Network-conscious GIF image transmission over the Internet

Traditional image compression techniques seek the smallest possible file si ze for a given level of image quality. By contrast, network-conscious image compression techniques take into account the fact that a compressed image will be transmitted over a packet-switched network that may lose and reorde r packets. We describe GIFNCa, a network-conscious revision of the popular GIF89a standard. As with GIF89a, GIFNCa compresses an image using LZW encod ing, however GIFNCa does so using an Application Level Framing approach. Th e data is segmented into path MTU-size data units, each of which can be ind ependently decompressed and displayed on its own. Under lossy network condi tions, when used in combination with an unordered transport protocol, GIFNC a permits faster progressive display at the receiver than GIF89a over an or dered transport protocol. This advantage comes in exchange for a small pena lty in overall compression. This paper defines GIFNCa, and presents prelimi nary experimental data concerning this tradeoff. The overall goal of this r esearch is to illustrate (1) the value of considering network characteristi cs in designing image formats, and (2) the value of unordered transport ser vice. (C) 1999 Elsevier Science B.V. All rights reserved.