A Novel Deep Learning Model for Security Enhancement in Cloud Systems

Cloud computing is increasing in popularity; the challenge lies with keeping security one step ahead to combat new cyber threats. Writing security protections that match a highly skilled attacker trying to break into a cloud environment can become hard. In this paper, we reviewed the existing cloud security approaches and presented the research gap. Based on the findings, we proposed an innovative blend approach combining Convolutional Neural Networks (CNNs), and Recurrent Neural Networks (RNN) with Long Short-Term Memory (LSTM) for cloud security improvement. Spatial features from cloud-based data visualization (network traffic heatmap or user behavior pattern) are extracted using CNNs. RNNs as well as LSTMs are used to characterize temporal relationships among the sequential data and can learn sequences of anomalies that evolve with time (such as intrusion attempts, and abnormal activities over consecutive samples. The new model uses CNN for spatial feature extraction and integrates it with RNN-LSTM for effective time-series scrutiny to detect static as well as dynamic types of security threats. This integration of these deep learning methods with cloud environments allows it to perform real-time observations of all the cloud environment detections and hence detect as well as mitigate any possible risks.