WD: A Sliding Window based Time Series compression algorithm
Jingwen Meng, Liang Liu, Yulei Liu, Ning Wang · 2023
The fast generation speed and large volume of time series data make the efficient storage and transmission of time series challenging. In addition, time series is also characterized by instability, however, many advanced time series compression algorithms are not able to resist such data fluctuations, and thus failing to achieve efficient compression in such scenarios. Therefore, for the data scenarios where the metric values change frequently, this paper proposes a sliding window-based time series compression, Window Delta. It aims to weaken the data volatility by a sliding window, so that the impact of data fluctuation on the compression results can be effectively solved. In the experimental stage, we evaluated the compression performance of Window Delta and observed a significant improvement in compression ratio compared to the state-of-the-art methods, ranging from 7% to 98%.