This repository contains a set of Bluetooth Low Energy (BLE) applications that facilitate data delivery from wearable sensors to upstream systems. In particular, this setup was deployed to interface relay laptops with multiple DFRobot Beetle BLEs, and stream sensor data to an upstream neural network, as part of a larger dance moves trainer system.

The code in this repository include a relevant selection of:

• C++/Arduino code to be run on sensors,
• Go code to run a BLE central application in physical proximity to sensors,
• Python code to transfer data to upstream components over the internet

The stack provides:

• Reliable custom protocol w/ automatic reconnection
• Windowed (of configurable size) data streams
• Built-in data collection
• Portability

This system has been tested with 4 active connections * 120 BLE 4.0 packets (containing 20 payload bytes each) / sec. However, it has been casually observed to be able to perform under much higher load.