Detection of Malware in Cloud Environment using Deep Neural Network
Prajna Kotian, Reena Sonkusare · 2021
There is a tremendous demand for Cloud Computing in organizations, educational institutions, etc. Cloud Services reduce the cost of setting up an office space. The users are only required to have a laptop. The maintenance of applications, servers, and storage are all taken care of by the Cloud Service Provider. As cloud services are gaining popularity, the volume of malware attacks on cloud services has doubled in 2019, according to the 2020 TrustWave Global Security Report. When the user clicks on the link or attachments, malware gets executed in the cloud environment. These provide a source for cybercriminals to gain unauthorized access to the machines. These, in turn, makes the system run very slow, thus consuming the CPU, memory, and network bandwidth of the machine. Thus, the dataset is split into CPU, memory, and network parameters and fed as an input to Deep Learning models. CNN is used for deep learning. This method helps in obtaining an accuracy of more than 95% using the 2D CNN model. The novelty of this paper is that Standardization and hyper-parameter tuning is used to improve the detection accuracy. SMOTE (Synthetic Minority Oversampling Technique) algorithm is used to reduce the imbalance in the dataset and thus obtain a better confusion matrix.