Toward Resilient Sensor Networks with Spatiotemporal Interpolation of Missing Data: An Example of Space Weather Forecasting
Masahiro Tokumitsu, Keisuke Hasegawa, Yoshiteru Ishida · Procedia Computer Science · 2015
This paper attempts to construct a resilient sensor network model for space weather forecasting. The proposed model is based on a dynamic relational network. A space weather forecasting is vital for a satellite operation because an operational team needs to make a decision for providing its satellite service. The proposed model is resilient for failures of sensors/missing data due to the satellite operation. In the proposed model, the missing data of a sensor is interpolated by other sensors associated. This paper demonstrates an example of the space weather forecasting involving the missing of the observation in a test case. In this example, the sensor network of the space weather forecasting continues a diagnosis by replacing faulted sensors with imaginary ones. The demonstrations showed that the proposed model is resilient against sensor failures due to suspend of hardware failures or technical reasons.