Analysis of the Characteristics of Agricultural Sensor Data

• 1、College of Informatics, Huazhong Agricultural University

Abstract：The integrity and availability of sensor data are essential for the development of Agricultural Internet of Things. In the agricultural senarios, the missing sensor data caused by the outdoor harsh environment can dramatically affect the integrity and availability of sensor data, so it is necessary to estimate the missing sensor data. To design the efficient missing sensor data estimation algorithms, the characteristics of agricultural sensor data should be firstly analyzed. Firstly, different types of sensor data are respectively transformed from time-space domain to DCT (Discrete Cosin Transformation) domian where the characteristics of these data are analyzed. Secondly, the relationships between different types of sensed object are described through claculating the correlation coefficients of different types of sensor data. The results show that the correlation of the same type of sensor data samples is more obvious in DCT domain than in time-space domain, and the strength of correlation between different types of sensor data is different. Consequently, the correlation of the same type of sensor data samples and the correlation between different types of sensor data can be used to design missing sensor data estimation algorithm.

Keywords： Agricultural Mechanization Engineering Agricultural Internet of Things Sensor Data Analysis Discrete Cosin Transformation Correlation Coefficient