Estimation of relative offset and drift between local clocks in wireless sensor networks

Abstract

In wireless sensor networks, each wireless sensor node records events occurred in its observation area with their observation time. Each wireless sensor node possesses its own local clock whose drift and offset are generally different from the others. In addition, it is difficult for the wireless sensor nodes to adjust drifts and offsets of their local clocks since transmission delay of messages between neighbor wireless sensor nodes are difficult to estimate due to CSMA/CA and RTS/CTS control for collision avoidance. Thus, it is difficult to achieve orders and intervals among events observed by different wireless sensor nodes. Moreover, even if multiple wireless sensor nodes observe the same event, their recorded observation times might be different and two observation records for the same event by multiple wireless sensor nodes are not always recognized as the records for the same event. Based on an assumption that obser- vation areas of neighbor wireless sensor nodes are overlapped, by using observation records of the commonly observed events by neighbor wireless sensor nodes, this paper proposes a novel method to estimate the relative drift and offset between local clocks of the neighbor wireless sensor nodes. Here, each sensor node only detects the occurrences of events and cannot achieve the locations where the events occur. Hence, commonly observed events between neighbor wireless sensor nodes are required to be detected. Our proposed method applies a heuristic that multiple observation records in neighbor wireless sensor nodes whose intervals are the same are estimated to be commonly observed events.