Missing Data Problem in the Monitoring System: A Review

Abstract

Missing data is a common phenomenon in sensor networks, especially in the large-scale monitoring system. It can be affected by various kinds of reasons. Moreover, incomplete data or information may affect the subsequent data processing and reasoning, resulting in a wrong decision. Hence, missing data recovery has always been a hot topic in the literature. However, although researchers have developed many different methods to recover missing data, this problem is still far away from being solved. To better track research progress and identify potential challenges, in this paper, we give a detailed review in the context of large-scale monitoring system. Mainly, we first introduce the basic concept of missing data, including the definition, causes, types, and performance evaluation. Then, a series of traditional and classical missing data recovery methods are analyzed and compared where their characteristics and scope of application are given. Furthermore, we present two current mainstream approaches from methodology to the existing literature, which are data recovery based on data mining algorithms and low rank algorithms, respectively. Finally, we conclude this paper with several promising directions for future research.