Part of my thesis
If you want to use or build on this project, and or have any questions at all, don’t hesitate to contact me.
Personal Area Networks applications on sports
The object of this dissertation is the study of the architecture of Body Area Networks (BAN) and Personal Area Networks (PAN), and their applications, with the aim of optimizing the performance of athletes, by enhancing personalized training practices. In the context of the thesis, subjects that will be studied include the collection of data for the recognition of the physical posture of the athlete’s body, as well as the design of smart IoT services with an emphasis on sports. Finally, an “e-sport” application will be designed and implemented which will collect and analyze data on the athlete’s body position in real-time.
Ichnaea was the goddess of tracing and tracking.
Ichnaea is an IoT solution that collects and analyzes body position data from an athlete in real-time. It is the “e-sport” application described in the thesis section. Although the specifications of the app are set, I plan to keep building on it as I explore my interests on the web front.
Clone the project:
git clone [email protected]:xrazis/ichnaea.git
First, make sure you have the docker service and docker-compose installed, then:
docker-compose -f ./docker-compose.yml up –d –build
In a few moments, Ichnaea will be up and running! Visit localhost:8080 to view the web dashboard.
I develop on Jetbrains products, so the run scripts should be available once you launch Webstorm.
If you use your host machine for a getaway just connect the Arduino to any USB port, and deploy the docker-compose file. Alternatively, install docker to a Raspberry, connect the Arduino, and run the client container standalone. Be sure to enter the correct IP in the client config so a socket connection can be established with the server. Find out more about the arduino setup down bellow, at the client section.
The client can either run on the host machine or on a Raspberry Pi that acts like a getaway for the Arduino. You can use an IMU connected to any compatible microcontroller with Johnny-five, see here on how to connect it.
The client does some basic calculations with the help of Johnny-five, then the data is then streamed to the server and
subsequently to the frontend. The frontend does the final calculations that are needed for the model visualization. This
way we avoid making any 'heavy' computations on the client device, thus allowing for small and power efficient getaways
like a Pi Zero with multiple microcontrollers attached on it. Implementation specific details can be found in the
thesis itself.
During the development process I used my host machine as a getaway, and an Arduino UNO with an Invesense MPU6050 on a HAT. Although this is a cumbersome solution, it proved a quick way to bootstrap a working solution and avoid soldering. Ideally I would use an Arduino Nano with a LoRa adapter in order to have a wireless connection to the getaway, maybe a future improvement!
To begin with, deployment is handled by docker and docker-compose. Abstracting the configuration to a handy .yaml
file, while keeping the dev environment the same across machines makes deployment a breeze. Running the project will
create the following services:
- Three Databases, each for its distinct purpose:
- Redis, used for caching and pub-sub.
- Influxdb, used to store the bulk of data produced from the client devices (athletes).
- A Grafana instance, attached to the influx cluster.
- MongoDB, used as the user store.
- Backend, a Node.js server that is responsible for data and user storage, client and SPA socket connection, and exposing an API.
- Frontend, an SPA built on Vue with Typescript.
- Client, a Node.js app that collects and sends the data to the backend.
The following services and respective ports are exposed:
| Service | Port | Usage |
|---|---|---|
| Influxdb | {ip}:8086 |
Web UI |
| Grafana | {ip}:3000 |
Web UI |
| MongoDB | {ip}:27017 |
Shell |
| Backend | {ip}:8000 |
Socket and API |
| Backend | {ip}:9229 |
Debugger |
| Frontend | {ip}:8080 |
Web UI |
The {ip} should be localhost if running locally, or the ip of the server you are running the services on (obv).
The following routes are exposed by the backend.
If the user exists in the DB, a new session is initiated.
Registers a user with the given params and initiates a new session.
Logouts current user.
Returns current user.
Updates a user matching the given _id.
Returns an array of athletes objects.
Returns an athlete object for a given _id.
Updates an athlete matching the given _id.
Deletes an athlete matching the given _id.
Returns all data in the specified time range.
Returns all data for a given user _id, in the specified time range.
-
When the docker-compose is up and running, and you observe the backend container output you will notice that a connection cannot be established with influxdb. That's because the influx container is not ready yet. Don't confuse that with the
depends_onoption. As described in the docker docs :However, for startup Compose does not wait until a container is “ready” (whatever that means for your particular application) - only until it’s running.
TL;DR waiting a few seconds for all services to become available will solve this, and data will be written to influx.
- Euler angles on Wiki / Mathworld
- Gimbal-lock on Wiki
- Quaternion on Wiki / Mathworld
- Interactive video to visualize quaternions
- Really nice explanation from Stanford's virtual reality class
- What is a complementary filter