Temperature-Assisted Clock Synchronization and Self-Calibration for Sensor Networks

Synchronization is a pre-requisite for many sensor network applications. However, it remains challenging in sensor networks due to both the limited resources and the dynamic environments. In this paper, we propose a new two-phase clock synchronization scheme. The first one is the external clock synchronization phase, during which nodes update their clock by exchanging timestamp messages with the reference clock. Different from the conventional solutions, we propose to directly remove the clock skew during the external synchronization to achieve a higher synchronization accuracy and lower computational complexity. The second one is the clock self-calibration phase, as the accumulated clock skew will make the synchronized clock drift away again, we need to compensate the clock skew to maintain the clock synchronization accuracy. However, the compensation is non-trivial as the clock skew may not be constant due to the changing environment. Thus we propose the temperature-assisted clock self-calibration (TACSC) to dynamically compensate the clock skew according to the working temperature. Extensive simulation demonstrates that the proposed synchronization scheme can achieve a much lower root mean square error in the external synchronization phase. Furthermore, during the clock self-calibration phase, the TACSC scheme can improve the synchronization accuracy by more than one order of magnitude, which is verified by both simulation and testbed experimentation.