Protecting Smart City Infrastructure : A Cloud Computing Security Model Based on Enhanced Auto Encoder Based Convolutional Neural Network (EAE-CNN) With Mdata Model

Cloud computing infrastructures have rapidly expanded to become a major concern for security and are often difficult to protect using traditional approaches. As cloud environments become more complex and highly distributed, signature-based security is becoming insufficient to counter advanced, and constantly evolving cyber threats. To enhance automated incident response, anomaly detection, and real-time threat detection, this research employs machine learning methods, deep neural models, and intelligent automation. It proposes a new attack detection framework, the Deep Learning–based Advanced Cyber Security Analysis Framework (ACAM), which is based on a multi-dimensional data association and intelligent analysis (MDATA) model, which represents dynamic and spatio-temporal information more comprehensively than a traditional knowledge graph. The four primary parts of the framework are an attack detection module, alerts correlation module, a subgraph generation module, and a knowledge extraction module. An Enhanced Autoencoder-based Convolutional Neural Network (EAE-CNN) is used in the attack detection phase. In all, these components are used to filter out false alerts and enhance the identification of multi-stage attacks. The experiment's results show that the suggested cyber resilience model improves the overall resilience of cloud systems by combining cybersecurity and cloud computing information security, achieves better results than existing methods.