Terra: Enabling Edge-Based Stream Processing for Resource-Constrained Environments

The world is becoming more and more data driven and with the rise of the Internet of Things, more and more data is being generated and transferred every day. The traditional way of data processing by collecting data by default in cloud data centres is becoming unviable due to increased network congestion and demands for privacy-aware processing and real-time responses. The concept of fog and edge computing was introduced to solve these challenges, but these often still require transfer of data from sensor nodes to processing nodes, and the option of processing data on the sensor device itself is still absent.
This work introduces Terra, which enables query processing on resource-constrained sensor nodes at the very edge of modern sensor network databases. Integrated into a state-of-the-art network database, this approach significantly minimises network traffic, leading to a big reduction in energy consumption. Through a series of real-world experiments on physical devices, we evaluate Terra’s performance and experimentally derive an energy cost model that verifies Terra’s capabilities. In particular, we show a significant reduction in network transfer and therefore energy cost when pushing aggregate computation onto sensor devices.