INSTITUTION NAME: Florida State University

WORK PACKAGE TITLE: IC

PROJECT TITLE: A Testbed for Network-Based Tools for Technical Collaboration

POC NAME: Geoffrey C. Fox
   EMAIL: gcf@cs.fsu.edu
   PHONE: 850 644 4587
     FAX: 850 644 0098

CTA or PEI: I/C

PROJECT DESCRIPTION:

[[[ This proposal should be considered as a placeholder, or a very
rough draft which will be refined in conjunction with the
Collaboration Whitepaper according to the following basic ground
rules:
a) We will participate in establishing 2 testbeds
   1) Access Grid or equivalent High End
   2) Real Networks or equivalent desktop
b) We have a general approach "Collaborative Portals" which is
   particularly close to Gateway but builds on TangoInteractive
   lessons. We want to understand what tools to integrate into this as we
   and all other extensive projects we are aware have found some
   reluctance to adopt collaborative tools.
c) We have some high level agreed goals: enable distance working
   together on: Administrative requirements for Staff meetings, Standing
   Committees, Focused meetings such as for Finance, Personnel,
   Operations, Training, etc.  and Technical requirements: Computational
   requirements, Visualization requirements, Peer collaborations, etc.
d) We don't know how to implement these desires -- what type of
   testbed, what type of tools, synchronous or asynchronous.  So we need
   a set of well defined experiments on test beds.
e) Integration of databases into collaborative portals figures
   prominently in Administrative Collaboration.  Also relevant to
   other CTAs
f) Technical collaboration ought to include Collaborative visualization
   using DICE
]]]

Advances in networking and computing have in recent years opened up
new possibilities for computer-based collaboration over long
distances. The PET program has worked extensively demonstrate the use
of synchronous collaboration frameworks (specifically Tango
Interactive) in distance education and training activities.  In this
highly structured environment, which provides a comfortable analogy to
the traditional face-to-face lecture experience, deployment and use of
computer-based collaboration tools has been quite successful.  

At the other end of the spectrum is the concept of general technical
collaboration.  In this case, the environment has very little
structure -- does the collaboration take place dekstop-to-desktop or
in special rooms (i.e. specially equipped conference rooms)?  Which
collaboration tools are useful and even how they might be used depends
strongly on the people and the project; it is easy to envision many
"specialty" tools being useful depending on the scientific domain or
other aspects of the work.  In addition, while it is not hard to find
computer-based equivalents for most of the tools used in face-to-face
technical collaboration, users often face problems related to
installation, learning curves, and fundamental limitations in
comparison to the real-world tools they're accustomed to. (It is rare
to find a conference room that is not equipped with a whiteboard and
people unfamiliar with its use; while in comparision imagine everyone
in the group might have to download and install a computer whiteboard
package, figure out how to use its cryptic interface, and then try to
draw something using the clumsy computer mouse.)

In this project, we attempt to look at general technical collaboration
in a way that will directly address the obstacles cited above.  We
will deploy a testbed environment, which will provide a focused set of
"core" tools, which will be chosen to provide as much similarity to
familiar real-world tools as possible.  The tools will be presented as
individual components which can be accepted or rejected by users
separately, rather than the take-it-or-leave-it approach when
monolithic collaboration frameworks are used.  This also facilitates
taking advantage of the rapid pace of advance of commercial tools in
this area.

To provide a certain amount of structure to the collaboration, we
recommend building the testbed around the Access Grid nodes which have
been discussed for ARL and Adelphi and will also be available at FSU
and NCSA, for example.  These sites provide equipment suitable for
high-quality audio/video conferencing, and sufficient display and
computer capability to handle most collaboration tools.  Working at
the AG locations also limits the number and variety of computer
systems on which which have to be installed and supported.  We will
also make limited use of more general desktop-to-desktop environments,
focusing primarily on ARL-FSU and ARL-ASC collaborative interactions.

Experiments in collaboration will be carried out using a mixture of
computer-based and traditional techniques.  We will focus our
attention on existing collaborations which already work effectively in
a distributed fashion in order to better understand how computer-based
tools can effectively be incorporated into their environment and which
tools and features are or are not important.  We plan to work
primarily with the distributed Gateway project, involving ARL, ASC,
FSU, NCSA, and OSC.  The core of this group (ASC, FSU, OSC) has
already been working together on this project for more than a year.

Technical collaboration, as with any other, can be divided into
synchronous and asynchronous interactions.  In a separate
"administrative collaboration" proposal, we emphasize asynchronous
tools which, in large part, can also be used for technical
collaboration.  For example, many groups make use of mailing lists to
communicate quickly and easily; a searchable mailing list archive can
be a convenient tool to capture these exchanges and make them easier
to refer to in the future.  A document repository can be used to
collect and share software documentation, presentations, reports, and
other documents among the group.  Because these types of tools, which
will be piloted in the administrative collaboration proposal, will be
easily deployable to the technical collaboration community, this
project will focus primarily on synchronous collaboration tools (many
of which are also relevant to "administrative" events like meetings
and briefings).

o Audio/video conferencing is an important tools.  The Access Grid
  equipment provides an excellent environment for high-quality a/v
  conferencing.  Various commercial tools for a/v conferencing will be
  tracked and tested as part of our I/C Core Support effort, and we
  expect to deploy tools for this testbed (i.e. those from
  RealNetworks) based on the results of those investigations.
o A shared web browser is, these days, such a basic tool for sharing
  web-based documents and making technical presentations that it has
  been included explicitly both in this project and in administrative
  computing.
o A whiteboard is an important general tool for technical discussions
  and brainstorming.  Strictly computer-based whiteboard tools are
  available which may be used effectively in conjunction with tablets
  now available in various sizes as an alternative to the mouse, which
  is quite clumsy for drawing.  In addition, real whiteboards can be
  interfaced to computers, which can then capture and transmit the
  whiteboard contents.  Many such whiteboards are rather expensive
  (thousands of dollars), however InFocus's Mimo (www.mimio.com) is a
  small, inexpensive ($500) and portable attachment for any existing
  whiteboard that allows it to capture whiteboard contents on a
  computer.  Because the real whiteboard is so simple to use, the
  computer equivalent must also be as easy and as similar as possible.
  For this reasons, we plan to evaluate both software whiteboards
  together with tablet input devices, and the Mimio product in order
  to understand user preferences.
o Display sharing systems such as NetMeeting, VNC, and others provide
  a general means to share a wide range of other tools, documents,
  etc.  We will deploy and evaluate selected tools in this area.

Though we are primarily talking these tools as standalone components
at present, it is also important to look forward to how they would be
integrated into an information or computing portal to provide
collaborative capablilities there.  In this context, we will use these
tools to refine the definitions of "portalML" and "resourceML" which
are being developed as part of Yr 5 portal efforts sponsored by ERDC
and ASC MSRC PET programs. PortalML is used to define and customize
the user's interface to a portal, while resourceML is used to define
the back-end resources which are accessible via the portal.  This work
will help set the stage for future testbed deployment of collaboration
tools integrated into information or computational portals as a
natural evolution and merger of concepts from collaboration and portal
technology.

The idea of computing portals is closely related to one further
important element of general technical collaboration: shared
visualization tools.  These tools, which may be general or specialized
to specific domains or problems, can be extremely useful when
collaborators need to analyze or discuss data in a graphical form.
While they can be thought of as an extension of the "shared display"
capabilities mentioned above, visualization tools are often far more
complex and resource-intensive than a tool like VNC.  Though
collaborative visualization tools are, in and of themselves, outside
the scope of this proposal, we plan to work with the SV team in order
to incorporate shared visualization tools and technologies into our
technical collaboration testbed.  Note that the focus on using Access
Grid nodes will also provide a good environment for visualization
tools.

PROJECT OBJECTIVES:

This project will explore some network-based tools that could be used
for general technical collaboration by geographically distributed
groups. While a wide range of specialized tools could be envisioned
for specific groups, we emphasize some of the more general tools that
would be valuable to most collaborations and note that the tools
described in our separate "administrative collaboration" proposal are
also relevant to technical collaboration.  Unlike previous PET work
aimed at general collaboration, this project will focus on a limited
set of tools in a testbed environment that provides more structure and
control than unrestricted desktop-to-desktop use of the tools.  This
will make it easier to support collaboration experiments, and easier
to get useful feedback from participants.

DELIVERABLES:

* Technical collaboration testbed utilizing Access Grid nodes at ARL,
  Adelphi, FSU and other sites.  Providing audio/video conferencing,
  shared web broswer, whiteboard, and shared display capabilities.
  Other capabilities, such as shared visualization will be
  incorporated as appropriate.
* Appropriate training and support for participants in collaboration
  experiments, to be provided jointly by FSU staff and the I/C on-site
  lead.
* A series of experiments, developed jointly with PET management,
  designed to "exercise" various aspects of the testbed and generate
  user feedback.
* A technical report analyzing the experience with the testbed tools
  and making recommendations with respect to the tools themselves,
  more general deployment, and drawing general conclusions with
  respect to computer-based technical collaboration in the HPCMP.
* A technical report on "collaborative portals"

CUSTOMERS/END USERS:

Ultimately all HPCMP users, but during Yr 5 the focus wil be on
controled testing.  We expect the ARL-ASC-FSU-NCSA-OSC Gateway project
to be one important group of users.  We will also try to identify a
limited number of users willing to participate in experiments between
Aberdeen and Adelphi.

BENEFIT TO THE WARFIGHTER:

The availability of tools which facilitate collaboration and the
efficient exchange of information among members of a working group
will result in higher productivity and efficiency for HPCMP
researchers working in groups and for those implementing the HPC
Modernization Program.  It will allow more resources to be devoted to
research and working directly with users.

PROJECT DEPENDENCIES AND SCOPE:

FSU and ARL on-site staff (particularly Derek Moses, part time on-site
I/C lead) will work together to develop the tools and to deploy them
into a testbed, but this project will only succeed if we can identify
users willing to participate in collaboration experiments and provide
feedback. For this we will require the assistance of the ARL PET team.

As described, this project assumes that Access Grid nodes will be
installed at ARL Aberdeen and in Adelphi, per our recommendation at
the Mid-Year Review.  AG nodes are also expected to be installed at
FSU, already exist at NCSA, and we believe that ASC and OSC have
sufficient hardware to allow them to implement something approaching
an AG node.  Use of AG infrastructure is not strictly required, but it
provides a high quality audio/video environment and helps to provide a
more structured environment which we believe will be beneficial to
experiments on technical collaboration.

There may be some additional cost to participating sites to acquire
certain commercial software packages (i.e. a/v conferencing tools) and
hardware (i.e. tablets or Mimio whiteboard capture system).  We have
allocated $2,000 in our budget to cover these costs at FSU, but do NOT
account for the costs other sites will incur.

This project will make extensive use of the on-site I/C lead, Derek
Moses.

RISK ELEMENT:

In our past experience with general technical collaboration, we have
found little uptake of collaboration tools.  We believe there are a
number of reasons for this.  Use of Access Grid facilities as the
focal point of collaboration will help structure the environment and
the interactions so that we can be more confident that the quality of
the collaboration services provided will be high.  Focusing on
individual tools or components will make for easier evaluation and
acceptance of them (in the past, prospective user have been presented
with a large, unfamiliar framework -- "take it all or leave it").
Finally, working with established collaborating groups will help in
obtaining feedback that highlights the effect of the collaboration
tools on the interaction.

REQUIRED FUNDING LEVEL
               Year X:
             Year X+1:
             Year X+2: