Noise Sensor Monitoring

This repository contains code for the noise sensor monitoring project. At Sunbird AI we're using this application to keep track of the sensors deployed in the field (i.e monitoring the battery level, signal strength). This application also handles communication between the sensors and an external server (receiving audio files, configuring the sensors e.t.c).

Technologies used

• Python is the primary programming language used
• Grafana used for visualization and creating the dashboard.
• InfluxDB is the time series database used to store key sensor metrics
• Eclipse Mosquitto is the MQTT broker used to facilitate communication between the sensors and the server.
• Docker and Docker-compose for spinning up containers and using prebuilt containers.
• AWS the system is deployed on an AWS EC2 instance.

Key python packages used

• Paho MQTT an MQTT Python client library.

Set this project up locally

• Clone this repository and cd into this project's directory.
• (Optional) Add your mosquitto configuration in the file ./docker/mosquitto/config/mosquitto.conf
• Make sure you have docker and docker-compose installed
• Create an external docker network named djangonetwork. docker network create -d bridge djangonetwork. This allows easy communication between the different containers that will be running.
• Run the command docker-compose up -d --build to run the containers.
• To view logs (in case something fails to work): docker-compose logs (TODO: create docker build file for the mqtt_client step)
• Visit http:\\localhost:8086 to setup influx db.
  • You'll need to setup a username, an organization and a bucket. Refer to the influx db docs.
• Visit http:\\localhost:3001 to view grafana and setup your dashboard.
  • Refer to the grafana docs for setting up an admin user
• Create .env file with the following info (Refer to the influxdb tutorial for how to get some of these):

INFLUX_DB_URL=http://localhost:8086
INFLUX_DB_TOKEN=<INSERT TOKEN HERE>
INFLUX_ORG=<YOUR-ORG>
INFLUX_BUCKET=<YOUR BUCKET>
MQTT_BROKER=localhost

MQTT Topics Description

Below are the MQTT Topics implemented in this project.

Topic	Example Message	Description
sb/sensor	{"deviceId": "SB1002", "dbLevel": 76...}	topic to which sensor sends its status. Required fields are devicdId, dbLevel, connected, batteryLevel, sigStrength

TODO: Add description of the other topics as they're implemented.

Setting up and running Django commands

• Ensure the containers are running. Use docker ps to see running containers.
• Apply the migrations using the command: docker-compose exec web python manage.py migrate
• Visit the app at http://127.0.0.1:8000/
• Use the following commands for running the tests:
  • Pytest tests (tests the functionality in the noise_sensors_monitoring package): docker-compose exec web coverage run --source=noise_sensors_monitoring -m pytest
  • Django tests: docker-compose exec web coverage run -a --source=users manage.py test
  • View coverage report docker-compose exec web coverage report -m

API documentation

This project includes an API that serves our public noise portal front end app.
The documentation for the API endpoints is in this repository's wiki, here.

Deployment Steps

• Install Heroku CLI.
• Connect to the noise-sensors-monitoring app on heroku.
• Run the command git push heroku main.
• Apply the migrations: heroku run python manage.py migrate.