Design, Implementation, and Simulation of Self -Healing Networks Using Deep Learning in SDN Architecture

In modern network paradigms, ensuring high availability, resilience, and adaptability is critical, particularly with the rapid proliferation of technologies such as the Internet of Things (IoT), cloud computing, and ۵G. As network infrastructures become increasingly complex and dynamic, traditional static and manual configurations are no longer sufficient to guarantee uninterrupted service delivery. This growing complexity necessitates the development of intelligent, autonomous systems capable of detecting, analyzing, and resolving network issues in real time without human intervention. This paper presents a novel self -healing framework based on deep learning within a Software -Defined Networking (SDN) environment. The proposed system integrates a hybrid Long Short -Term Memory and Convolutional Neural Network (LSTM -CNN) model, designed to capture both temporal patterns and spatial features in network traffic for precise anomaly detection. Detected anomalies automatically trigger network reconfiguration through the SDN controller, enabling real -time mitigation and recovery. The framework is practically implemented using the Mininet emulator and Ryu SDN controller, with simulated scenarios including Distributed Denial -of-Service (DDoS) attacks, core link failures, and sudden traffic surges. Evaluation results demonstrate high detection accuracy (۹۶.۷%) and significant performance gains compared to rule -based and threshold -based methods, highlighting the framework’s effectiveness and potential for real -world deployment.

کلیدواژه ها:

Software -Defined Networking (SDN) ، self-healing networks ، anomaly detection ، LSTM -CNN ، Mininet ، Ryu controller ، deep learning ، network recovery