Practical Implementation of M4 for Web Visulization Service
"Vast volumes of time series are becoming more common
in part to recent research advancements in big data analytics and
sensor/monitoring networks. However, tranmission of time series
for visualization services can cause serious bandwidth wastage and
extensive network delays if there is no efficient data management.
Existing work such as M4 aim to solve this problem by providing
high data reduction rates through data aggregation and guaranteeing
the reliability of visualization results at the same time. However,
current work on M4 does not consider verification in a more
practical environment; for example experimentation on web-based
servers that are openly accessible by users. In this paper, we propose
inter-pixel gradient-based M4 (IGM4) for enhancing the existing
M4 scheme, and conduct a study to reduce the amount of data
and delay without distorting the results of visualized graph. We
build user-friendly web-based system to dealing with the data processing
technique, and perform test on the empirical environment.
Finally, we present the results of performance evaluation through
comparison among the original data, M4, and IGM4 reflecting the
various kinds of time series and resolutions."