La présentation est en train de télécharger. S'il vous plaît, attendez

La présentation est en train de télécharger. S'il vous plaît, attendez

Présenté par : AHMED Toufik Université de Versailles - France

Présentations similaires

Présentation au sujet: "Présenté par : AHMED Toufik Université de Versailles - France"— Transcription de la présentation:

1 Présenté par : AHMED Toufik Université de Versailles - France
DVMA : Un Algorithme de Marquage pour les Flux MPEG4 dans un Réseau DiffServ Auteurs: Toufik AHMED, Guillaume BURIDANT, Ahmed MEHAOUA Présenté par : AHMED Toufik CNRS-PRiSM Lab. Université de Versailles - France

2 Résumé Contexte et Motivations Architecture de MPEG-4 Contributions :
Transport de MPEG-4 sur RTP/IP DVMA : Algorithme de marquage de flux MPEG4 Evaluation de performance Conclusion

3 Contexte Conditions pour une application multimédia dans les réseaux :
Besoin de QoS de bout en bout Optimisation des resources réseaux Protocoles TCP/IP : RTP, RTCP, et RTSP Modèles de gestion de la QoS sur IP : IntServ et DiffServ

4 Objectifs Transport de flux vidéo MPEG 4 sur réseaux IP :
Profile d’encapsulation RTP (Couche Transport) Mécanisme de marquage DiffServ (Couche Réseau) Signalisation OSI MPEG-4 DMIF / IETF SIP (Couche Session)

5 Architecture de MPEG-4 ... ... Compression Layer Sync Layer IP ATM
Interactive Audiovisual Scene Composition and Rendering SL ... Primitive AV Objects Description Information Object Descriptor ISO/IEC Visual ISO/IEC Audio Compression Layer Upchannel AU Sync Layer ISO/IEC Systems ISO/IEC DMIF Delivery Layer IP ATM MPEG-4 is an emerging digital multimedia standard with associated protocols for representing, manipulating and transporting natural and synthetic multimedia content such ad audio, video and data over a broad range of communication infrastructures including IP and ATM. The characteristic of MPEG-4 is to provide an object-oriented representation of multimedia content for the support of new ways of communication, access, and interaction with digital audiovisual data, and offering a common technical solution to various problems such as telecommunications, broadcast, interactive services and essentially VoD. MPEG-4 terminal architecture is composed of three layers: Compression Layer, Sync Layer and Delivery Layer. The Compression layer generates representation of content data called ES (Elementary Sreams), the hierarchical relations, locations and properties of ESs in a representation are described by dynamic set of ODs (Object Descriptors). ODs are themselves conveyed through one or more ESs. MPEG-4 scene is described by Scene Description (SD) information that addresses the organization of audiovisual objects within a scene, in terms of both spatial and temporal location. Scene Description is performed using a Virtual reality language (VRML) called BIFS (Binary Format for Scene). ESs is composed of AUs (Access Units). An AU is the smallest data entity to which timing information can be assigned. As an examples, A valid MPEG-4 Elementary Stream could be an MPEG-1 video, labeled with MPEG-4 system information in its headers, therefore, an AU would then be one video frame I, B or P labeled with priority information, timing information, and others. ...

6 Transport de MPEG-4 sur IP
Compression Layer Elementary Stream SL Layer SL Layer + RTP extension Standart MPEG-4 Approche Alternative Delivery Layer RTP P lug-in UDP IP RTP Fragmentation rules 1 2 3 : Approach [8] Approach [9] : Our Approach RTP/UDP/IP 1-2 Approche [ATT] Approche [ [INRIA/Microsoft] Approche [NTT] 4 : Notre approche 4

7 DVMA : Algorithme de marquage de paquets vidéo sur DiffServ (audio & OD)
if stream is “audio stream” then (application of property 2) if coder rate is “low rate” then DSCP=AF Low Drop Prec //example AF11 if coder rate is “medium rate” then DSCP=AF Medium Drop Prec //example AF12 if coder rate is “high rate” then DSCP=AF High Drop Prec //example AF13 if stream is “objects descriptor, scene descriptor” (application of property 4 ) then DSCP = EF //descriptions streams are significant, no loss //it’s necessary that these streams will be available //as soon as possible in MPEG-4 player to interpret //correctly the scene Policing Audio AF1x (a) Property 1: Making distinguish between important information and less important one such as Video, Audio, OD and BIFS. In case of congestion, the network will be relieved by eliminating the less important information, This property led us to distinguish between marking of video, audio and OD in Diffserv. (b) Property 2: Stream that accept losses (Video/Audio) will be transmitted using AF PHB. Currently, AF defines 4 classes within each class 3 degrees of drop precedence according to the importance of the packet. We propose to map Video into one class and Audio to another class and so on. (c) Property 3: MPEG-4 uses hierarchical coding for resolving scalability and heterogeneity issues. Different coding modes exist: Space Scalability, Temporal Scalability SNR Scalability. We must make distinguish between main video stream (base layer) and improvement stream (enhancement layer) (d)  Property 4: Stream such as: Scene description and Media objects description are very sensitive and need to be preserved from errors and loss during transmission. We propose to transmits these streams over TCP and if possible with EF PHB! (2)               DVMA Algorithm We assume that audio and video are carried with two distinct AF classes. This enables network nodes to better cope with congestion by given some routing privilege to audio compared to video. However, a problem of media synchronization appears. The solution is to synchronize the audio and video at destination using RTP Timestamps. OD BIFS EF /AF ?

8 DVMA : Algorithme de marquage de paquets vidéo sur DiffServ (Video)
Policing if stream is “video stream” (application of property 3) if “base layer video stream” (level 1 = mimimum QoS) then DSCP = AF low Drop Prec //example AF21 if “ enhanced layer video stream 1” (level 2 = medium QoS) then DSCP = AF Medium Drop Prec //example AF22 if “enhanced layer video stream 2” (level 3 = maximum QoS) //example AF23 minQoS Video medQoS AF2x maxQoS

9 Performance Evaluation - Testbed -
Our IP Diffserv network testbed is composed of a video server application that plays MPEG-4 video stream for many heterogeneous clients. The server and clients are connected through TCP/IP network. We exploit some test-scenario on the transmission process. We quantify QoS measurement and we compare between transmission process in Best Effort service and in Differentiated Service.

10 Architecture des routeurs IP Diffserv
Edge routers It accept traffic into the network. They can characterize, police, and/or mark customer traffic between other edge or core routers. Within the Diffserv domain, service association is performed according to the DSCP value in each packet’s IP header. Therefore, the video application must mark the packets correctly to obtain a particular level of service within the Diffserv region. Core routers They are configured to perform packet classification based on DSCP, packet scheduling, (3) queue management, (4) policing and (5) packet dropping.

11 Signalisation MPEG-4 DMIF
We developed an MPEG-4 interactive video on demand system consisting of a video server and a multimedia client communicating over the previously described IP Diffserv network testbed as illustrated in Figure 4. The client and server code is based on Java Media Framework developed by Sun Microsystems The MPEG-4 Server consists of MPEG-4 pump, DMIF Instance for IP network, and tools for RTP/UDP/IP stack.

12 Signalisation DMIF et QoS
Paramètres de QoS spécifiés par le propriétaire du service MPEG-7 peut être utilisé pour automatiser cette fonction Paramètres de QoS portés par les primitives DA_channelAdd() comme QoSDescriptor Ports RTP/IP assignés à chaque flux (session) PHB assigné à chaque session.

13 Signalisation DMIF phase 1: Initiation de la Session

14 Signalisation DMIF phase 2: Ajout de canaux

15 Codage vidéo MPEG-4 hiérarchique
(1)               MPEG-4 Video In our simulation, the MPEG-4 sequence is obtained by using a set of multimedia components such as audio and video Elementary Streams ESs. Video ES is from the highly bursty MPEG-1 Star Wars movie [20]. Voice, Object Description, Scene Description and Data ESs were generated according to an exponential distribution. Three video streams were generated that are associated with: ·        The MPEG-4 base layer video stream that offers a minimum presentation quality (i.e. minQoS), ·        The MPEG-4 enhanced layer 1 video stream, that improves minQoS to a better quality (i.e. MedQoS) ·        The MPEG-4 enhanced layer 2 video stream, that improves medQoS to the maximum quality (MaxQoS).

16 Variation des pertes de paquets en fonction de la charge
Diffserv Best Effort

17 Pertes de paquets : Scénario IP Best Effort
Load 95 % Load 80 %

18 Perte de paquets : Scénario IP DiffServ
Load 95 % Load 80 %

19 Délai de bout en bout : Scénario IP best effort
Load 80 % Load 95 %

20 Délai de bout en bout : Scénario IP DiffServ
Load 95 % Load 80 %

21 Conclusion Contributions : Résultats des tests :
Un algorithme de marquage de paquets vidéo MPEG-4 pour routeur IP DiffServ a été proposé et validé Configuration de domaines DiffServ selon plusieurs PHBs. Résultats des tests : Réduction de la perte des données sensibles Réduction du délai de bout en bout (~50%)

22 Travaux Futurs Passerelles de signalisation DMIF-SIP pour des conférences multimédias : Interconnexion DMIF2SIP Interconnexion Interworking Function DVMA dynamique et adaptatif Utilisation de BB (bandwidth Broker / COPS Server) et de surveillances d’évènements réseaux.

Télécharger ppt "Présenté par : AHMED Toufik Université de Versailles - France"

Présentations similaires

Annonces Google