Research on ICMPv6 DDoS Attack Detection Based on Integrated Feature Selection and LSTM-GRU
Yupeng Cheng, Huahu Xu · 2024
This paper proposes a method for detecting ICMPv6 DDoS attacks based on an integrated feature selection and LSTM-GRU model. With the widespread adoption of the IPv6 protocol, ICMPv6 has become a crucial component of network communication; however, its inherent characteristics make it susceptible to DDoS attacks. The paper begins with a review of existing detection methods, identifying limitations in traditional approaches when addressing emerging threats. To overcome these challenges, the study designs an integrated feature selection algorithm that combines IGR and CHI to select the most valuable features for detection. Based on this, an LSTM-GRU hybrid model is constructed, leveraging LSTM's ability to capture long-term dependencies and GRU's computational efficiency, significantly enhancing the speed and accuracy of DDoS attack detection. Experimental results demonstrate that the proposed method achieves an accuracy of 98.5% in detecting ICMPv6 DDoS attacks, outperforming single models and exhibiting strong adaptability and practicality in dynamic network environments.