Gradient Clock Synchronization in Sensor Networks

Abstract

In this paper, we examine gradient clock synchronization in a system model that is typical for sensor networks: We assume message-delay uncertainties to be negligible, and communication to be infrequent. The gradient property requires the difference between any two network nodes' clocks to be bounded from above by a non-decreasing function of their distance. This means that every node has to be synchronized better to nearby nodes than to faraway nodes. We provide a lower bound for the achievable synchronization quality in our model and discuss its relation to an existing bound in a different model.