Kyung-sub Min

Probabilistic context prediction using time-inferred multiple pattern networks

We propose a probabilistic method for context prediction of mobile users based on their historic context data. The proposed method predicts general context based on the probability theory through a novel graphical data structure, which is a kind of weighted directed multi-graphs. User context data are transformed into the new graphical structure, in which each node represents a context or a combined context and each directed edge indicates a context transfer with the time weight inferred from corresponding time data. The periodic property of context data is also considered. We bring a nice solution to context data with such property. Through simulation, we could show the merits of the proposed method.

A multiple-file write scheme for improving write performance of small files in Fast File System

Fast File System (FFS) stores files to disk in separate disk writes, each of which incurs a disk positioning (seek + rotation) limiting the write performance for small files. We propose a new scheme called co-writing to accelerate small file writes in FFS without sacrificing its advantages. The scheme collects writes to small files in the same directory and stores them together to disk in a single large disk write. Co-writing multiple files in one disk I/O can reduce disk positioning times resulting from small file writes in FFS. Our experiments on OpenBSD show that the co-writing slashes the cost of small file writes in FFS by up to 33%, in both real-world and synthetic benchmarks.