Data Prediction-Based Virtual Sensor in Digital Twin Scenario using Deep Learning Approach

Digital twin offered a lot of new possibilities in new technologies. One of the digital twin implementations in the internet of things is the virtual sensor. Separating the physical and logical aspects of the sensor and implementing a data prediction model on the logical side can lower the number of data transmissions between sensor nodes. A lightweight data prediction model that is suitable for typical iot resource-constrain devices is preferable. The result shows that L-MLPDP as the proposed model achieves the best result with 0.113 RMSE, 0.081 MAE, 0.072 MAPE, and 0.11 ms computing time.