1
0
mirror of https://github.com/CovidBraceletPrj/CovidBracelet.git synced 2024-12-06 09:35:43 +01:00
Go to file
2021-12-08 18:09:00 +01:00
.github/workflows Run build and test actions on pull_requests too 2021-07-23 11:46:07 +02:00
img Add article to readme 2021-12-08 16:44:36 +01:00
include/tls_config Init entropy for desktop tests 2020-08-18 22:47:28 +02:00
lib Add protobuf library and make sure project compiles 2021-12-08 18:09:00 +01:00
src Add protobuf library and make sure project compiles 2021-12-08 18:09:00 +01:00
test Add platformIO files and adapt file structure 2020-08-18 22:46:49 +02:00
zephyr Simulate flash for native posix 2021-07-23 11:52:57 +02:00
.clang-format Format ens files and add .clang-format file 2021-05-25 18:06:15 +02:00
.gitignore Simulate flash for native posix 2021-07-23 11:52:57 +02:00
basestation.py initial code release 2020-06-08 11:15:05 +02:00
LICENSE Initial commit 2020-06-07 21:45:48 +02:00
platformio.ini Set fix platform version, so Zephyr v2.5.0 is always used 2021-07-23 11:47:50 +02:00
README.md Apply suggestions from code review 2021-12-08 16:44:36 +01:00

Contact Tracing Wristband Actions Status: test Actions Status: build

This project aims to add support the Exposure Notification (EN) protocol for tracing contacts from Apple and Google on low-power wristbands. It maintains full compatibility with smartphone Apps based on the EN protocol.

Note: this is a proof of concept and not ready for production
Contributions Welcome!

Features

  • Builds on Zephyr OS targetting the NRF52 BLE SOCs
  • Easy project setup using PlatformIO
  • Sends and receives exposure beacons as specified by Google and Apple for Covid Contact Tracing
    • Rolling, encrypted, anonymous beacons
    • Compatible with Apple iOS and Android phones
    • With rolling proximity identifiers and encrypted metadata
    • Proximity identifiers and BLE addresses change every 10 minutes
  • Stores own keys for 14 days
  • Stores contacts of more than 5 minute duration
  • Upon infection upload keys to a public database
  • Retrieve keys of infections from database
    • computes rolling proximity identifiers
    • compares to stored contacts to check for exposure

Note: as we for now do not use the flash for key storage, this currently only works on nrf52480 or you can just store a very small number of keys. Moving the keys to flash is on the TODO list and will fix this.

Getting Started

This project is based on platformIO for an easy setup process, see: https://platformio.org/platformio-ide.

Testing

To run the tests for the desktop environment select the task test for desktop or run:

platformio test -e desktop

Demo Video

Video Demo

Roadmap

Flash Support

  • Store long-term contacts in flash

Device Initialization

The devices need to be initialized:

  • Fix entropy: keys are always the same on boot up
  • Synchronize time
  • Set device name, user id or so in Flash (-> persistend configuration)

Device Synchronization

  • Extend beyond the simple (existing) basestation
  • Synchronize time
  • Get infections from DB, Check their signatures
  • Secure GATT services and authentication of base statation in general
  • Read keys form national databases

Extract Keys from Device

In case of an infection, the keys need to be extracted from the device:

  • Secure extraction of keys (Base version existing)

Device Updates Over the Air (OTA)

Securely updating many running devices over the air with signed firmware will greatly improve maintainability. Related feature branch: feature/ota

Logging

  • extensive logging: crash, reboot, battery level, charging state, contacts, memory useage, flash usage
  • Energy efficiency

Protocol Specifications

  • set scanning interval to the correct value, for now we just use the default
  • set advertisement interval, correct value, for now we just use the default: should be 200-270 milliseconds
  • set transmit power and store this in the corona beacons
  • extensive compatibility testing with Apple iOS and Android devices
  • BLE advertisements sets

Other

  • Contininous integration testing
  • More platforms: with display etc.

Possible Platforms

Possible platforms for real-world deployment many, as many of the cheap fitness trackers base on NRF52 or chips with similar capabilities. However, many would need the firmware to be shipped to manufactures.

  • Watch UI
  • Pine Time could be good for testing

Display

A basic visual interface based on LVGL is included. It can be tested on POSIX-compliant platforms by building it with the zephyr/build_native.sh script. SDL must be installed to provide a virtual display.

No risk contacts Low risk High risk