Wireless Sensor Network for habitat monitoring on Skomer Island

In the natural sciences, research often relies on extensive manual investigation. Such methods can be error-prone and obviously don't scale well. The development of autonomous data acquisition systems such as Wireless Sensor Networks (WSN) has provided a method to significantly reduce manual work and, as such, has the potential to enable researchers to address previously infeasible scientific questions. However, making the transition from WSN deployments in a laboratory to real-world deployments is still very challenging. Creating robust, error-free systems that are able to run autonomously in real-world environments without manual supervision has proven to be complex and, therefore, the number of successful collaborations between computer scientists and natural scientists is still limited. Here, we describe our successful attempt to design and deploy a WSN to monitor seabirds on Skomer Island, a UK National Nature Reserve. We summarize the evolution of the system over a period of three years, share insights on selected design decisions, and discuss both, our experience and the problems we have encountered.