Energy-Efficient Routing with LSTM-Attention DQN for Anomaly Detection in Networks

Energy-efficient routing is significant to extend the lifespan of IoT networks, which are typically constrained by limited energy resources. Anomaly detection in communication networks is essential for identifying security threats such as DoS, R2L, Probe, and U2R attacks. This paper proposes an Energy Efficient Routing with LSTM-Attention Mechanism-based DQN for efficient and accurate anomaly detection. By integrating Long Short-Term Memory (LSTM) with an Attention Mechanism and Deep Q-Networks (DQN), the system improves detection performance while ensuring energy efficiency. The system is evaluated against traditional approaches, including AK-NN, DPC-DBN, Decision Trees (DT), and RNN-LSTM, using performance metrics such as precision, recall, F-measure, and accuracy. The results show significant improvements: precision (1.08), recall (0.44), F-measure (0.83), and accuracy (0.52) over conventional models. These findings highlight the proposed system's ability to detect various attacks effectively, while optimizing energy consumption. Its significance lies in providing a robust, scalable, and real-time solution for anomaly detection, particularly in resource-constrained environments, making it a valuable tool for modern network security.