ESP-IPS Server

Purpose

This is an Indoor Positioning System that relies on the Bluetooth chips of the famous ESP32 microcontrollers. The idea is to use their capacity to send Bluetooth pings in broadcast mode while scanning for other pings, hence building an RSSI (Received Signal Strength Indicator) buffer. The server, hosted in this repository, then collects the data through MQTT and processes the lot.

It features two modes :

• Training : knowing the distance between each ESP32 (beacon and station included), and having a bunch of RSSI between each of them, a linearised model is trained allowing for future distance estimates
• Real Time Positioning : In this mode, beacons are usually placed on moving objects (e.g. : toy cars) and stations fixed at known positions. The server collects all RSSI data between the beacons and the stations, and using the previously trained model estimates every distance and tries to calculate an estimated position for every beacon. This last feature used an optimization algorithm (gradient-descent) to find the best estimate.

Unfortunately, the system never really worked properly. Probably since the antennas used (although straight) weren't made for this purpose, and for the RSSI surely isn't a precise enough metric.

I developped this project in classe de première, on the suggestion of my CS lab work teacher in NSI who gave me the opportunity to attend his BTS class' lab work.

Instructions

Config

To start it you will need a .env file like this

# Influx Client/Server Config
INFLUX_HOST= (Optional if no changes are made to docker-compose.yml)
INFLUX_TOKEN=
INFLUX_ORG=
INFLUX_BUCKET=
INFLUX_USERNAME=
INFLUX_PASSWORD=
MODE= (Optional if not in setup)

# MQTT Client Config
MQTT_PORT=
MQTT_HOST= (Optional if no changes are made to docker-compose.yml)

# API Configuration
API_HOST= 
API_PORT= (Optional if no changes are made to docker-compose.yml)
EXPOSE_PORT=

# General Config
RSSI_BUFFER_SIZE=
INIT_RSSI_BUFFER_SIZE=
RSSI_DISTANCE_ORDER=

Setup

For setup, you will need to add a MODE=setup entry to the env file Then you can start influxdb using docker-compose:

docker-compose up influx -d

Once it is started, you can stop it

docker-compose down

Running

Then you can start it using docker-compose :

docker-compose up -d

You're done ! The high precision IPS server is now started, and you can enjoy it using the GraphQL API.