Adaptive hybrid framework for low-latency DDoS defense in consumer-centric SDN architectures
Abstract Software-Defined Networking (SDN) increases networking flexibility, scalability, and programmability by separating the control plane from the data plane. Nevertheless, this structure is also vulnerable to DDoS, which may result in performance degradation and service unavailability. We propose a new adaptive hybrid real-time DDoS detection and mitigation framework designed for consumer applications with tight low latency constraints such as telemedicine, online gaming, and video streaming services. The suggested architectural model incorporates four well-known techniques (LSTM-based DeepPredict-DDoS, Reinforcement Learning-based Adaptive DeepPredict-DDoS, Genetic Algorithm-based DeepPredict-GA, and ARIMA-based DeepPredict-ARIMA) through an innovative decision-making engine. Experimental results show an average detection accuracy of 97.08%, reduced latency by 80%, and packet loss rate as low as 0.027%. These properties make the solution scalable, efficient, and effective for consumer-level SDN systems. Graphic Abstract