| Summary: | 碩士 === 長庚大學 === 電機工程研究所 === 89 === This paper presents a hierarchical sampling selection (HSS) algorithm to select k-th smallest item in large data, distributing in the storages in the Internet. It is hard to sort the whole data in a single station because it wastes the finite memory and bandwidth in the Internet. The main idea of HSS algorithm uses two ways to reduce the memory and bandwidth cost on the selection in the Internet. One way is sampling the data in the storages to reduce the amounts of data transmitted in the Internet. The other way uses the hierarchical routing characteristic [1,2,3] of the Internet to reduce the cost. No matter the total data amount, the transmission of data decrease to the most 2 × N / M－1 blocks. The length of the block can adjust according to the maximum transmission unit in the hierarchical network. The HSS algorithm is the best method to selection k-th smallest item in large data in the actual network environment.
|
|---|
