Our research provides a solution to some known problems in sensor data system: 1) data is often locked away in proprietary systems, if it is even preserved at all, and 2) there are limited tools to automate processing the vast amount of raw data. Based on big data analytics, the architecture we designed can operate smoothly on low-cost yet relatively high-performance nodes, such as Raspberry Pi.
The new system has two main features. Firstly, it is fully distributed, which means the data storage and processing happen on each node and ready to scale. Secondly, it is fault-tolerant. A single node failure will not affect the data and the system operation.