Innovation-based timestamp validation for reliable sensor network synchronization

Sensor networks need a common timescale to be shared among their constituent elements. Accuracy and reliability are key issues in time dissemination, the latter aspect being specifically addressed in several synchronization schemes. In the paper, we present an approach to reliability assessment based on monitoring the innovation of the Kalman filter employed as the local clock servo in a sensor node. The assumed clock model is described by two state variables and its equations are general enough for most typical devices and operating conditions. Results provided demonstrate the capability to detect unreliable time information, thus improving synchronization performances.