iGrid 2000 GiDVN Demonstration Korean Side Plan

July 6, 2000

Seungmo Choe (APAN-KR/KOREN NOC, KT)
Jaewha Lee (APAN-KR/KOREN NOC,KT)
Yung Yi (Seuol National Univ.)
Sungkwan Youm (Korea Univ.)

1. General description
   This demonstration applies QoS to the GiDVN demonstration traffic. This demonstration is an attempt to provide DiffServ support for DV over the JP-KR link that will be provisioned for supporting QoS traffic. The demonstration will provide for a combination of production-level IPv4 traffic (BE - best effort traffic) and GiDVN demo traffic (premium service traffic) on the same link. The current JP-KR link bandwidth is 8Mbps and the day-to-day average traffic seems to average around 3Mbps. If the IPv4 traffic does not provide for sufficient volume congestion on the link during the demo, additional traffic will be produced. The potential sources include video server sources at the Seoul National University and Korea University, and 4-way videoconferencing between Yokohama, SNU, KU and the APAN-KR NOC. APAN-KR is a consortium providing a means for Korea to participate in the APAN, and KOREN is an advanced research & education network, which support APAN-KR members such as SNU and KU and the original KOREN members in Korea.

2. Collaborators
   
   • Jaehaw Lee, Seungmo Choe (APAN-KR/KOREN NOC)
   • Yung Yi, Yanghee Choi, Wang Lijing (Seoul National University)
   • Sungkwan Youm (Korea University)
   • Kazunori Konishi, Yasuichi Kitatsuji (APAN-JP, KDD)

3. Scenario

4. Network Topology
   iGRID 2000 GiDVN Demonstration Topology -- Korean Part

   

5. Preparation for DiffServ/QoS enabled DV
   • Video Source 1: SNU video charger MPEG1
   • Video Source 2: KU Real video MPEG1
   • DiffServ enabled routers: SGP_qb, TGP_qb, Seoul-XP, Tokyo-XP, and DiffServ routers in Japan
   • KR-JP link bandwidth: 2Mbps for this premium service demo from the total 8Mbps
   • Video Contents: real time video and stored video data for demonstration.(provided by SNU and KU)
   • DiffServ routers should be configured to classify the video stream from the source 1 to premium traffic. (The video stream 2 will be classified to best-effort traffic.)
   • H.323 video conferencing system: This will be used for communication among the collaborators.
     • Participants: SNU, KU, NOC and Yokohama site
     • H.323 MCU: Japan
     • Client: NetMeeting.

6. Demonstration Procedure
   1. SNU and KU send video sources.
   2. Normal State – Both Video 1 and Video 2 have normal video quality.
   3. Yokohama site requests APAN-KR NOC to send background traffic with H.323 VC.
   4. APAN-KR NOC sends background traffic.
   5. Yokohama site observes if the background traffic affect video quality with H.323 VC.
   6. Yokohama site shows the graph of KR-JP link traffic load and downgraded video quality.
   7. Yokohama site requests APAN-KR NOC to stop sending background traffic with H.323 VC.
   8. APAN-KR NOC sends background traffic.
   9. Yokohama site observes if the quality of video is recovered.

7. Measurement

8. Application level
   • Real Video
     • Required BW and Average received BW
     • Loss: # of packets
     • Retransmission: # of request / success
   • Video Charger
     • Network Transfer Rate
     • Packet Loss Ratio: # of Received Packet and # of Loss Packets
   • The difference of video quality between EF and BE streams can be recorded with camcorder.

9. Network level
   • Delay: RTT with PingER
   • Load: Router Command Line Interface
   • Jitter (Delay variation): Perl script provided by KU
   • Loss: Router Command Line Interface