CarbonSens: Monitoring Key Atmospheric Pollutants and Climate-Active Gases using IoT, EO and AI

Carbonsens weaves IoT, AI, and Web into a vibrant platform for tracking Earth's air. Its sleek dashboard, AI chatbot, and 7-day gas forecasts empower users to monitor key atmospheric pollutants and climate-active gases like CO2, CH4, and more. Split into three GitHub repos—AI for analytics, IoT for sensors, and Web for UI/backend—it’s a bold step toward a greener future.

---

1. Project structure

├── AI            # ChatBot & AI forcasting
├── Documents     # Project documentation & images
├── IoT           # IoT system architecture and Arduino firmware  
├── Web           # Full-stack web application 
├── .gitignore    # specify which files git should ignore
├── README.md     # Project overview and setup instructions
└── CHANGELOG.md   # Version history and updates

1.1. AI

• ChatBot & AI forcasting
• Integrated in the web's backend : backend/API/views.py

1.2. IoT

• The iot/ folder includes the microcontroller code to:

  • Interface with sensors (BME680, MQ-4, MG811, TSL2591, GPS)
  • Package data using CayenneLPP
  • Send data over LoRaWAN using LMIC with OTAA

1.3. Web

Frontend

• Displays interactive dashboards, chatbot UI, and forecasting visuals.
• Contains every service display for user ensuring a good UI & UX

Backend

• Handles APIs (SignUp/SignIn, chatbot, forecasting), EO data processing, and IoT integration using Django REST Framework.
  • API : Contains AI forcasting and ChatBot functionalities
  • data : Contains data handling for EO
  • IoT : Contains scripts to link Iot device to the webapplication

1.4. Documents

• project-related documentation and images

---

2. End-to-End Solution Architecture

The system monitors gas concentrations ( CO, NO2, CH4, CO2...) using Earth Observation (EO) and IoT (Internet of Things) data, offering:

• Real-time dashboards for gas levels.
• AI-powered chatbot for analysis and queries.
• 7-day gas emission forecasts.
• AI Assistant

2.1. EO System

Earth Observation data, mainly from sentinel 5P imagery, is processed to compute gas concentrations for CO, CH4, NO2, O3 and SO2

2.2. IoT System

Physical sensors deployed in the field collect environmental and gas concentration data. The data is transmitted via LoRa to a LoRaWAN Stack in a virtual server in the cloud. For a detailed view of the IoT system and its components, see the IoT README

2.3. Web Application

Serves as the interface for users to visualize real-time sensor data, EO insights, and analytics. For a detailed view of the web application's internal architecture, see the Web README

---

3. Tech Stack

Layer / Component	Technology / Tool
Frontend	React.js
	Tailwind CSS
	Vite
Backend	Django
	Django REST Framework
	Object-Relational Mapping (ORM)
Docker	Dockerfile
	Docker+ MQTT Broker
	Docker+Cron
	Docker-compose
IoT Data Pipeline	Arduino C++
	LoRa / LoRaWAN, ChirpStack
	Devices : Arduino Mega 2560, BME680, MG-811, MQ-4, TSL 2591, LoRa/GPS Shield
	CayenneLPP
	MQTT (Mosquitto)
AI/ML	Python (scikit-learn RandomForest)
	Time series forecasting with feature engineering (sin/cos transforms)
	Groq API (LLM with custom prompt engineering for chatbot)

---

---

4. Setup

4.1. Clone the Repository

git clone https://github.com/zaydbf/GAIAthon.git
cd GAIAthon

4.2. Environment Configuration

Create & Configure .env with Copernicus, Groq API credentials and VM vAriables. (in Web/backend/.env)

Check .env.exemple for more details .env.exemple

4.3. Build and Run

cd Web
docker-compose build
docker-compose up

• Frontend: http://localhost:5173
• Backend: http://localhost:8000

---

5. Changelog

See CHANGELOG.md for detailed version history and updates.