Intrusion Detection Systems (IDSs) play a fundamental role in ensuring the reliability of green Cyber-Physical Systems (CPSs). Yet, existing IDS approaches did not capture the Inconsistent Sequential Pattern (ISP) in green CPS, thus resulting in high misclassification and limited sustainability. To address this issue, this study proposes a sustainability-driven ISP-aware intrusion detection framework that integrates ISP analysis and an energy-aware optimized deep ensemble learning model. The proposed framework encompasses the major layers, including the perception layer, transport layer, network layer, and control layer. The perception layer is responsible for authentication, whereas the transmission layer acts as an intermediate node between the perception layer and the network layer. Here, the network layer is generalized well enough to perform ISP-aware intrusion detection. Further, the proposed approach employs behavioral similarity modeling to enable fine-grained intrusion detection across CPS layers. Moreover, an optimized deep ensemble model is employed to detect intrusions while ensuring computational efficiency and generalization. Finally, the commands are executed through the actuator in the control layer. Thus, the proposed work obtains noticeable accuracies of 99.0654% and 99.4523% regarding network and IoT data, respectively. Hence, the experimental results confirm that the incorporation of ISP analysis significantly enhances IDS reliability and resilience in green CPS environments.
Sagar et al. (Sun,) studied this question.