IaaS, malware become the most dangerous threat to IaaS Infrastructure. In proposed method, two RNN architectures of deep learning are considered for detecting malware in cloud virtual machine such as LSTMRNNs (Long Short Term RNN) and GRU (Gated Recurrent Unit RNN). Behavioral features of CPU, disk utilization and memory are learnt. Running applications in real time environment in online cloud platform with no restriction and capture all behaviors of normal and benign applications. This model achieves higher detection rate of 99% over 42,480 datasets. Based on the ordering of input data, the performance varied.