A Hybrid CNN-LSTM-ANFIS Model Optimized With PSO For Efficient Ddos Detection

Distributed Denial-of-Service (DDoS) attacking method is among the most common and severe threats impacting today’s networks. Such attacks target to flood servers, disrupt services, and significantly inconvenience users. In response to this research problem, this study introduces a novel approach using a CNN-LSTM hybrid deep learning model. In addition, an Adjustable Neuro- Fuzzy Inference System (ANFIS) is incorporated with the system to strengthen the decision-making part depending on the flow of attacks and the network condition and Particle Swarm Optimization (PSO) is used to optimize hyperparameters of the proposed hybrid model, including filter size, layer number, dropout rate, and the learning rate of the CNN-LSTM model. The optimization tunes the parameters of ANFIS, including membership functions and fuzzy rules, hence improves the performance in detecting dynamic DDoS attacks. The proposed model is validated with NSL-KDD dataset containing seven data layers, which exhibited a detection accuracy of 99.30%, a much higher rate than the existing methods. The proposed approach claims that the integration made can perform automatic parameter optimization thereby minimizing human intervention, enhancement of detection rate, convergence rate, and robustness in identifying new emerging DDoS threats.