Time synchronization for wireless sensor networks

Recent advances in miniaturization and low-cost, low-power design have led to active research in large-scale networks of small, wireless, low-power sensors and actuators. Time synchronization is a critical piece of infrastructure in any distributed system, but wireless sensor networks make particularly extensive use of synchronized time. Almost any form of sensor data fusion or coordinated actuation requires synchronized physical time for reasoning about events in the physical world. However, while the clock accuracy and precision requirements are often stricter in sensor networks than in traditional distributed systems, energy and channel constraints limit the resources available to meet these goals. New approaches to time synchronization can better support the broad range of application requirements seen in sensor networks, while meeting the unique resource constraints found in such systems. We first describe the design principles we have found useful in this problem space: tiered and multi-modal architectures are a better fit than a single solution forced to solve all problems; tunable methods allow synchronization to be more finely tailored to problem at hand; peer-to-peer synchronization eliminates the problems associated with maintaining a global timescale. We propose a new service model for time synchronization that provides a much more natural expression of these techniques: explicit timestamp conversions . We describe the implementation and characterization of several synchronization methods that exemplify our design principles. Reference-Broadcast Synchronization achieves high precision at low energy cost by leveraging the broadcast property inherent to wireless communication. A novel multi-hop algorithm allows RBS timescales to be federated across broadcast domains. Post-Facto Synchronization can make systems significantly more efficient by relaxing the traditional constraint that clocks must be kept in continuous synchrony. Finally, we describe our experience in applying our new methods to the implementation of a number of research and commercial sensor network applications.