Intrusion Detection System Using Long-Short-Term Memory with Mish Activation Function in Cloud

Intrusion Detection System (IDS) in cloud involves the monitoring and analyzing of system activities and network traffic to determine the unauthorized access or malicious behavior. IDS is significant in protecting sensitive data while maintaining the security and integrity of cloud resources. However, the IDS in cloud face struggles with accurately classifying complex, evolving the attack patterns because of the dynamic nature of cloud infrastructure and huge data volume. This research proposes Long-Short-Term Memory with Mish Activation Function (LSTM-MAF) to detect the intrusions accurately. LSTM-MAF enhances the complex detection, non-linear patterns, as well as the model’s adaptability by evolving threats which leads to a reliable and accurate performance. Initially, the NSL-KDD and CSE-CIC-IDS2018 datasets are used to evaluate the model’s performance. Min-max normalization normalizes the data in the pre-processing phase that preserves relationships in data. Then, the Honey Badger Algorithm (HBA) is established to select the most appropriate features effectively. The proposed LSTM-MAF achieves a better accuracy of 99.871% and 99.94% using NSL-KDD and CSE-CIC-IDS2018 datasets when compared to the existing methods namely, Hybrid Network-based IDS (HNIDS) and Correlation-based F-Differential Evolution (CFS-DE).