Extremely large-scale sensing applications for planetary WSNs

The potentially widespread adoption of Wireless Sensor Networks presents a unique challenge both for the communication infrastructure of the Internet as well as for backend systems responsible for processing the data and making it available in useful form. In this paper we explore this space by making careful estimates on the traffic volumes of sensing data in a number of scenarios explored in the literature. Examples of the scenarios covered are patient monitoring, structural monitoring, vehicular applications and environmental monitoring. The results show that if there were a "flag day" for sensor network deployments in the near future they would as a whole dominate over other forms of mobile and wireless network traffic at least until the middle of the next decade. However, the overall data volumes would still appear to be manageable, although only barely. Based on these forecasts, we comment on the feasibility of processing data from large sensor deployments. We also present first results from our ongoing work towards developing a highly scalable framework for storing and processing data obtained from planet-scale WSN deployments. In particular, we study how MapReduce-based data processing could be used to deal with the scalability challenge, and argue using selected case studies that platforms such as Hadoop could indeed be used to deal with such data volumes.