CDI-Type II: Toolkits, Components And Applications For Building Virtual Organizations

Martin Greenwald, Massachusetts Institute of Technology - PI

1. Project Summary:

This proposal seeks to significantly enhance the effectiveness of collaborative research and graduate education through the development and integration of new communication technologies, rich presence displays and social networking software. The focus will be on increased quality of interpersonal and group interactions, particularly informal, ad hoc communication, which are essential to the functioning of teams but are not well supported by current tools. Components and toolkits developed under this project will be deployed and evaluated in a diverse set of research and education applications that span scientific and engineering disciplines and which are national and international in scope. The project would be a partnership between university based researchers and educators and DOE funded laboratories operating major scientific facilities, which are themselves the focus for broad multi-institutional collaborations and graduate education. The project team has substantial experience in facilitating distance learning, remote participation and remote collaboration. From that experience has come the observation that existing tools are not capable of providing the degree of engagement and immersion that are required to make virtual teams truly effective. For distance learning and distributed research collaborations, the weaknesses of current technologies translates into poorer educational experiences and less productive research. To analyze the causes and provide remedies, we propose the following approach.

  1. Develop toolkits for integrated multi-media and enriched presence displays, including standards-based, modular, role-aware, presence-aware, web-friendly, multi-platform, multi-media communications components. Integrate these tools for rapid “inflation” of virtual groups using data stored in social networking applications
  2. Implement presence via integration of “virtual worlds” (or other geographic metaphors)with communications tools, environmental sensors, screen sharing and external data (e.g. calendars). A key element of this task are sensor networks and other tools for linking virtual and real worlds.
  3. Create a partnership between university and government laboratories for construction and deployment of application prototypes, built from these components. Application areas would include: distance learning involving the MIT-Portugal program and remote laboratory instruction (iLabs); graduate education using facilities remote from the university and multi-institutional; internationally-distributed research collaborations in fusion energy and high-energy physics.
  4. Evaluate these tools in real-world research and education settings including investigation of learning and interaction effectiveness.

Intellectual Merit: Through innovative use of emerging technologies, the proposed research should have a profound impact on the educational and scientific disciplines involved. The team assembled has extensive experience in developing, providing and applying tools and technologies that foster group interactions and remote participation. The real needs of research and education will be addressed directly. Extended discussions over several years has led members of this proposal team to appreciate the strong overlap in interests and commonality of the problems each faces. We expect significant synergies between research and education objectives, between tool developers and applications, among the different scientific disciplines and among national and international components of this project. We anticipate that experience in each sphere will inform developments in the others.

Broader Impact: While the proposed activities target a particular set of educational and research disciplines, their breadth along with the common interests and requirements that they represent, suggest that approaches developed by this project would have wide application. The basic problem that this project seeks to address –effective informal and ad hoc communication in extended groups – is not unique. We note that in many areas of science, research and graduate education are carried out through large, multi-institutional, geographically distributed collaborations. The proposed approach, to develop components and toolkits that are standards-based, modular, extensible and multi-platform should allow for easy adoption by other communities. As appropriate, we will submit protocols and data formats for standardization.

2. Project Description (consisting of sections below)

2.A. List of Participants (1 page)

Martin Greenwald – Massachusetts Institute of Technology, PlasmaScience & FusionCenter (PSFC)

Vijay Kumar – Massachusetts Institute of Technology, Office of Educational Innovation and Technology

Phillip Long – Massachusetts Institute of Technology, Office of Educational Innovation and Technology

Joe Paradiso – Massachusetts Institute of Technology Media Lab

Henning Schulzrinne - ColumbiaUniversity, Dept. of Computer Science

Eric Gottschalk – Fermi National Accelerator Laboratory

David Schissel – General Atomics

Stan Kaye - Princeton Plasma Physics Laboratory – PrincetonUniversity

Jon Menard- Princeton Plasma Physics Laboratory – PrincetonUniversity

2.B. Description (5 pages)

2.B.1 Introduction: Science, Technology, Engineering and Math in the Flat World

Increasingly, education in the Science, Technology, Engineering and Math (STEM) disciplines is following the trajectory of scientific research, broadly collaborative, distributed and global in participation. Students are finding the tools needed for their research dislocated from their academic ‘home’ institutions forcing them to leave one community and join another. The thin threads of email, phone and occasional video chat or synchronous online web conferencing are weak connectors between them. A significant challenge for US research and education leadership is how students will retain value from their local academic community with its distinctive education and research programs while conducting research using instruments distributed around the globe. In the research community, despite decades of experience working in large multi-institutional, geographically diverse teams, only partial success has been achieved. Modern communications and computer technologies hold out the promise to “virtualize” distributed, collaborative activities, binding educators, students and researchers into tightly knit and efficient teams despite geographic separation.

2.B.2 Objectives

The project team, with a diverse background in science education, communications protocols, media and scientific research, has identified a set of common interests and problems. Our experiences suggest that the greatest obstacle to effective virtual communities involves communication and engagement among geographically dispersed participants. To address this challenge, we propose construction of modular components that exploit the convergence of physical and logical communications channels so that phone, audio, video, email, messaging, screen sharing and data can be integrated into a common collaboration framework. Rich presence would be implemented via integration of “virtual worlds” or other geographic metaphors using information from environmental sensors, social networking tools and directory services allowing people, places, roles and data streams to be identified, located, scheduled and connected into a flexible collaboration fabric. Collaborative applications would be constructed from modular components, deployed into real working environments, evaluated and modified based on experiences.

We see“presence” as a mechanism for enriching communication and for projecting community. Previous research has shown that becoming an effective member of a knowledge community happens through interaction and engagement [1-3]. Presence, complicit in the conveyance of tacit knowledge critical for the formation of community as well as conducting complex social and technical interactions. The proposed research will leverage virtual worlds to provide a test environment for the study of complex social interactions associated with learning activities and as a using it as a tool through which these interactions will be distributed in a virtual community. Key affordances of virtual worlds include: Geographical Metaphor:leveraging an intuitive navigation framework [4]; Immediacy: feedback, fundamental for practice development; Interactivity: content is community built, modified and refined; Persistence: environments created are stable; Shared Space: many people can participate at once, promoting collaboration development of social capital for knowledge sharing; Socialization/Community: knowledge and social arrangements arecreated, maintained and shared. Information relevant to remote collaborators can be collected in a variety of manners. Most existing systems in big science applications use real-time data from the experiment itself that is rendered or available on remote displays, allowing distant users to observe and participate. On the other hand, much information relevant to facilitating collaboration comes not from experimental data, but rather from inferring the state of the participants and the activity occurring in the spaces where they are working. This information can often be very dynamic, and either inefficient, impossible, or very annoying for participants to manually set, hence automatic extraction of such social context is desirable for this application.

Improving remote participation in large scale scientific research is another key objective of this proposal. Partnering with DOE funded laboratories, which will provide major facilities to serve as test beds, discipline specific applications will be built, deployed and evaluated. Experiments in fusion sciences feature large teams involved in near-real-time decision making during operations. This puts a premium on good communication and engagement throughout the entire team. As early as 1992, a fusion group at the University of Wisconsin[5] demonstrated the ability to remotely view operations and control a major diagnostic on the TFTR tokamak at Princeton. Soon after, full remote operation of the C–Mod and DIII–D tokamaks was demonstrated from a control room set up at LLNL[6,7].Today, remote participation is routine with run time often given to remote session leaders. However, interactions among the distributed team are still relatively weak. For high-energy physics (HEP) the goal is to deploy integrated communications tools as application prototypes and evaluate their usefulness for work on the Large Hadron Collider (LHC) and Compact Muon Solenoid (CMS) experiment at CERN. Prototypes will be deployed and evaluated for operations by implementing them in the LHC@FNAL Remote Operations Center (ROC) at Fermilab. The ROC, which was officially dedicated in October 2007, is the first center established for remote operations for the LHC. The purpose of the ROC is to maximize the effectiveness of scientists and engineers working remotely on LHC and CMS commissioning and operations, scheduled to begin in 2008. The ROC has three primary functions: to provide a physical location with remote access to information that is available in control rooms and operations centers at CERN, to serve as a communications conduit between CERN and members of the LHC community in the US, and to serve as a focal point for US LHC and CMS outreach and education activities.

2.B.3 Approach and Work Plan

The Project Team: This project teams computer scientists, software engineers, educators, students and researchers from a range of scientific and engineering disciplines to provide common solutions to a jointly held set of problems. The team contains specialists capable of analyzing and developing the components required, others with extensive experience in building and supporting applications and still others responsible for defining the ultimate educational and research missions and evaluating the success of the proposed solutions. Adoption and evaluation of tools by partners at DOE supported laboratories will be funded by DOE, and will, in effect, provide major facilities as test beds for this work.

2.B.3.a Communication and Collaboration Components

Synchronous and asynchronous collaboration tools have existed since the 1970s and are now commonly installed on most PCs, ranging from IM-focused tools such as iChat and MS Messenger to more business-oriented Java-based environments such as WebEx. However, these approaches remain unsatisfactory for scientific collaboration, as they are either single-purpose or closed environments. They make it difficult or impossible to integrate external applications or data sources and typically require teams to cobble together a set of tools, each with their own user management and user interface conventions. Since well-established tools exist in this sphere, when possible, we will leverage and integrate tools and protocols allowing them to be part of an integrated environment, with appropriate external controls and data exchange mechanisms. Thus, we propose to architect, prototype and test an integrated, presence-enabled, extensible, open, cross-platform and secure collaboration environment designed for distributed, multi-organizational scientific teams, combining synchronous and asynchronous collaboration tools. As appropriate, we will submit protocols and data formats for standardization. We describe each of the features in turn below.

Synchronous and asynchronous: Most scientific collaborations alternate between high-intensity synchronous interactions, such as voice calls or IM sessions coupled to machine control and data sharing, and asynchronous collaborations, such as the exchange of documents, source code or data sets. In our vision, a group should have easy, unified access to the standard tools, including chat rooms, voice and video conferencing, source code management, Wikis, a shared calendar, a mailing list, blog, electronic logbooks, an issue tracker and document repositories. Setting up such a collaboration environment should only require simple web-based configuration, in one location, and should be easily manageable by non-technical staff. Group members should be able to view the collaboration from a variety of vantage points, such as a historical timeline, by creator or through full-text search.

Extensible tools: Unlike most business and personal interactions, scientific collaboration require access to data and scientific instruments. We plan to make it easy to integrate Java- and AJAX-based tools into the collaboration environment, so that group members can easily find the tools and add their own.

User management: Scientific teams often form on short notice and for varying lengths of time, from days to years. They are also dynamic, with team members joining and leaving. We propose to leverage emerging open authentication standards, such as OpenID [8], and social networking sites, possibly based on OpenSocial. A team leader should be able to invite and grant access to team members by selecting from her list of friends or entering some list of identifiers. Currently, OpenID and similar efforts are focused on HTTP-based applications; we plan to develop protocols and mechanisms that allow other application protocols, such as SIP, RTSP and XMPP, to leverage these infrastructures.

Role-based naming: In many cases, science experiments require access to key staff by function and skills, rather than a specific named individual. Our system will be designed to make it easy to designate groups of individuals where each member can help, e.g., .

Screen sharing: Tools are available for sharing all or part of computer screens. We propose to modify and repackage one or more of these to create an extensible, embeddable component to complement the communications tools. Under consideration are open-source H.261 and open source Flash encoding. These operate under multiple OS environments with a general API for easy tool integration.

Rich-presence-enabled: During intensive collaborations, tracking the status and reachability of participants becomes crucial, so that questions can be resolved quickly. Current IM applications provide little useful information beyond online/offline, offer only a single level of privacy controls, and rely on manual entry of data. Building on our work in rich presence [9,10], we plan to create an extensible set of sensors that can automatically update presence information, while respecting the privacy of participants [11]. For some forms of scientific collaboration, location information of the participants is crucial, so we plan to add location-enabled presence [12], including privacy protections [13].

Throughout the course of this project, we intend to apply and refine the concept of “Dual Reality,” as developed and initially explored by authors at the MIT Media Laboratory [14,15]. Dual Reality is a paradigm by which sensor/actuator networks tunnel information across the real/virtual divide by means of metaphoric representation. Our current work has explored applications in browsing sensor net data, where graphical metaphors in shared 3D virtual environments (such as Second Life) are driven by measured signals (e.g., motion, sound, light, vibration) and inferred phenomena (e.g., activity level, occupancy) in actual sensor-equipped buildings. We have similarly explored mappings by which the virtual world of information projects metaphorically into real spaces via displays, actuator arrays, and robotics. Abstracting the sensor data through appropriate metaphors (rather than presenting or plotting raw data) aids in efficient browsing and interaction, as salient phenomena are more readily observable. Beyond Dual Reality is "Scalable Virtuality," where the manifestation of virtual phenomena in the real world becomes a function of available and appropriate information portals. We see a unified and seamless digital information space that virtual events project into, ranging hand-held devices, to full-up immersive 3D virtual world. In the case of physics experiments, this is especially relevant, since participants will be often mobile and will have access to different kinds of interfaces at different times. Interactions can be modified depending on personal availability and urgency of the communications.

The Media Lab team has also developed a host of wearable sensors and signal processing algorithms for determining relevant social signals and group dynamics, which can be very relevant to remote presence and distributed team interaction in running physics experiments, as they correspond to parameters like interest, affiliation, and interruptability [16]. The Media Lab team is now in the process of developing and deploying a large sensor net system with wearable and infrastructure components intended for automatic context-driven media generation [17] – this system will be adapted to function as a sensor infrastructure to be installed in participating laboratories and work areas to enable relevant events, phenomena, and social signals to be automatically abstracted into virtual environments.