Blockchain-Enhanced Federated Learning for Secure Malicious Activity Detection in Cyber-Physical Systems

Detecting and securing against malicious activity in cyber-physical systems (CPS) while optimizing resource utilization pose significant challenges. In this paper we propose an innovative approach that combines water wave optimization (WWO) for feature selection, deep recurrent neural networks (RNN) for training, and blockchain-based security to enhance the efficiency and accuracy of malicious activity detection in CPS. The WWO technique is applied to extract the most relevant features from the intricate and high-dimensional CPS data, effectively reducing dimensionality and improving computational efficiency. These selected features are then inputted into a deep RNN model, capable of capturing temporal dependencies and patterns within the data, thus enabling effective identification of malicious activities. To strengthen the security of the CPS system, we integrate blockchain technology as a foundation for secure data storage and management. By leveraging blockchain, we establish a decentralized and tamper-proof ledger that ensures the integrity and transparency of detected activities. The immutability of the blockchain provides robust protection against malicious tampering or unauthorized modifications, enhancing the overall security of the CPS system. We evaluate our proposed system architecture using a real-world CPS dataset, showcasing its superiority in accuracy and efficiency compared to traditional feature selection methods. The WWO-based feature selection significantly reduces the feature space while preserving critical discriminative information necessary for detection. The deep RNN model, trained on the selected features, achieves high accuracy in classifying and identifying malicious activities. Moreover, our approach exhibits robustness and adaptability in the face of evolving attack patterns. By training the deep RNN on a diverse and extensive dataset, the model becomes more resilient to unknown attacks, thereby enhancing CPS security. The integration of water wave optimization for feature selection, deep RNN for training, and blockchain-based security offers a promising avenue for improving malicious activity detection in CPS.