Enhancing Cloud Security with Federated CNN-LSTM: A Novel Approach to Intrusion Detection

Intrusion detection in cloud environments faces challenges from scalable architecture, virtualized resources, and high data throughput. As cyber threats evolve, particularly within cloud infrastructures, there is a critical need for advanced, adaptable security mechanisms. Vulnerable to a wide range of attacks due to their scalable, decentralized nature and shared resources, cloud systems require robust security solutions. This paper introduces an innovative intrusion detection framework using distributed collaborative learning via federated learning, tailored for the unique security challenges of cloud environments. It enables nodes to independently train and update intrusion detection models without central data aggregation, enhancing privacy and collaborative learning while boosting attack detection capabilities across the network. Employing a convolutional neural network (CNN) within this federated structure, the proposed method significantly enhances privacy, security, detection accuracy, and system scalability. By utilizing distributed intelligence, this group learning approach surpasses traditional centralized learning, achieving a balance between local knowledge and comprehensive detection efficacy.