Data Approximation for Time Series Data in Wireless Sensor Networks

Data prediction approaches are proposed in many fields to approximate time series data with a tolerable error. These approaches typically build prediction functions based on assumptions of the data variation. Nonetheless, if the variation of real-world time series data does not follow the assumption, the performance of data prediction will be limited. This paper presents a lightweight data approximation approach for time series data. This approach utilizes binary codes to represent original values, directly shortening their lengths in the cost of data precision. Then the author implements this approach in the WSN scenario. Two types of application layer messages and their transmission scheme are presented. These approaches are employed in WSN applications to: (1) report abnormal conditions in time, and (2) reduce data transmissions independently of data variations. Series of simulations are built on the basis of five real datasets. Simulation results based on five real datasets validate the performances of the proposed approach.