Anomaly Detection Model for Smart Home using WSN

Abstract

The rapid advancement of home automation systems has seen the integration of various sensors newlineand devices to enhance security, comfort, and efficiency. However, the increasing complexity newlineand connectivity of these systems also introduce potential vulnerabilities and the need for newlineeffective anomaly detection mechanisms. This research focuses on the detection of anomalies newlinein home automation systems using a combination of ESP32 microcontrollers, LDR (Light newlineDependent Resistor), Gas and DHT11 sensors to generate data, which is then uploaded to newlineThingSpeak for further analysis. The primary objective is to leverage machine learning newlinealgorithms specifically fuzzy clustering, k-Nearest Neighbors (k-NN), and k-medoids to newlineanalyze the data, detect anomalies, and improve key performance metrics such as accuracy, newlinedetection rate, and throughput. Home automation systems have become increasingly popular newlinedue to their ability to provide enhanced security, convenience, and energy efficiency. These newlinesystems typically involve a network of interconnected sensors and devices that monitor various newlineenvironmental parameters and automate household functions. In this study, we utilize an array newlineof sensors including ESP32, LDR, Gas, and DHT11 to monitor different aspects of the home newlineenvironment. Data generated from these sensors are transmitted to ThingSpeak, an IoT newlineanalytics platform, for real-time monitoring and storage. Anomalies in home automation newlinesystems can indicate potential security breaches, malfunctions, or inefficiencies. Therefore, newlineaccurate and timely detection of these anomalies is crucial. Machine learning algorithms have newlineshown great promise in identifying patterns and deviations in large datasets. This research newlineapplies fuzzy clustering, k-NN, and k-medoids algorithms to the sensor data to detect anomalies newlineand improve overall system performance. Data from the ESP32 microcontroller, LDR, Gas, newlineand DHT11 sensors are collected at regular intervals and transmitted to ThingSpeak.