Hybrid Deep Learning Network for Attack Detection in Cloud Environment Using Bot-IoT Dataset

The rapid development of new emerging computing technologies has prompted many enterprises to outsource their data and computational needs. Such services must always adhere to safety standards that include privacy, accessibility, and integrity; therefore, an extremely secure infrastructure is among the most critical features of cloud computing environments. By comprehending malware activity across its entire behavioral spectrum, conventional security measures are significantly surpassed. A hybrid Deep Learning (DL) model, Convolutional Neural Network-Gated Recurrent Unit (CNN-GRU), is proposed for detecting attacks in cloud data. The BoT-IoT dataset is used to validate the solution by characterizing various types of malicious behaviors. The proposed model is compared with traditional Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) models in terms of performance reliability and execution time. Achieving a maximum accuracy of 98.8% and detecting attacks in just 11 seconds, the hybrid DL model demonstrates promising potential for enhancing security monitoring and identifying malicious behavior.