SmartGuardML: ML-based MQTT Data Analysis Approach for Threat Prediction in Smart Homes

Smart home devices, geared with sensors and connected over the network, have modernized everyday life by enhancing efficiency, convenience, and comfort. Sensors and Internet-Of-Things (IoT), integrated into smart home devices help automate tasks and thereby, reduce human intervention. However, a surge in the usage of these smart home devices has led to new cyber-security challenges and vulnerabilities such as unusual network traffic, data privacy, potential tampering of devices, and unauthorized access to data. Thus, retaining the integrity and confidentiality of data has become essential. To mitigate these risks and potential attacks, an MQTT data analysis-SmartGuardML is proposed, wherein ML-based Classification techniques are leveraged, involving Random Forest(RF), Support Vector Machine (SVM),, Decision Tree(DT), Naive Bayes(NB) and Logistic Regression(LR) and cyber attacks namely Denial-of-Service (DoS), Flooding, SlowITe, Bruteforce and Malformed data are detected. The performance evaluation of SmartGuardML is scrutinized through various metrics such as accuracy, precision, recall, and f1-score. Along with these metrics, models are also evaluated using receiver operating characteristic (ROC) graphs, Precision-Recall(PR) graphs and confusion matrix to visually analyze the model’s performance. This approach SmartGuardML enhances cybersecurity in smart home environments by ensuring robust protection against unauthorized access and potential threats, thus safeguarding sensitive data and ensuring the privacy and safety of residents.