Machine Learning and Internet of Things for Smart Living: A Comprehensive Review and Analysis

The role of machine learning is to provide intelligent processing, analysis, and prediction for the data being used in the smart Internet of Things (IoT) applications to enhance quality and efficiency. This study aims to analyze the taxonomy of machine learning algorithms used in the specific type of IoT smart living applications. This chapter demonstrates an analysis of the data extracted from the 52 peer-reviewed scientific publications describing IoT observations on machine learning algorithms (clustering, classification, regression), optimization metrics, data size, data type (text, audio, video, image), data collection method, data processing location (cloud, fog, edge), and routing protocol (hierarchical clustering, point-to-point, multi-hop routing). The results show that most of the studies utilized cloud computing as their data processing location in IoT (Smart City: 55.88% and Smart Health: 60.61%) over Fog computing environments (Smart City: 11.76% and Smart Health: 15.15%). The chapter has shown that the classification is more used (Smart City: 45.46%, and Smart Health: 70.37%) than clustering and regression techniques in the IoT smart living applications. Furthermore, this work shows that among the four types of data formats (text, audio, video, and image) have been used in IoT applications, text data in Smart City (66.67%) and Smart Health (61.11%) was the primary format that has been used. Future work should focus on the adaptation of fog computing while applying machine learning for different smart applications, which will lead to a decrease in network traffic and optimized bandwidth.