Publishing, storing and displaying IMU sensor readings from an Arduino Nano 33

(Areas de Aplicación del IoT - UIB - MUSI 2020-21)

This repo provides the code and configuration needed to:

• Connect an Arduino Nano 33 to a MQTT broker (mosquitto) with TLS broker authentication
• Read the Nano 33 IMU sensor values and publish them via MQTT
• Load the published values in an Influx database using a Telegraf agent
• Display the recorded values on a Grafana dashboard
• The Nano is programmed with three possible operating modes:
  • alltime: will publish IMU readings at the specified sample rate for as long as it is powered on
  • ifmov: will publish IMU readings at the specified sample rate only if movement is detected
  • sleep: won't publish IMU readings at all
• The operating mode can be changed remotely by publishing an MQTT message on the topic sensors\imu1\mode, with the payload being the name mode name (e.g. "alltime")

Screenshot of Grafana's dashboard:

Usage

Update the parameters in credentials.h and upload the imu_sensor.ino sketch to your Nano board (requires the WiFiNINA, PubSubClient and Arduino_LSM6DS3 libraries).

You will need properly configured instances of mosquitto (or any other mqtt broker), Telegraf, Influxdb and Grafana; running on a server connected to the Internet or to the Nano's local network. This repo was tested running the four services on a VM with a Raspbian OS**. The initial configuration section below assumes that you have already installed and set up the basic configuration for these services, and only provides details on the adjustments needed to make them work over TLS or for this specific application.

**Note that the four services don't neccesarily have to be on the same server as they communicate to each other over the network.

Nano operating logic

At the top of the imu_sensor.ino sketch the following parameters can be adjusted:

• pubTopic: topic where the IMU readings will be published
• subTopic: topic where the Nano will subscribe to receive instructions
• threshold: minimum gyroscope absolute value to consider that the Nano is moving, used when the Nano is operating in ifmove mode. Initial calibration showed that, even when stationary, the sensor would sometimes measure values in the -10 to +10 dps range, hence the current 10 threshold.
• sample_rate: published readings per second. Note that according to the Arduino_LSM6DS3 documentation, the sampling rate is 104 Hz so this is the maximum publishing rate that would make sense.

As opposed to other type of sensors, most of the IMU data use cases require a high sampling rate (e.g. infer a particular type of movement), which can present a challenge on the storing side. Hence, the ifmove operating mode, which will only send data when movement is detected.

Some additional clarifications on the sketch logic:

• The IMU readings are published as a single string in the Influx format, which allows the Telegraf to parse and load them directly into a measurements table in the Influx database.

• On ifmove mode, values are published during one second after the last value above the movement threshold is measured. This is to ensure that all the information of a given movement is captured.

• Given how the logic is coded, sending any instruction other than "alltime" or "ifmov" will make the Nano go into sleep mode.

Initial configuration

Generating the authentication certificate

We want the broker to authenticate in order to establish a connection with our Arduino Nano. We first need to create a Certified Authority (we could also use a public one). To do so, we will follow the following steps using the OpenSSL libray.

Generate a CA key and and a self-signed certificate (the root certificate):

openssl req -new -x509 -days 365 -extensions v3_ca -keyout ca-master.key -out ca-master.crt

Generate the broker's key:

openssl genrsa -out broker.key 2048

Generate the broker's certificate signature request (Important: when prompted to provide the certificate details, the Common Name field has to be the broker's server domain name or the servers IP address, depends on how you pass the server's address in credentials.h):

openssl req -out broker.csr -key broker.key -new

Sign the broker's CSR with the CA:

openssl x509 -req -in broker.csr -CA ca-master.crt -CAkey ca-master.key -CAcreateserial -out broker.crt -days 180

Configuring mosquitto for TLS

See the mosquitto.conf in this repo.

A copy of the root certificate, the broker's certificate and the broker's key must be stored in mosquitto's server. The config file mosquitto.conf must be updated to indicate the path to these files, as well as the TLS version to use.

Since we are doing broker authentication only (not the Nano), don't include the line require_certificate true. Otherwise the broker will reject the connection.

The current configuration expects the Nano to also provide a username and password, hence the allow_anonymous false. It also has an access control allowing only certain users to read and write on certain topics.

Upload the root certificate to the Nano board

Follow the steps on this tutorial, and when prompted to enter the website from where to fetch the SSL certificate, just add the IP and port of the mosquitto server (or the domain name if it has one).

Configuring Telegraf

See the telegraf.conf in this repo.

After creating an Influx database to store the IMU measurements, the influxDB credentials and the database name must be updated in telegraf.conf.

The MQTT credentials and details must also be provided. The Telegraf agent must subscribe to the topic where the Nano is publishing the IMU readings (in this example it is sensors/imu1/values/#). To connect over TLS the server address must start with ssl:// and the the path to the root certificate must be added.

Configuring Grafana

The only setting required in Grafana which is somewhat specific for this use case (apart from obviously creating the dashboard with the IMU data charts), is to change the min_refresh_interval setting from the default 5s to 100ms on grafana.ini.

An exact dashboard like the one I built can be created by importing the grafana_dashboard.json included in this repo.