Mitigating Unknown DDOS Attacks with RPL: A Computationally Efficient IDS Framework

The proliferation of internet usage has heightened cybersecurity challenges, particularly Distributed Denial-of-Service (DDoS) attacks. Traditional security measures, including Intrusion Detection Systems (IDS), are often inadequate against sophisticated threats. This study enhances IDS capabilities using advanced machine learning and deep learning models, including CNN-RPL, CNN, FTC, and FusionNet. The proposed CNN-RPL model integrates Convolutional Neural Networks with Reciprocal Points Learning for robust Open-Set Recognition, achieving superior accuracy in detecting known and unknown DDoS attacks. Experimental results on the CICIDS2017 and CICDDoS2019 datasets demonstrate high accuracy, proving the model's effectiveness in safeguarding against evolving cyber threats.