Multi-Model Secure Redundant Storage for IoT Data Using Random Forests, Deep Neural Networks, and Adaptive Particle Swarm Optimization

Le résumé fourni par la source

With the increasing deployment of IoT systems, secure and efficient data storage has become a critical challenge. This paper proposes a multi-model secure redundant storage approach for IoT data by integrating Random Forest (RF), Deep Neural Network (DNN), and Adaptive Particle Swarm Optimization (APSO), refers to the complete proposed system that integrates Random Forest (RF) for feature extraction, Deep Neural Network (DNN) for anomaly detection and sensitivity classification, and Adaptive Particle Swarm Optimization (APSO) for dynamic storage strategy adjustment. The RF extracts key features from high-dimensional data, DNN detects and classifies anomalies, and APSO dynamically adjusts storage parameters for optimized redundancy. The model was evaluated on the SmartHomeIoTData-v1.0 dataset, comprising 1,000 devices and over 1,000,000 data entries across temperature, humidity, and status metrics. Compared to baseline models (KNN, SVM), our approach improves accuracy from 90% to 95%, increases storage resource utilization to 75%, and reduces data loss probability to 0.01%. These results demonstrate enhanced system security, efficiency, and responsiveness on resource-constrained devices.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.