Hybrid Approach For Anomaly Detection Using Clustering Mechanism

Abstract

This paper explores the design and implementation of an IoT-based home
automation system using the ESP32 microcontroller, integrated with DHT11,
LDR, and gas sensors. The primary objective is to collect environmental data such
as temperature, humidity, ambient light levels, and air quality, and transmit this
data to the ThingSpeak cloud platform for real-time monitoring and analysis. By
leveraging Wireless Sensor Networks (WSN), the data is fetched from ThingSpeak
and analyzed in MATLAB using advanced clustering algorithms, specifically
focusing on fuzzy clustering, k-medoids, and k-means, to detect anomalies with
high accuracy and superior detection rates. The ESP32 microcontroller, known for
its powerful processing capabilities and integrated Wi-Fi, serves as the system's
core. The DHT11 sensor monitors temperature and humidity, the gas sensor
detects various gases to ensure safety, and the LDR sensor measures ambient light
levels for energy-efficient lighting control. Data transmitted to ThingSpeak is
visualized in real-time and retrieved for further analysis in MATLAB. Fuzzy
clustering is emphasized for its ability to handle uncertainties and provide
nuanced anomaly detection by assigning membership levels to data points for
different clusters. K-medoids, robust to noise and outliers, uses actual data points
as cluster centers, while k-means, although sensitive to noise, is also employed for
partitioning data into clusters. The system's performance is evaluated based on
throughput, latency, and detection rate. High throughput ensures efficient data
processing, low latency allows near real-time insights, and a high detection rate
minimizes false positives and negatives. this project demonstrates significant
improvements over existing home automation systems, highlighting the potential
of IoT and advanced data analysis techniques in enhancing the functionality,
reliability, and safety of smart homes.