Content aware multimedia communications [Elektronische Ressource] / von Alexander Eichhorn

Content aware multimedia communications [Elektronische Ressource] / von Alexander Eichhorn

-

English
259 Pages
Read
Download
Downloading requires you to have access to the YouScribe library
Learn all about the services we offer

Description

Content-Aware MultimediaCommunicationsDissertationZur Erlangung des akademischen GradesDoktor-Ingenieur (Dr.-Ing.)vorgelegt der Fakultat fur Informatik und Automatisierung¨ ¨der Technischen Universit¨at IlmenauvonDipl.-Inf. Alexander EichhornEinreichung am 26. Juni 2007Disputation am 5. Dezember 2007GutachterProf. Dr.-Ing. habil Winfried Kuhnhauser, TU Ilmenau¨Prof. Klara Nahrstedt, University of Illinois at Urbana-ChampaignProf. Dr. Sc. Thomas Plagemann, University of Oslourn:nbn:de:gbv:ilm1-2007000398Content-Aware MultimediaCommunicationsDipl.-Inf. Alexander EichhornA dissertation submitted in partial fulfillmentof the requirements for the degree ofDoktor-Ingenieur (Dr.-Ing.)Faculty of Computer Science and AutomationTechnische Universit¨at IlmenauSubmitted at 26. June 2007Defended at 5. December 2007Committee in ChargeProf. Dr.-Ing. habil Winfried Ku¨hnhauser, TU IlmenauProf. Klara Nahrstedt, University of Illinois at Urbana-ChampaignProf. Dr. Sc. Thomas Plagemann, University of OsloAcknowledgementsThis work has its roots in the teaching, help, patience, and inspirationof a great number of people, to whom I wish to express my gratitude.First of all my deep thanks go to my family who supported me all thetimeswithaidandencouragingwords. Ihighlyappreciatetheirenduringbelieve in me to successfully finish this thesis. I would also like to thankDr. Ulrich Klettforconvincing me ofstarting to studycomputerscienceat all.

Subjects

Informations

Published by
Published 01 January 2007
Reads 30
Language English
Document size 6 MB
Report a problem

Content-Aware Multimedia
Communications
Dissertation
Zur Erlangung des akademischen Grades
Doktor-Ingenieur (Dr.-Ing.)
vorgelegt der Fakultat fur Informatik und Automatisierung¨ ¨
der Technischen Universit¨at Ilmenau
von
Dipl.-Inf. Alexander Eichhorn
Einreichung am 26. Juni 2007
Disputation am 5. Dezember 2007
Gutachter
Prof. Dr.-Ing. habil Winfried Kuhnhauser, TU Ilmenau¨
Prof. Klara Nahrstedt, University of Illinois at Urbana-Champaign
Prof. Dr. Sc. Thomas Plagemann, University of Oslo
urn:nbn:de:gbv:ilm1-2007000398Content-Aware Multimedia
Communications
Dipl.-Inf. Alexander Eichhorn
A dissertation submitted in partial fulfillment
of the requirements for the degree of
Doktor-Ingenieur (Dr.-Ing.)
Faculty of Computer Science and Automation
Technische Universit¨at Ilmenau
Submitted at 26. June 2007
Defended at 5. December 2007
Committee in Charge
Prof. Dr.-Ing. habil Winfried Ku¨hnhauser, TU Ilmenau
Prof. Klara Nahrstedt, University of Illinois at Urbana-Champaign
Prof. Dr. Sc. Thomas Plagemann, University of OsloAcknowledgements
This work has its roots in the teaching, help, patience, and inspiration
of a great number of people, to whom I wish to express my gratitude.
First of all my deep thanks go to my family who supported me all the
timeswithaidandencouragingwords. Ihighlyappreciatetheirenduring
believe in me to successfully finish this thesis. I would also like to thank
Dr. Ulrich Klettforconvincing me ofstarting to studycomputerscience
at all. Your words and thinking of me encouraged me a lot and I will
always keep your advises.
I am deeply indebted to my supervisor Prof. Winfried Ku¨hnhauser
for offering me an opportunity to work with the Distributed Systems
Group at TU Ilmenau. I would like to thank you for the lasting sup-
portthroughoutmystudies,alltheguiding,discussionsandcooperation.
Even if the duration has been longer than originally planned, I appreci-
ate the belief you have expressed in my ability to finish the work and for
providing me with the required funding.
A lot of people have contributed to this work in various ways. Many
of the ideas presented herein evolved during discussions with my friends,
colleagues and students. I would particularly like to thank my friend
Andreas Franck for the numerous enlightening discussions during the
last years and for his detailed, insightful comments on nearly every page
of this work. I would also like to thank my friends and colleagues Dr.
Thorsten Strufe, Dr. Carsten Behn and Mario Holbe for their much
appreciated ideas, support, and feedback as well as Katja Wolf for their
great administrative support. In addition, I would like to thank all
colleagues in the Institute of Computer Science at TU Ilmenau for the
pleasant working environment.
Writing this thesis would not have been possible without the efforts of
many friends and students who contributed to the development of the
Nojaplatform. SpecialthanksgotoCarstenKonig, ChristianBrien, Se-¨
bastian Kuhn, Alexander Hans, Christoph Bohme, Alexander Senier,¨ ¨
Michael Roßberg, Melanie Friedrich, Holger M¨oller, Florian Kriener,
Jens D¨onhoff, Tino Jungebloud, Ives Steglich, and the numerous other
people who worked for the Noja project.
vvi
Finally, I like to express my deep gratitude to all my friends, espe-
cially my flat-mates Sebastian Daume, Klemens Gob¨ el, Stefan Hertwig,
Janosch Lax, Jana Kludas, Alexander Schultz, Carsten Riesel, Sandy
Stehr and Oliver Hitzegrad, all friends who lived in the K4, and all the
others I forgot to mention. Although I was often busy and non-present,
neither physically nor mentally, I highly appreciate the mental support I
received from all of you. I am truly glad that all of you have influenced
my way of living and thinking.Abstract
Thedemandsforfast,economicandreliabledisseminationofmultimedia
information are steadily growing within our society. While people and
economyincreasinglyrelyoncommunicationtechnologies, engineersstill
struggle with their growing complexity.
Complexity in multimedia delivery originates from several sources.
The most prominent is the unreliability of packet networks like the In-
ternet. Recent advances in scheduling and error control mechanisms
for streaming protocols have shown that the quality and robustness of
multimedia delivery can be improved significantly when protocols are
aware of the content they deliver. However, the proposed mechanisms
requireclosecooperationbetweentransportsystemsandapplicationlay-
ers which increases the overall system complexity. Current approaches
also require expensive metrics and focus on special encoding formats
only. A general and efficient model is missing so far.
This thesis presents efficient and format-independent solutions to sup-
port cross-layer coordination in system architectures. In particular, the
firstcontributionofthisworkisagenericdependencymodelthatenables
transport layers to access content-specific properties of media streams,
such as dependencies between data units and their importance. The
second contribution is the design of a programming model for stream-
ingcommunicationanditsimplementationasamiddlewarearchitecture.
The programming model hides the complexity of protocol stacks behind
simple programming abstractions, but exposes cross-layer control and
monitoring options to application programmers. For example, our in-
terfaces allow programmers to choose appropriate failure semantics at
design time while they can refine error protection and visibility of low-
level errors at run-time.
Based on examples we show how our middleware simplifies the inte-
gration of stream-based communication into application architectures.
An important result of this work is that despite cross-layer cooperation,
neither application nor transport protocol designers experience an in-
crease in complexity. Application programmers can even reuse existing
streaming protocols which effectively increases system robustness.
viiKurzfassung
Der Bedarf unsere Gesellschaft nach kostengunstiger und zuverlassiger¨ ¨
Kommunikation wachst stetig. Wahrend wir uns selbst immer mehr von¨ ¨
modernen Kommunikationstechnologien abhangig machen, mussen die¨ ¨
Ingenieure dieser Technologien sowohl den Bedarf nach schneller Ein-
fuhrung neuer Produkte befriedigen als auch die wachsende Komplexi-¨
¨tat der Systeme beherrschen. Gerade die Ubertragung multimedialer¨
Inhalte wie Video und Audiodaten ist nicht trivial. Einer der prominen-
testen Grunde dafur ist die Unzuverlassigkeit heutiger Netzwerke, wie¨ ¨ ¨
z.B. dem Internet. Paketverluste und schwankende Laufzeiten kon¨ nen
die Darstellungsqualit¨at massiv beeintrac¨ htigen. Wie ju¨ngste Entwick-
lungen im Bereich der Streaming-Protokolle zeigen, sindjedoch Qualitat¨
¨und Robustheit der Ubertragung effizient kontrollierbar, wenn Strea-
mingprotokolleInformationenu¨berdenInhaltdertransportiertenDaten
ausnutzen.
ExistierendeAnsat¨ ze,diedenInhaltvonMultimediadatenstromen¨ be-
schreiben, sind allerdings meist auf einzelne Kompressionsverfahren spe-
zialisiert und verwenden berechnungsintensive Metriken. Das reduziert
ihren praktischen Nutzen deutlich. AuSSerdem erfordert der Informati-
onsaustausch eine enge Kooperation zwischen Applikationen und Trans-
portschichten. DaallerdingsdieSchnittstellenaktuellerSystemarchitek-
turen nicht darauf vorbereitet sind, mussen entweder die Schnittstellen¨
erweitert oder alternative Architekturkonzepte geschaffen werden. Die
Gefahr beider Varianten ist jedoch, dass sich die Komplexitat eines Sys-¨
tems dadurch weiter erhohen kann.¨
Das zentrale Ziel dieser Dissertation ist es deshalb, schichtenubergrei-¨
fende Koordination bei gleichzeitiger Reduzierung der Komplexitat zu¨
erreichen. Hier leistet die Arbeit zwei Betrage zum aktuellen Stand der¨
Forschung. ErstensdefiniertsieeinuniversellesModellzurBeschreibung
vonInhaltsattributen,wieWichtigkeitenundAbhangigkeitsbeziehungen¨
innerhalb eines Datenstroms. Transportschichten konnen dieses Wissen¨
zur effizienten Fehlerkontrolle verwenden. Zweitens beschreibt die Ar-
beit das Noja Programmiermodell fu¨r multimediale Middleware. Noja
¨definiert Abstraktionen zur Ubertragung und Kontrolle multimedialer
ixx
Strome,¨ die die Koordination von Streamingprotokollen mit Applikatio-
nen ermogl¨ ichen. Zum Beispiel k¨onnen Programmierer geeignete Fehler-
semantiken und Kommunikationstopologien auswah¨ len und den konkre-
ten Fehlerschutz dann zur Laufzeit verfeinern und kontrollieren.