A Conv-BiLSTM Sandwich Network for DDoS Attack Detection

In recent years, DDoS attacks have presented a significant threat to numerous organizational websites, including those of enterprises and network service providers, by flooding the servers with an excessive amount of internet traffic. This flood of traffic can cause the intended server to slow down or become completely inaccessible to authorized users, which leads to major functional and financial disturbances for businesses and organizations. Hence, there is a need for the detection of such networks before completely disrupting the system. We proposed an efficient hybrid deep learning mechanism in order to detect DDoS attacks. The proposed hybrid model combines CNN and BiLSTM to create a novel architecture for the system. The proposed method achieves a false positive rate of 0.02% and a true positive rate of 99.37%, resulting in a minimum loss of 0.54% and an accuracy rate of 99.80%.