Characterization of an Acoustic Sensor Network to Improve Clock Synchronization Strategies

The prevalence of bird species is especially valuable as indicator for biodiversity, environmental health, and ecosystem changes. The development of reliable AI-based species classifiers and low-cost Autonomous Recording Units (ARUs) has significantly improved the capabilities of automation in species monitoring and localization. One ongoing challenge in the automated evaluation process is the issue of clock drift among the distributed ARUs, which can affect the accuracy and synchronization of data collected across multiple devices. Specifically, the clock drift of AudioMoths ARUs is of interest as they are commonly used in this context. In this paper, we characterize the clock drift in AudioMoth ARUs and explore different clock synchronization strategies. Additionally, we investigate the impact of the clock drift and the time synchronization strategies on bird call localization as an example in a simple TDoA localization simulation.