Survey on IoT Data Preprocessing

Abstract

Internet of Things (IoT) is a growing technology in all fields of science and engineering. The amount of data emitted by the sensors used in the various fields is high. Therefore, efficient knowledge from such large datasets is a clear requirement of many users. This large data is far from perfect; it has many defects (such as noise, missing values, outliers etc.) and is not suitable for analysis because it can lead to incorrect conclusions. So, data preprocessing is a required technique for such data. Data preprocessing is an important and essential step, the main goal of which is to dedicate techniques to clean, refine, repair and improve that raw data. This paper proposes a survey on IoT data preprocessing and its techniques. This paper discusses exiting research on data preprocessing in the IoT context and, introduces the background of IoT data preprocessing and present literature reviews of the advanced research on data preprocessing techniques. The classification of various preprocessing approaches with techniques is clearly depicted in the figure. Various approaches of preprocessing cleaning, transformation, reduction and integration are described. In addition, methods for such approaches in IoT data preprocessing are also discussed. IoT Data preprocessing techniques on various applications are tabulated. Finally, issues and challenges, most useful in future work, are discussed.