IoT-Based Air Pollution Monitoring and Decision Support System

This project describes an IoT based system for monitoring air pollution and provide decision support in an urban setting of Vantaa, Finland. The overall purpose of the system is to monitor air quality continually, generate decisions based on the continuous flow of ‘raw’ environmental data (like traffic, weather etc.) to improve public safety and intelligent management of public areas.

The CPS (cyber-physical system) concept allows users to connect the digital system with the physical environment through a network of connected devices (sensors), by using wireless communication networks (internet) to communicate with remote cloud service for processing.

import pandas as pd import matplotlib.pyplot as plt import numpy as np import datetime

Run to view results

System Architecture (Important for marks)

The system consists of multiple CPS layers:

1. Physical Layer Includes the real-world environment such as air pollutants (PM2.5, PM10, NO₂), traffic, and weather conditions.

2. Sensing Layer Sensors continuously measure pollutant concentrations and send data at regular intervals (every 5–10 minutes).

3. Communication Layer Data is transmitted using IoT technologies such as 4G or LoRa networks, enabling reliable and scalable connectivity.

4. Data Processing Layer Data is stored and processed using cloud or edge computing. In this project, processing is simulated using Python.

5. Decision Layer A threshold-based algorithm classifies air quality into:

Normal Warning Alert

6. Acting Layer The system generates alerts to notify users, such as citizens or city authorities.

np.random.seed(42) samples = 144 # 24 hours, 10 min interval time = np.arange(samples) pm25 = np.random.normal(18, 4, samples) pm25[42:55] = pm25[42:55] + np.linspace(10, 24, 13) pm25[72:78] = pm25[72:78] + 38 pm25[96:110] = pm25[96:110] + np.linspace(8, 32, 14) pm25 = np.maximum(pm25, 1) pm10 = pm25 * 1.6 + np.random.normal(5, 5, samples) no2 = pm25 * 1.2 + np.random.normal(10, 6, samples) pm10 = np.maximum(pm10, 1) no2 = np.maximum(no2, 1) data = pd.DataFrame({ "Sample": time, "PM2.5": pm25, "PM10": pm10, "NO2": no2 }) data.head()

Run to view results

plt.figure(figsize=(14, 5)) plt.plot(data["Sample"], data["PM2.5"], color="black", label="PM2.5 signal") plt.title("PM2.5 Sensor Signal in Vantaa") plt.xlabel("Sample") plt.ylabel("PM2.5 concentration (µg/m³)") plt.grid() plt.legend() plt.show()

Run to view results

plt.figure(figsize=(14, 5)) plt.plot(data["Sample"], data["PM2.5"], label="PM2.5") plt.plot(data["Sample"], data["PM10"], label="PM10") plt.plot(data["Sample"], data["NO2"], label="NO2") plt.title("Simulated Air Pollution Measurements") plt.xlabel("Sample") plt.ylabel("Concentration") plt.grid() plt.legend() plt.show()

Run to view results

The x-axis is Sample number. You have 144 samples, which represents 24 hours of monitoring with a 10-minute sampling interval.

The y-axis is pollutant concentration.

There are three lines:

Blue line = PM2.5 Fine particles. This is your main decision variable.

Orange line = PM10 Larger dust particles. It is usually higher than PM2.5.

Green line = NO2 Traffic-related gas. It follows a similar pattern to PM2.5 and PM10.

Main observation

The graph shows that pollution is not constant. Most of the time the values stay at a normal/moderate level, but there are clear peaks around:

sample 45–55 sample 72–78 sample 100–110

These peaks represent possible traffic pollution events, street dust, or temporary weather-related pollution buildup.