piPeCounter - A Raspberry Pi people counter

Real-time application for estimating the number of people in an area using Bluetooth and WiFi signals.

DEPRECATED

I recommend checking out this repo instead: https://github.com/IanMercer/pi-sniffer

Installation

The setup instructions are based on Raspbian Buster, as this project aims for a Raspberry Pi Zero W as platform.

Prerequisites

To run this application some prerequisites must be met.

Monitor mode

In order to be able to put the Raspberry Pi's onboard WiFi into monitor mode to scan the air, one need to replace the kernel driver shipped with Raspbian by the driver from the nexmon project. Unfortunately, one need to compile the driver by oneself. The compilation and installation process is well described in the project's README file. Follow all the given steps for chipset bcm43455c0 (not bcm43430a1) driver.

Upon success of the driver installation, set up the monitoring interface wlan0mon:

sudo iw phy $(iw dev wlan0 info | gawk '/wiphy/ {printf "phy" $2}') interface add wlan0mon type monitor
sudo ifconfig wlan0mon up

One can test ones effort by scanning the air at once:

sudo tcpdump -i wlan0mon

Dependencies

Initially, Pipenv is used for management of virtual python environments, so you don't mess up your system's python installation.

# For pipenv
sudo apt-get install python3-setuptools pipenv

Because this project takes advantage of the Bluetooth stack, a few development dependencies must be met. As they don't come precompiled, they need to be compiled. Please, be patient as this will take some time… See also Installation of pybluez on Raspberry Pi 3

# For pybluez
sudo apt-get install libbluetooth-dev python-dev libglib2.0-dev libboost-python-dev libboost-thread-dev

Clone the repository and install all dependencies by Pipenv for the server and the scanner, respectively. As this process will surely take a while… a timeout may occur. To get around that issue, increase the timeout (by setting the environment variable PIPENV_TIMEOUT to five minutes), which should be sufficient.

git clone https://github.com/AlexNaga/rpi-people-counter.git

export PIPENV_TIMEOUT=300   # Five minutes in seconds

# Install dependencies for the server
cd rpi-people-counter/server
pipenv install

# Install dependencies for the scanner
cd ../scanner
pipenv install

Data storage

MongoDB is used to store sensor data.

sudo apt-get install mongodb

After installation, make sure the database service (MongoDB) is up and running.

systemctl status mongodb.service

Optionally, start and enable the service, if it is not already done by the installation process.

sudo systemctl start mongodb.service
sudo systemctl enable mongodb.service

MQTT broker

Mosquitto is an open source message broker that implements the MQTT protocol.

sudo apt-get install mosquitto

After installation, make sure the service (Mosquitto MQTT) is up and running.

systemctl status mosquitto.service

Optionally, start and enable the service, if it is not already done by the installation process.

sudo systemctl start mosquitto.service
sudo systemctl enable mosquitto.service

Getting started

Running the server

1. Go to the server folder: cd rpi-people-counter/server
2. Make a copy of the configuration file: cp config/config.ini.example config/config.ini …and change the database credentials (if needed).
3. Then run:

pipenv run gunicorn run:gunicorn_app --workers=1 --worker-class gevent --bind localhost:8000

Running the scanner

The scan process is realized by the program tshark:

sudo apt-get install tshark

1. Go to the scanner folder: cd rpi-people-counter/scanner
2. Make a copy of the configuration file: cp config/config.ini.example config/config.ini
3. Then run:

pipenv run python main.py

Architecture

System overview

Scanner view

Server-side view

Client-side view