Data Prediction Algorithms in Real-Time Using Internet of Things and Named-Data Networks

Current paper presents a new perspective on Wireless Body Area Network (WBAN) data processing. There are three aspects considered herein: data management, data security and post-processing using Machine Learning (ML). The model was used to predict values which can be useful indicators for remote patient monitoring or athlete performance tracking. We selected FIWARE platform for data management. The security layer was resolved using a proxy server and authentication methods for both sensors and users, whilst Named-Data Networks (NDN) was needed for external communication. Data prediction was offered by Deep Echo State Networks (DESN), which offers good performance in terms of RMSE, training time and prediction time. For experiments, the WESAD dataset used different measurements collected over a longer period from different body sensors (temperature, blood pressure, EKG). The proposed approach is a suitable alternative for data management using the API developed by us, while NDN secures data retrieval from outside the platform.