SoC based IoT sensor network hub for activity recognition using ML.net framework
Alexandru Iulian Alexan, Anca Alexan, Stefan Oniga · 2020
Nowadays single-board computers - SBC and System on a Chip SoC's get smaller and provide more processing power, yielding perfect alternatives to low-cost local hubs for small sensor networks. Even though sensors can now be connected directly to the cloud, for example, the ESP32 SoC, a local hub can provide a simple interface for multiple sensors to reach the IoT cloud. One of the best options for a local hub is the Raspberry Pi SBC, based on the Broadcom BCM2837 SoC, due to its small form factor and power consumption. We propose a system based on the Raspberry Pi SBC that interconnects connects two SensorTag CC2650 Bluetooth modules to the cloud, a MongoDB database. The obtained data is then processed offline using Microsoft's ML.NET machine learning framework to detect the user's activity. This framework was chosen as it's free, cross-platform, and open-source.