Distributed Denial of Service Attack (DDoS) is an attack by compiling multiple systems on the internet with infected zombies/agents and forming a network of botnets. DDoS attacks resulted in financial losses, lost productivity, brand damage, downgrades of credit and insurance ratings, and disrupted customer and supplier relationships. In addition, IoT technology is also vulnerable to large-scale DDoS attacks. To prevent DDOS attacks, a model that can detect DDoS attacks is needed. In this research, we propose Deep Q-Network (DQN) to detect DDoS attacks. DQN is a reinforcement learning algorithm that combines deep learning and q-learning. The application of DQN is used to improve the accuracy of attack detection on the dataset. In this paper, the dataset used to detect DDoS attacks or not is the CICDDoS2019 dataset provided by the Canadian Institute for Cybersecurity. Based on the comparison of the methods carried out, the results of the proposed DQN method can detect 11 DDoS attacks and benign/normal data with better accuracy values compared to the LR and SVR methods. The results showed that the proposed model had an accuracy value of 96% and was better than LR and SCR methods
Copyrights © 2022