An Efficient Cloud Storage Model for GOP-Level Video Deduplication using Adaptive GOP Structure

Abstract

Cloud storage systems may be used by users and businesses to transfer their enormous volumes of data for storage, processing, and analysis. Cloud storage may be used efficiently by employing the deduplication technique to prevent duplicate copies. Users may save any type of material, including audio, video, images, text, and more. The handling of deduplication techniques for these distinct data types may vary depending on their characteristics, sizes, etc. Video deduplication is one of these, and it’s essential for reducing memory waste in the cloud storage system. There are many levels of video deduplication that may be done. In this research work, we proposed a GOP-level deduplication system using an adaptive GOP structure. Since GOP is the level of deduplication, it is inefficient to build GOP with a fixed size. It can fail to discover a pair of identical frames. The adaptive GOP structure will yield GOPs with closer relation among their frames than with a fixed-size GOP structure. The proposed technique is compared with the fixed-size GOP structure with the GOP sizes of 8, 10, 12, and 15 and the proposed technique achieved a 2.18% PSNR gain which is relatively higher than other fixed-size GOP structures.