I. Introduction
Subsampling methods are useful for saving computation and storage source for large sample datasets, see [15], [21], [22], [25]. Technological advances have enabled an extraordinary speed in data generation and collection in many scientific fields and practices, such as in astronomy, economics, and industrial problems. However the growth rate of the storage memory and the computational power is still far from sufficiently handling the explosion of modern data sets. The demand for extracting a small sample from a large amount of data arises routinely.