Fine-grained network time synchronization using reference broadcasts
ACM SIGOPS Operating Systems Review - OSDI '02: Proceedings of the 5th symposium on Operating systems design and implementation
Timing-sync protocol for sensor networks
Proceedings of the 1st international conference on Embedded networked sensor systems
The flooding time synchronization protocol
SenSys '04 Proceedings of the 2nd international conference on Embedded networked sensor systems
Estimating clock uncertainty for efficient duty-cycling in sensor networks
Proceedings of the 3rd international conference on Embedded networked sensor systems
Low-power high-accuracy timing systems for efficient duty cycling
Proceedings of the 13th international symposium on Low power electronics and design
A tale of two synchronizing clocks
Proceedings of the 7th ACM Conference on Embedded Networked Sensor Systems
The true cost of accurate time
HotPower'08 Proceedings of the 2008 conference on Power aware computing and systems
Hi-index | 0.01 |
Time synchronization is an essential service in distributed computing and control systems. It is used to enable tasks such as synchronized data sampling and accurate time-of-flight estimation, which can be used to locate nodes. The deviation in nodes' knowledge of time and inter-node resynchronization rate are affected by three sources of time stamping errors: network wireless communication delays, platform hardware and software delays, and environment-dependent frequency drift characteristics of the clock source. The focus of this work is on the last source of error, the clock source, which becomes a bottleneck when either required time accuracy or available energy budget and bandwidth (and thus feasible resynchronization rate) are too stringent. Traditionally, this has required the use of expensive clock sources (such as temperature compensation using precise sensors and calibration models) that are not cost-effective in low-end wireless sensor nodes. Since the frequency of a crystal is a product of manufacturing and environmental parameters, we describe an approach that exploits the subtle manufacturing variation between a pair of inexpensive oscillators placed in close proximity to algorithmically compensate for the drift produced by the environment. The algorithm effectively uses the oscillators themselves as a sensor that can detect changes in frequency caused by a variety of environmental factors. We analyze the performance of our approach using behavioral models of crystal oscillators in our algorithm simulation. Then we apply the algorithm to an actual temperature dataset collected at the James Wildlife Reserve in Riverside County, California, and test the algorithms on a waveform generator based testbed. The result of our experiments show that the technique can effectively improve the frequency stability of an inexpensive uncompensated crystal 5 times with the potential for even higher gains in future implementations.