Efficient Hybrid Sensor Data Recovery Scheme
Ei Khaing Win, Tomoki Yoshihisa · 2019
In the emerging Internet of Things (IoT) technologies, network-connected sensors or sensor-attached devices are used for sensing purposes and a large number of devices utilize the sensor data stream to analyze the situations of the real world. In case of the client-server model, to keep the reliability of the result of sensor data analysis, the server has to re-transmit the lost data to the loss-encountered receivers. Reducing the number of streams for the lost data recovery is important to tackle the scalability challenge. To overcome the client-server model problems, the peer-to-peer model emerges as a paradigm. In the peer-to-peer data recovery, the assisted peer selection is an important issue. Although there are many existing peer selection methods, these do not consider the fixed communication bandwidth and data availability of the assisted peer cooperatively. Hence, in this paper, we introduce a hybrid sensor data recovery scheme considering each peer's fixed communication bandwidth and data availability. According to our simulation results, we confirmed that our method introduces less number of average recovery streams with faster catch-up time.