Deep learning-based intrusion detection systems for securing cloud-native and distributed architectures

Cloud-native and distributed architecture has also deeply changed modern computing environments allowing for scalability, flexibility, and microservices deployments at scale. But this shift has also given rise to specific security challenges, such as dynamic attack surfaces, lateral movement of threats and advanced cyberattacks against containerized and serverless infrastructures. Traditional intrusion detection systems (IDS) have limitations in handling the scale, speed, and diversity of data generated during these incidents. In order to overcome these limitations, this work proposes a deep-learning-based intrusion detection framework with a particular emphasis on ensuring that it is secure in cloud-native and distributed systems. Utilizing state-of-the-art neural network architectures; Convolutional Neural Network (CNN), Recurrent Gnural Network (RNN) and hybrid deep learning models, this new model accounts for both spatial and temporal features of network traffic data. It combines real-time monitoring, anomaly detection, and adaptive learning capabilities to improve the accuracy of detection and minimise false positives. On standard datasets, we show that our new approach significantly overcomes classical machine learning methods in precision, recall and detection rate. Additionally, the model is scalable and cloud-native, while only importing sends for real-time deployment, enabling pervasive security monitoring. This work advances intelligent, autonomous cybersecurity solutions for next-generation cloud infrastructures.