Air quality monitoring station

Architecture

Monitoring Station	⇨	Message Queue	⇨	Data Collector	⇨	Data Publisher	⇨	Static Website
Raspberry Pi Zero + SDS011		AWS SQS		GitHub Actions		GitHub Actions		GitHub Pages

1. The Monitoring Station reads PM2.5 and PM10 from the sensor and sends the metrics to the Message Queue.
2. These messages are collected by the Data Collector.
3. The Data Publisher computes the hourly averages and publishes the data as a static website on GitHub Pages.

Components setup

Message Queue

Create an account in AWS and:

1. Create a SQS FIFO queue named airquality.fifo.
2. Create an user for the monitor and give it the following permissions:
   • SQS:SendMessage
   • SQS:GetQueueUrl
3. Create an user for the collector and give it the following permissions:
   • SQS:DeleteMessage
   • SQS:GetQueueUrl
   • SQS:ReceiveMessage

Monitoring Station

Composed by SDS011 and Raspberry Pi Zero with a Debian-based distribution (I use DietPi).

1. Setup AWS credentials

   • Setup monitor AWS credencials for the monitor IAM.

2. Install dependencies

   • Run pip3 install boto3 pyserial

3. Install the service

   • Copy monitor/airquality.py to /root/airquality.py.
   • Copy monitor/airquality.service to /etc/systemd/system/airquality.service.
   • Run systemd install airquality.service && systemctl daemon-reload

It sends a message every 10 seconds, so it fits AWS SQS free tier.

Data Collector

It collects the data sent to the queue every 15 minutes and stores it in a git branch named state. The branch must be checked out as a git worktree under .state folder.

It's implemented on GitHub actions. Setup the following secrets:

• AWS_ACCESS_KEY_ID
• AWS_SECRET_ACCESS_KEY

Data Publisher

It computes the hourly averages and publishes the results as a static website on GitHub Pages. The data and the website are stored into a branch named gh-pages. The branch must be checked out as a git worktree under .gh-tree folder.

The chart is built with Plotly.

Git worktrees

Collected and published data are stored on different branches that are managed as git worktrees. They are handled by stateful-action on the GitHub workflow.