Distributed cooperative HTTP Caching in Mobile Networks

Conference: Mobilkommunikation – Technologien und Anwendungen - 18. ITG-Fachtagung
05/15/2013 - 05/16/2013 at Osnabrück, Deutschland

Proceedings: Mobilkommunikation – Technologien und Anwendungen

Pages: 6Language: englishTyp: PDF

Personal VDE Members are entitled to a 10% discount on this title

Authors:
Drechsler, Chris; Windisch, Gerd (Chemnitz University of Technology, Chemnitz, Germany)

Abstract:
Mobile network operators are currently observing a tremendous increase of mobile data traffic due to the increased usage of smartphones or other Internet-enabled devices (e.g. tablets, laptops). Recent studies have shown that HTTP is the dominant protocol in transferring content to users and that multimedia content account for a major part of the transferred bytes. A natural solution for reducing the transmission costs and for improving the QoE is caching of frequently requested content. To achieve high (byte) cache hit rates a reliable detection mechanism of duplicated HTTP transfers and a reasonable large cache size is needed. To overcome this drawback we propose a distributed caching architecture where the clients perform the caching of frequently requested HTTP content by allowing a partial use of the user equipment’s memory for local caching (which is free of charge from the operator’s perspective). In this contribution we engage clients to share the content of their local caches (e.g. the browser cache) via smart proxies within the operator network in order to build a large distributed and cooperative cache. These smart proxies can be located at eNodeB sites so that users within the cell coverage area of the eNodeB can directly support each other in retrieving frequently requested HTTP content. We also introduce a new method for precisely identifying duplicate transferred content over HTTP independently of the used URL or of commonly applied mechanisms like personalization of HTTP messages (e.g. via cookies or parameters in the query string) and content negotiation. Beside this we show how caching can be achieved even in the presence of HTTP message personalization and how content producers keep full control over their content although it is cached by network operators or clients.