Enhancing Cloud Security: A Novel 1D CNN-Informer Framework for Intrusion Detection with Hunger Game Search Optimization

Storage and data access services offered by the cloud are major reasons for cloud computing's (CC) current popularity. As the number of threats to networks grows, security and privacy become paramount. Businesses and other organizations can take use of cloud computing's scalability, flexibility, and low-cost Tob storage infrastructure. When anything seems to have changed unexpectedly, an anomaly-based Intrusion Detection System (IDS) implementation will recognize it and quarantine the entries to protect the database's integrity. When it comes to sophisticated networking settings, approaches for clustering and classification based on machine learning are utilized for scalability and anomaly-based intrusion detection system assault categorization. Intrusion detection models built using machine learning are quick, efficient, and flexible enough to handle both known and new threats, such as zero-day assaults. In order to reliably anticipate attacks, this research suggests a new data processing and enhanced one-dimensional convolutional neural network (1D CNN)-informer architecture. As part of the data preprocessing phase, To standardize the formats of both structures' data and couple the 1D CNN and informer using cross-entropy loss. To speed up the training process and get to the least cost function faster, To normalize using the minimum-maximum feature scaling method. To make the Informer model stronger at learning the sequence links among data and effectively reduce prediction variability, a relative position encoding approach is developed. The suggested model's parameters are fine-tuned using the classification accuracy-boosting Hunger Game Search Optimization Algorithm (HGSOA). Using the revised cybersecurity CSE-CIC-IDS2018 dataset, the research employs these models to handle binary-class classification. Improving intrusion attack detection accuracy for intrusion detection systems in the cloud, among other performance metrics, is the goal of this study. In addition, To compare our findings to those in the relevant literature.