Most developers assume connecting a consumer smartwatch to a web app is simple: call Chrome’s Web Bluetooth API, listen to the standard heart rate service, and read the numbers.
If you are using an expensive medical chest strap, that works. But if you try it with an accessible, off-the-shelf consumer smartwatch, you get absolute silence.
Here is why: to preserve battery life, budget smartwatches keep their optical green sensors asleep. They refuse to broadcast continuous vitals over standard Bluetooth channels unless their proprietary companion phone app commands them to.
For our NASA Space Apps Challenge 2026 project (Aura-Med), our challenge was keeping astronauts healthy on long-duration missions to Mars. When radio communications face a 3-to-22 minute speed-of-light delay, crew members cannot wait for Houston doctors, and they cannot rely on cloud servers.
We needed a web workstation that could communicate directly with an everyday $25 smartwatch through a browser, with zero phone apps in the middle.
Here is the story of how we reverse-engineered the watch's proprietary protocol and decoded live biometric telemetry directly inside Google Chrome.
The Problem: The "Closed Garden" of Consumer Wearables
When we examined the watch's Bluetooth Low Energy (BLE) advertisements, we immediately hit three major technical walls:
-
No Standard Heart Rate Service: The watch did not expose the public Bluetooth SIG heart rate service (
0x180D). Instead, it routed all communication through a proprietary, unadvertised vendor service. - The Dormant Sensor Barrier: Even after establishing a connection, the optical pulse sensor remained dark. The watch was waiting for a specific digital handshake that only the manufacturer’s native app knew how to send.
- Compressed Binary Envelopes: The watch didn't stream human-readable text. It packed incoming sensor readings into compressed Protocol Buffer (Protobuf) binary frames wrapped in a custom multi-packet transport layer.
🛠️ The Engineering Breakthrough: The 3-Step Solution
Rather than giving up and forcing users to run an Android phone in the background, we treated the browser as a full-fledged hardware controller.
1. Custom GATT Service Mapping
Through packet inspection, we identified the primary transmission and notification pipes on the watch. By explicitly whitelisting these proprietary vendor UUIDs inside Chrome's navigator.bluetooth request, we opened a direct, low-level data channel between the browser and the watch's onboard microprocessor.
2. The Bi-Directional "Wakeup" Handshake
We discovered that the watch operates like a digital walkie-talkie. To start the sensor stream, our web app had to emulate the companion mobile app's startup sequence:
- Wake Firmware: It sends an initial activation command that wakes the watch from standby.
- Energize Sensors: It transmits a secondary command that turns on the optical green photoplethysmography (PPG) LEDs for continuous pulse tracking.
-
Keep-Alive Loop: It exchanges two-way confirmation receipts (
Ready ACKsandCompletion ACKs) for every chunk of incoming data, coupled with a periodic heartbeat ping to prevent the watch from returning to sleep.
3. Client-Side Protocol Decoding
Once the raw binary chunks began flowing into the browser, we had to parse them. Standard Protobuf decoding relies on pre-compiled schema files using heavy server-side compilers.
To keep the application ultra-lightweight and 100% offline, we wrote an in-browser recursive Varint decoder in pure JavaScript. It traverses the raw byte stream on the fly, unpacks the serialized tree, and extracts:
- Continuous Heart Rate (BPM)
- Real-time Step Counts (for physical activity monitoring)
- Blood Oxygen Saturation (SpO2)
- Device Battery Health
In less than 50 milliseconds from the physical heartbeat, the biometric reading is decoded and rendered live on the astronaut’s dashboard.
Why This Matters
Hardware manufacturers often lock their devices behind proprietary clouds and walled-garden mobile apps.
By taking the time to understand the underlying transport framing and characteristic negotiation, we proved that ordinary, low-cost consumer hardware can be liberated using open web standards.
You don't need a $1,000 specialized medical device or a locked-down smartphone app to track real-time health data—just an off-the-shelf watch, a web browser, and an understanding of low-level protocols.
Aura-Med was created for the 2026 NASA Space Apps Challenge by Team Afrinit.
Check out our project live: Aura-Med Live Demo
Top comments (0)