Enhanced Intrusion Detection in IoT Smart Homes: Leveraging Binary and Multi-Class Classification Models

Abstract

This study uses the MQTT-IoT-IDS2020 dataset, which contains normal traffic and attack traffic such as scan_A, scan_sU, Sparta, and mqtt_bruteforce attacks. This dataset is statistically extracted based on the unidirectional-based features packet header flow feature and has 19 features. This study used 10 best algorithms, namely ADABOST, eXtreme gradient boosting classifier (XGBC), stochastic gradient descent classifier (SGDC), random forest (RF), Naïve Bayes (NB), multi-layer perceptron classifier (MLPC), decision tree (DT), logistic regression (LR), linear discriminant analysis (LDA), and K-Nearest Neighbor (KNN) using binary class and multi-class. Using this classification algorithm, researchers measure the value of accuracy, precision, recall, F1 score, classification time, and receiver operating characteristic (ROC) curve to obtain the best classification algorithm. Measurement of accuracy value is done by dividing the dataset into 80:20 for training data and testing data, then validating the measurement of accuracy value with k-fold.