Presentation is loading. Please wait.

Presentation is loading. Please wait.

A Quality-Driven Decision Engine for Live Video Transmission under Service-Oriented Architecture DALEI WU, SONG CI, HAIYAN LUO, UNIVERSITY OF NEBRASKA-LINCOLN.

Similar presentations

Presentation on theme: "A Quality-Driven Decision Engine for Live Video Transmission under Service-Oriented Architecture DALEI WU, SONG CI, HAIYAN LUO, UNIVERSITY OF NEBRASKA-LINCOLN."— Presentation transcript:

1 A Quality-Driven Decision Engine for Live Video Transmission under Service-Oriented Architecture DALEI WU, SONG CI, HAIYAN LUO, UNIVERSITY OF NEBRASKA-LINCOLN HAOHONG WANG, MARVELL SEMICONDUCTORS AGGELOS KATSAGGELOS, NORTHWESTERN UNIVERSITY IEEE Wireless Communications, Aug. 2009

2 Outline  Introduction  Service-oriented Architecture(SOA)  Current real-time video transmission  Proposed SOA system  Case study  Experiment result  Conclution

3 Introduction – Service-Oriented Architecture  SOA has been regarded as a promising distributed network management method in large-scale heterogeneous communications networks  Entire video communication system can be decomposed into many different services provided by one or more service providers.

4 Introduction  Two types of live video applications:  Video streaming application (Youtube)  pre-encoded and packetized at the same server  Cannot adapted to changes such as network congestions.  Interactive video application (videoconferencing)  videos are coded on-the-fly  source content and network conditions are jointly considered to determine the optimal encoding modes

5 Prosoped SOA system

6  Decision engine can retrieve the user profile information and services from the broker network, optimize the service configuration, and implement different capacities of applications. User perceived video quality Available services Different capacity of apps

7 Media Signal Processing Service  Based on different user profiles and available network resources, decision engine selects different media signal processing algorithms (services) to deal with user requests.  Extracting the ROI  Downsampling  Filtering the high-frequency component  Encoding or transcoding a video sequence  Dropping the current frame

8 Performance Evaluation Service  Network-centric metrics such as throughput, delay fail to provide an efficient and accurate evaluation  Different importance of video bitstream  Continuous and smooth playback  Error resilience and concealment  Application-centric metrics such as expected end-to-end video quality are the most straightforward and reasonable.  Calculation of video quality is based on some predefined rate-distortion function or model.

9 Network Service  Path selection  Multiple paths in a multihop network that may provide different levels of reliability  Decision engine integrate some existing routing protocol, such as optimal link state routing (OLSR), into a workflow to find the optimal transmission path.

10 Network service  Resourse allocation  Multimedia data of a given video stream have different levels of importance to the user-perceived video quality  Various resourse allocation and scheduling approaches have been developed. Such as time slot/bandwidth allocation, packet ordering, and retransmission.  The decision engine needs to choose an appoach such that the user-perceived video quality is maximized while the utilization enhanced.

11 Case study  An SOA-based live video communication sysytem 1. N-frame video sequence C ={g 1, …, g N }. Each video frame can be divided into a foreground and a background. Foreground part being the ROI. 2. Wireless network model as a DAG G(V, E) with node set V and edge set E. 3. packet k over G delay deadline is associated with frame decoding deadline T max.

12 Case study(cont.) 4. Always checks the total delay of packet k at node v. If exceeds T max, packet k should be discarded. 5. Use pixel recursive algorithm(ROPE) to performance evaluate, estimating the expected distortion. The contributions of foreground and background distortion to the user-perceived video can be weighted by λ k.

13 Case study(cont.) 6. The scheduling service Φ k for packet k is based on the video quality evaluation result. Priority scheduling approach first scheduled the foreground packet for transmission. 7. The maximum number of retransmissions Π k (v,u) for packet k over link (v,u) is jointly determined by the packet delay constraint T max and the total delay

14 Case study  Each packet k generated by the media signal processing service and transmitted by the network is characterized by:  The source coding service S k  The transmission path selection service P k  The scheduling service Φ k  The packet delay deadline T max  The quality impact factor λ k

15 Object function  Expected distortion for packet k can be written as E[D k ] = Q k ( S k, P k, Φ k, T max, λ k )  Object function for decision engine V is the generated workflow by decision engine for end user.

16 Experimental Result  Identification of the ROI is performed by the following stages  background subtraction  split-and-merge  morphological operations.

17 Experimental Result  Simulation parameters  H.264/AVC JM 12.2  Video Clip: “Mother and Daughter.”  30-node network deployed over a 1000 m × 1000m  Source and destination are chosen randomly  Transmission range: 150 m  Generate 50 topologies and run 50 computations to obtain the average.  Packet delay deadline T max : 0.033s

18 Experimental Result Two network-centric routing service: PLR-based: packet loss rate as routing metric Delay-based: packet delay as routing metric

19 Experimental result Without priority scheduling: foreground and background are the same With priority scheduling: foreground has a 4.5 dB PSNR better than whole video without IRI 9.5 dB PSNR better than background

20 Experimental result (a)Original (b)Using content analysis and priority scheduling (c)Without using content analysis and priority scheduling

21 Conclusion  Traditional multimedia communication systems are lacking the flexibility of end-to-end QoS for various multimedia applications, especially for live video applications.  A quality-driven decision engine for real-time video transmissions based on SOA jointly considered and optimized various kinds of data processing services by the decision engine.  Experimental results show that the proposed quality-driven service-oriented decision engine can provide better end-user experience.

Download ppt "A Quality-Driven Decision Engine for Live Video Transmission under Service-Oriented Architecture DALEI WU, SONG CI, HAIYAN LUO, UNIVERSITY OF NEBRASKA-LINCOLN."

Similar presentations

Ads by Google