afjcjsbx / weatherly

Weatherly

Weather forecasting machine learning with IoT devices and fog computing

Weatherly is a distributed weather forecast application that uses Recurrent neural network(RNN) on fog nodes and iot sensors for temperature monitoring.

To run a Fog Node:

Build container image running the build.sh script or use the following command:

docker build -t fog-node:latest .

Run the Fog Node container with the following command:

docker run -e CITY=Rome -e BROKER_ADDRESS=192.168.1.2 -e BROKER_PORT=1883 -e BROKER_USERNAME=afjcjsbx
-e BROKER_PASSWORD=password -e INFLUXDB_ADDRESS=1.2.3.4 -e INFLUXDB_PORT=8086 -e INFLUXDB_DATABASE=home 
-e INFLUXDB_USERNAME=afjcjsbx -e INFLUXDB_PASSWORD=password -p 8086:8086 -p 1883:1883 fog-node

or use docker compose:

version: '3.6'

services:

   fog-node-1:
       image: fog-node:latest
       container_name: fog-node-1

       restart: always
       ports:
           - "8086:8086"
           - "1883:1883"
       environment:
           - CITY=Rome
           - BROKER_ADDRESS=broker.emqx.io
           - BROKER_PORT=1883
           - BROKER_USERNAME=username
           - BROKER_PASSWORD=password
           - INFLUXDB_ADDRESS=1.2.3.4
           - INFLUXDB_PORT=8086
           - INFLUXDB_DATABASE=home
           - INFLUXDB_USERNAME=username
           - INFLUXDB_PASSWORD=password

To run a sensor:

Build Jar from iot-sensor-mqtt module and run:

java -cp "iot-sensor-mqtt-1.0-SNAPSHOT-jar-with-dependencies.jar" com.afjcjsbx.iotsensor.SensorPython

Arguments

Parameter	Example	Description
--clientId	"weather-sensor-1-name"	Client ID to use when connecting
--lat	41.89	Weather locality latitude in coordinates
--lon	13.59	Weather locality longitude in coordinates
-e or --endpoint	"broker.emqx.io"	Local fog node ip or Mqtt cloud service
-apikey or --apikey	52cb6acs8ejvn42hc4b92bffc187beac	openweathermap.org private API key
-t or --topic	/weather/Rome	Topic name to publish data

Libs

Java libs used in sensor:

• Apache HttpComponents
• Project Lombok
• gson

Python libs used:

• influxdb-python
• numpy
• pandas
• TensorFlow
• Keras
• paho-mqtt