Proposed Collaboration on Access Grid Ubiquitous Computing

1
Proposed Collaboration on Access Grid /
Ubiquitous Computing
JongWon Kim and Juwon Park The Temporary
Genkai-Hyeonhae Meeting Feb. 27th,
2003 Networked Media Laboratory Dept. of
Information Communications Kwang-Ju Institute
of Science Technology (K-JIST) jongwon_at_kjist.ac.
kr http//netmedia.kjist.ac.kr
2
Contents

• Collaboration on Access Grid
• Collaboration on Ubiquitous Computing

3

• Collaboration
• on
• Access Grid

4
Access Grid

• Access GRID
• Support group-to-group interaction across the Grid

5
Usage Areas

• The Academic and Research, Government, Private
  Sectors
• Making Remote Collaborations Work across
  Boundaries
• Check Multi-Sector Collaboration over the Access
  Grid by J. T. von Hoffman (Boston Univ)

6
Collaboration Issue for H-G Joint Projects

• Access Grid (4)
• Status 2
• Contact Koji Okamura (Kyushu Univ.)
• JongWon Kim (K-JIST)
• Participants
• Kyushu Natl Univ, AIST
• KISTI, K-JIST, Chonnam U, HYU
• Description (to be revised)
• The Access Grid, as a communication channel for
  users from remote sites, can give researchers and
  scientists in both Korea and Japan a seamless and
  immersive collaboration environment. With this
  project, we want to explore the potential of
  Access Grid in supporting collaborative research
  between two countries. To realize it, we will
  collaborate to research ways to build low-cost
  Access Grid system (e.g. mini-AG or PIG) and to
  promote its deployment. Also, we will conduct a
  joint research to improve current AG systems to
  better support enhanced video/audio quality,
  interaction, security, and inter-operablity.
• Expected Results (to be revised)
• - Provide a seamless colloaboration channel
  for researchers and scientists.
• - Promote the adoption of multicast technology
  over H-G link.
• Required bandwidth
• - Dependent on the number of site, camera per
  site, video rates
• (normally full access node requires
  bandwidth around 1.5Mbps)

7
Relation with AG and other projects
8
NGrid Project lead by KISTI
GGF Grid
APAN APGrid
Euro APGrid
ANL/NCSA
APEC APGrid
NGrid KISTI(20022006)
Grid Tech-center
GNOC
National Grid Infra
Grid Standard
Grid Middleware Research
GFK
N BT Grid
N Supercmpt Grid
Security
N NT Grid
N Clusters Grid
Resource Mngt.
N CT Grid
Data Mngt.
N Appls. Grid
N ET Grid
Inf. Service
Access Grid
N IT Grid
VR Grid
N ST Grid
9
KISTI AG

• First AG in Korea (2002)

Int. AG Community
KISTIs Access Grid Node
Digital Video
RGB Video
Display Computer
Shared App. Control
Network enabled IP multicast
NTSC Video
Video Capture Computer
Digital Video
Analog Audio
Digital Audio
Mixer
Audio Capture Computer
RS232 Serial
Control Computer
Echo Canceller
10
K-JIST Mini AG

• Personal Interface to Access Grid (PIG) (Oct.
  2002)
• With AG 1.2 release (note that PIG software is
  still very delicate)

11
Towards Enhanced Access Grid e-AG

• An experimental, on-going trial to enhance the
  functionalities of Access Grid by augmenting
  improved support for immersive 3D Display, Human
  Computer Interaction, and Networking.
• With e-AG, we hope support
• All functionalities of Access Grid
• Stereo video delivery and display
• Collaborative use of 3D computer graphics.

12
Immersiveness Enhancement for Access Grid via
Stereo Video Delivery

• For providing the immersiveness, we include the
  stereo video delivery functionality to EAG
• Developing 3D video delivery module enables the
  reliable and real-time
• The developed system supports
• Reliable networking of real-time media with
  reliable transport via ARQ, FEC, and their hybrid
• Versatile display of stereo video in Left,
  Interlaced, Left Right, and Above Below views.

13
Immersiveness Enhancement for Access Grid via
Computer Graphics

• For the graphics side of 3D, EAG is being
  enhanced by supporting CG-based sharing in
  various communication architectures.
• Combination of push repository in server memory
  and file repository method is implemented to
  support multi-user collaborative environment.
• UIC EVLs QUANTA is adopted as a baseline
  networking and data management module
• VR software tools such as CAVELib or VR Juggler
  are also selected and utilized to build immersive
  3D display.

14
Collaboration Issue for H-G Joint Projects

• On-line Joint Technical Conference on Access Grid
• Joint Project (Long-term)
• Joint participation in AG 2.0 and beyond

15

• Collaboration
• on
• Ubiquitous Computing
• - Seamless Mobile Service? -

16
Whats UbiComp?

• Whats Ubiquitous Computing (UbiComp)?

Distributed Systems
Mobile Computing
Ubiquitous Computing

• Smart space context-aware
• Invisible smart agent
• Localized scalability physical distance-based
• Uneven condition heterogeneous infrastructure

• Remote communication
• Fault tolerance
• High availability
• Remote info access

• Mobile networking mobile IP, ad hoc protocol,
  etc.
• Mobile information access BW-adaptive, data
  consistency, etc.
• Energy-aware systems process scheduling, memory
  management, etc.
• Location-sensitive systems location sensing,
  location-based services, etc.

Courtesy of K-JIST U-VR Lab
17
UbiComp Challenges ubiquitous Networking

• Infrastructure ubi-networking
• Beyond desktops/servers
• Connected to networks
• Networked appliances
• Sensors
• Historical sites other locations

18
Collaboration Issue for H-G Joint Projects

• Ubiquitous (Wireline/wireless) Computing (18)
• Status 2
• Contact Morioka (hmorioka_at_isit.or.jp)
• JW Kim (jongwon_at_netmedia.kjist.ac.kr
  )
• Participants
• ISIT
• KJIST
• Description
• We are planning to build a prototype version
  of ubiquitous environment utilizing
• Mobile IP and IEEE 802.11a/b (and other L2
  media) connecting Fukuoka, JAPAN and Gwangju,
  KOREA.
• On top of the built ubiquitous environment
  we will explore example applications such as user
  tracking
• and IP mobile video streaming.
• Expected Result
• In our joint effort, instead of trying to
  build hot spots, we will focus on building real
  ubiquitous
• environement with mobility and security
  support. We hope to let people in both countries
• to better locate each other and communicate
  ubiquitously.

19
Project Scope Objective

• Mobile Media Delivery via Wireless Multicast and
  Network Adaptation

Wireless Aware Media Application
Group Communication (1-to-N, M-to-N)
Efficiency Better Bandwidth Utilization
Mobile IP / Wireless Multicast
Wireless Network Adaptation
Reliable Multicast
Advanced Streaming (Wireless Multicast)
Reliability Better Quality
20
Project Scope Objective (Cont.)

• For Streaming Media Applications over
  Wired/Wireless Integrated Networks
• Real-time streaming of stored contents only (but
  can include on-line streaming) to enable
  buffering (up to 3sec)
• Maybe extended to conferencing (limited scope) at
  later phase of project
• Will address the challenges such as
• Mobility How to track users movement and
  sustain seamless streaming?
• Multicasting How to enable sharing of wireless
  channel efficiently?
• End-to-end QoS How to maintain the quality of
  media streaming over highly fluctuating wireless
  network?
• Anticipated Duration 2002. 12 2006.12 (4 years)

21
Implementation Phase I (2003) One-to-one
Streaming

• Environment
• Mobile IPv4
• IEEE 802.11b
• Server Darwin or Helix Server
• Client Notebook
• MPEG-2/4 streaming
• Objectives
• Implement the streaming server/client that
  provides seamless streaming (i.e., no service
  disruption during handoff).

Media Server
Adaptive streaming client
22
Implementation Phase II (2004) One-to-many
Streaming

• Environment
• Mobile IPv6
• IEEE 802.11b or 802.11a
• Client Extension to PDA?
• MPEG-2/4 streaming
• Source Specific Multicast (SSM)
• Objectives
• Implement the streaming server/client providing
  that provides seamless multicast streaming.

Media Server
Multicast
Adaptive streaming client
23
Mobility Challenge Mobile IP
CN

• Smooth and Fast Handoff ?

Forwarding buffer
7
6
5
forwarding
FAnew
FAold
Bind
Output buffer
to FAnew
Handoff
24
Mobility Challenge Handoff-aware Streaming Client

• Adaptive buffer control
• To reduce the discontinuity of streaming
• Increasing the buffer level of client before
  handoff
• Question
• What is a proper buffer level?
• What is a handoff duration?

Handoff is initiated
Handoff is initiated
Buffer Level
Buffer Level
No playback
time
time
Handoff is completed
Handoff is completed
lt No buffer level control gt
lt Buffer level control gt
25
Mobility Challenge Handoff-aware Streaming
Client (Cont.)

• Estimation of handoff duration by adopting
  K-JISTs Transient Time Analysis
• Source traffic rate
• Handoff load
• Background traffic load
• Beacon time period (router advertisement)
• Buffer level control
• Using the handoff duration (HTP) to calculate the
  proper buffer level.
• Minimizing the buffer level
• Providing no media discontinuity
• ? No service disruption during handoff

Collecting parameters
Calculating HTP
Estimating Buffer Level
Bc Current Buffer Level Bt Target Buffer Level
BcltBt
Decrease Buffer
Increase Buffer
26
K-JIST Transient Time Analysis How long time
does Handoff take?
Source Traffic Turning Point

• Handoff Transient Time

New stream
Old stream
Handoff Initialization
New stream
to
tn
Case 1
Old stream
STP
HTP
tf
Case 2
STP
UTP
HTP
Old Router
New Router
Case 3
UTPUnstable Time Period STPSilence Time
Period HTPHandoff Time Period
UTP
STP
HTP
HTP may be a good reference to the buffer level.
27
K-JIST Transient Time Analysis Handoff Period
Time (HTP)

• Timing Analysis
• UTP max(0,Tlast_of_old- Tfirst_of_new)
• max(0, (to - tn) qo tf)
• STP min (Tfirst_of_old, Tfirst_of_new)
• HTP UTP STP Tfirst_of_new Tfirst_of_old
• Queueing Model

qo
(a) Before handoff
(b) After handoff
28
K-JIST Transient Time Analysis Analysis Result
29
Seamless Streaming Aid by Adaptive Playout?

• Merits of Client Adaptation
• Can prevent excessive buffer overflow and
  underflow
• Can reduce discontinuous playback time
• Can maintain small buffer size.

• Role of Clients for Adaptive Playback

Buffer level
30
Seamless Streaming Network Adaptation

• Robust/Scalable Media for Wireless Media Delivery
  and Associated Network Adaptation

Layered Video Encoding
Network Adaptation
Receiver
R-D / Corruption Model
Layered RPI
Wireless Network
Delivery With Network Adaptation (Rate,
Delay, Loss)
Source Rate/ Error Resilience
Receiver adaptation
R-D Analysis
Video pre- processing
Video packet Stream
Layered Encoding
Network Feedback
Network Monitoring Feedback Handling
Frame Complexity
Quality Control
End-to-End Feedback
Application Feedback
31
End-to-end QoS Challenge Wireless LAN

• Relative Differentiated Services and the
  Proportional Differentiation Model for IEEE
  802.11x WLAN
• Service Differentiation Upstream
• Different IFS (Inter Frame Space)
• Different Back-off Time
• Different Frame Size
• Service Differentiation Downstream
• Scheduler in Access Point
• qi Performance measure for class I
• c1 lt c2 lt.lt cN Quality differentiation
  parameter
• Proportional Delay Differentiation Model
• Proportional Loss Differentiation Model

Scheduler
q1
q2
q3
Classifier
AP
32
Collaboration Issue for H-G Joint Projects

• With H. Morioka_at_ISIT, Fukuoka, Japan
• How to combine ISITs Fast and Secure Handoff
  Idea ? (Short-term)
• With implementation Deploy ISITs handoff at
  K-JIST and test its performance for seamless
  streaming
• MISP (mobile Internet Service Protocol) for
  Authentication, Address Assignment, and
  Encryption
• Joint Project (Long-term)
• Build ubiquitous computing environment ?

33
ISIT Handover with 2 Interfaces (1)
34
ISIT Handover with 2 Interfaces (2)
35
ISIT Handover with 2 Interfaces (3)
36
ISIT Fukuoka Mobile Broadband (FMBB)
37
Thank You!
38
Research Network in Korea Korean Partner
Carrier KJCN Japanese Partner Carriers Fukuoka
Gigabit Highway GENKAI Project QGPOP part
KIST
Korea
KISTI
Repeaterless
KJCN
KJIST
Shimonoseki Univ. of Fishery
250 km
Busan

Gwangju
Kitakyushu City High-speed LAN
GbE
To Tokyo
GENKAI/Hyeonhae Project
Kitakyushu
150Mbps
2.4Gbps
Nogata Iizuka Tagawa
Fukuoka
QGPOP (ISIT)
APAN Internet2
Japan
Kurume
FGH AP
Submarine cable landing station
Ohmuta
Network equipment (Routers etc.)