Integrating Blockchain and Machine Learning for Secure Data Aggregation in Wireless Sensor Networks

ABSTRACT Wireless sensor networks (WSNs) face challenges in managing large volumes of data transmissions because of their resource‐constrained nature. They are typically deployed in unattended environments, making them highly vulnerable to security threats and attacks. This study presents the blockchain and machine learning (ML) algorithms‐based integrated secure data aggregation (BM‐SDA) framework, which is designed to detect intrusions while ensuring secure data aggregation and storage in WSNs. BM‐SDA incorporates a fast neural learning model (FNL), which is a single‐layer feedforward neural network, to enhance real‐time attack detection. It employs the synthetic minority oversampling technique (SMOTE) for data balancing and integrates the Kalman filter for noise reduction. To optimize feature selection, a particle swarm optimization (PSO)–based approach is also employed. Additionally, the blockchain module secures WSN's operations by managing (i) node registration, (ii) authentication, (iii) revocation, (iv) storing encrypted data hashes, and (v) credentials. Exhaustive simulation results on a WSN dataset demonstrate that BM‐SDA outperforms existing models by achieving 99.2% accuracy, 99.1% precision, 99.3% recall, 0.97 as a Matthews correlation coefficient (MCC), and a 35.8% reduction in transaction cost, proving its effectiveness for secure, scalable, and cost‐efficient WSN deployments.