MACRO UNIVERSITY

A Framework for a
Federation of Virtual Universities

S. K. Chang, University of Pittsburgh, USA
T. Arndt, Cleveland State University, USA
F. R. Guo, National Taiwan University Medical College, Taiwan
S. Levialdi, University of Roma, Italy
A. C. Liu, Feng Chia University, Taiwan
J. H. Ma, Aizu University, Japan
T. Shih, Tamkang University, Taiwan
G. Tortora, University of Salerno, Italy

(Draft, April 4, 1999)

1. Introduction

Nearly every academic institution these days has set up a web site and is in the process of starting, or has already started, a distance learning program on the Internet. Coupled with a digital library, many an academic institution is experimenting with the "virtual university". In some countries the local government has taken the initiative to form a consortium of universities offering on-line courses from each institution. These consortia are also called virtual universities or virtual campus [4]. Thus the term "virtual university" has become ambiguous, referring sometimes to the distance learning courses of a single academic program, sometimes to the distance learning program of an academic institution, and sometimes to a consortium of distance learning programs.

More serious than the confusion in terminology, however, is the unreal expectations from the administrators of academic institutions that a virtual university can be the "cash cow" of the school. Some administrators believe once they have a virtual university, they can dismiss half of the instructors and thus become more cost effective. These people fail to realize the importance of providing a learning environment on the Internet, which is not an easy task. Just to offer some courses on the Internet does not provide a learning environment. The students can easily get confused and disoriented, if left alone on the Internet. Beneath the "virtual university", there needs to be another layer, offering a student a personalized and personable learning environment.

On the other hand, many visionaries have already pointed at the possibility of combining the educational resources from a large number of academic institutions, thus creating a super rich learning environment. Without losing sight of the individual student's micro-universe, it is hoped that the coupling and coalition of numerous academic institutions will constitute a macro-universe for the students.

It is for these reasons that we introduce a new framework for distance learning consisting of three layers: the micro-university [3], the virtual university, and lastly the macro university. A micro-university is a self-contained learning environment, usually on a single PC (or lap-top, notebook, palm-top, PDA, etc.), for a single student. A micro-university is designed to serve a single student, but of course many instances of the same micro-university can be created to serve many students.

A virtual university is a collection of individualized learning environments so that students can engage in learning activities from home, remote locations, etc. A virtual university is usually owned and operated by an academic institution. A virtual university therefore has more administrative functions than a micro-university.

The Macro University is a framework such that many virtual universities can pool their resources together, thus creating a very rich learning environment for a large number of students from all over the world. The following is the motto for the proposed framework of Macro University: "Macro University enables any student to learn from any place at any time".

2. Characteristics of the Macro University

The characteristics of the Macro University can best be explained by the Three Elements (tien-di-ren) and Three Levels of Understanding (shi-jian-hui). The technical design principles can be derived from these Three Elements and Three Levels of Understanding in all three layers of the Macro University. No matter what layer we are dealing with -- the Macro University layer, the virtual university layer, or the micro-university layer, these principles should be respected.

The Three Elements are:

  • Adaptability (environment, Tian1-Shi2)

    MACRO-U can operate under different resource constraints, such as network bandwidth, error rate, workstation's limitations, etc.

  • Scalability (organizational characteristics, Di4-Li4)

    MACRO-U can be for individual instructor operating by himself, small to large institutions, and federation of institutions

  • Universal Accessibility (people, Ren2-He2)

    MACRO-U delivers education to any one at any time in any place regardless of linguistic skills, physical handicaps, age differences, etc.

    Three levels of understanding (Shi4-Jian4-Hui4) are:

  • Awareness (awareness, Shi4)

    MACRO-U enhances user awareness

  • Visualization (cognition, Jian4)

    MACRO-U supports user's dynamic visualization of the three elements

  • Understanding (wisdom, Hui4)

    MACRO-U enchances user's understanding

    To explain the three levels of understanding, consider the following example. A student can be brought into awareness of a tool, such as the crane. The student first observes the general characteristics of the crane. He then tries to observe the details, and tries to imagine (or visualize) how the crane behalfs in action. By jumping back and forth between the higher level and the lower level abstractions of the crane, the student finally accomplishes true understanding of the mechanical principles of the crane.

    It is our belief that the Macro University (and the micro-universities) should offer different levels of visualization tools, simulation tools, etc. so that true understanding can be accomplished.

    3. Macro University's Software Architecture

    MACRO-U consists of a federation of virtual universities, each of which consist of multimedia micro-universities called MMUs, which operate on dynamically changing collections of tele-action objects or TAOs [1], [2]. 
                            --Virtual University--Micro-University 1
                        |
     Macro University --
                        |
                        |                    --Micro-University 2
                        |                    |
                        --Virtual University--
                                             |
                                             --Micro-University 3

    A micro-university is a learning environment for an individual student. It is designed to optimally serve the needs of that student. A virtual university consists of one or more MMUs. In the limiting case a virtual university is the same as a single micro-university. But usually a virtual university consists of many micro-universities and in addition can perform many complex administrative functions.

    MMUs can be small, medium or large. It can be a single software module, a few software modules, or a complex configuration of software modules. For example, the current MMUs of the Macro University include: Private Office, VCR, Knowledge Table Manager, Virtual Library, Adlet System, etc.

    4. Networking Infrastructure

    Virtual University software systems are established on the Internet. With the growing popularity of Internet, distance learning or lifelong education courses are widely available. However, to support Macro University operations effectively on the Internet, there are some problems yet to be solved. The following are phenomena observed on current virtual university applications over Internet:

    4.1. End-to-end delay

    To transmit binary large objects (BLOBs) such as real-time video, there are several reasons cause the delay. Video and audio information are captured, compressed, put to packets before it is transmitted on packet switching based network. On the other side, the data stream are unpacked, decompressed, put to buffer, and played on one or more remote workstations. All of the above processes are time consuming, especially in packet transmission. An efficient coding and compression scheme as well as suitable buffer sizes are necessary.

    4.2. Heterogenous network bandwidth

    Since Internet is a heterogenous environment, different network infrastructures support different bandwidths. Current solutions to BLOB transmission hardly support multi-resolution of multimedia information upon demand. Bandwidth of Internet is wasted in some situations.

    4.3. Without Quality of Service (QoS) guarantee

    Internet bandwidth is used in a best effort solution manner. An individual application consumes the required bandwidth as much as possible until Internet bandwidth is exhausted. However, unlike e-mail information, virtual university communications required real-time transmission with a reasonable quality of service. But QoS is not guaranteed on Internet.

    4.4. Lack of a good prioritizing mechanism

    Internet supports a uniform delivery class. Since transmissions are of the same priority, non-real-time applications may consume immediate network bandwidth which is deadly required by other time critical applications. A good prioritize mechanism should be embedded in the transmission protocol.

    4.5. No general prediction of network bandwidth

    Since Internet is decentralized, it is hard to have prediction of bandwidth and granted bandwidth to applications. If bandwidth can be predicted and reserved, real-time transmissions are possible.

    4.6. Package loss

    Due to abnormal situations (e.g., network failure), packages are lost. In some cases, in order to avoid congestion, some packages are dropped. Package loss brings a problem to real-time communication. A fault tolerant solution is required.

    4.7. Week multicasting

    Virtual university is designed for multiple workstations. Current Internet technology fails to support efficient multicasting. An efficient and robust multicasting strategy is required.

    To cope with the problems, researchers of networking and multimedia computing communities thus work together toward a future solution. Internet2 is a project with over 100 corporate partners and affiliate members. The solution promises end-to-end broadband network services to be established among the participating universities and national laboratories. One of the goals of Internet2 is to support national research activities as well as distance education and lifelong learning. Objectives of Internet2 focus on broadband media integration, interactivity, and real-time collaboration among broadband communication network users.

    In spite of the construction of the Internet2 infrastructure, there are other considerations toward the success of virtual university. All BLOBs are compressed before they are transmitted. Due to the heterogenous network bandwidth of Internet, compression techniques require progressive coding of image, video, and audio. Upon the demand of a network infrastructure and an individual workstation, different resolutions of multimedia information are transmitted. Adaptive routing techniques are thus necessary in such a transmission. There are two problems need to be solved. Firstly, a multimedia record needs to incorporate different resolution of information. In JPEG 2000, Discrete Wavelet Transform (DWT) provides a solution of progressive still image coding. However, in MPEG 2, even the temporal domain can be progressive (e.g., dropping B-frames), the spatial domain of MPEG 2 is hardly to be incorporated with the DWT techniques. Also, audio progressive coding techniques are not developed yet. However, if DWT techniques are used effectively, it is possible to extract a portion of multimedia record depends on the bandwidth to deliver just enough information. But, the adaptive routing techniques need to face another challenge -- the estimation of immediate network bandwidth and useful routes. Distributed software agents are required to estimate network bandwidth on the fly and provide the intelligence to the routers. The router needs to operate based on some QoS requirements of real-time multimedia information and to operate based on a rate adaptive protocol (increase or decrease transmission rate upon demand) of non-real-time transmission.

    MACRO-U operates on the metropolitan network and local area networks including: MBorn, xDSL, ISDN, Fast Ethernet and 100VG AnyLan.

    5. Specification of Tele-Action Objects for Open Information Exchange

    MACRO-U is a worldwide consortium of Virtual Universities. As such there will be a wide variety of computational platforms used by students of MACRO-U. In order to minimize the problems associated with exchange of data in a heterogeneous computing environment, an open standard for information exchange in the MACRO-U framework is required. Fortunately, there exists an emerging standard for Electronic Data Interchange (EDI) - XML, eXtensible Markup Language. Virtual Universities are structured as a collection of MMUs where each MMU is a collection of TAOs. TAOs are described by TAOML, and TAOML is itself an XML application. Thus, we can use standards-based EDI in the MACRO-U.

    The preliminary TAO prototyping facility implemented TAOML as an extension of HTML. The TAOML was then interpreted by a TAOML Interpretor which produced standard HTML as output. This output was combined with an HTML template file (TPL) to produce the user interface portion of the TAO. The advantage of this method is that the TAO uses the universal user interface - HTML. It is viewable on any computing platform containing a web browser. Furthermore, through the widespread use of the web, HTTP has become the universal network protocol. TAOs (and by extension MMUs) are universally available. The disadvantage of this method is the need for the TAO interpretor and the lack of required markup tags for TAOs in HTML. The frontend to this TAO environment is a visual specification tool which allows the user to specify a TAO in a user-friendly way and which automatically converts the specification to TAOML.

    In MACRO-U the TAOs will represent the salient objects for a distance learning environment: class slides; recorded and live lectures; exams; syllabi; recorded and live demos; etc. Each of these objects will have its own characteristics which need to be described in order to present the TAO at the student's computer. In XML, each of these objects can be described by a schema called a Document Type Definition (DTD). An XML parser can then verify the structure of XML documents by referring to the appropriate DTD. XML separates the structure of the document (given by XML) from the presentation of the document given by XSL (eXtensible Style Language) a stylesheet language designed for use with XML. XSL enables formatting information to be associated with elements in the source document to produce formatted output. A variety of stylesheets can be produced to meet the various needs of the students of MACRO-U: synchronous vs. asynchronous learners; low- bandwidth vs. high-bandwidth connections; workstation vs. PDA; visually impaired students; etc.

    XML carries metadata about the document in the form of character data. Thus the metadata is easily interpreted and can also be searched by programs written, for example, in Java. This possibility greatly enhances the exetensibility and interoperability of MMUs expressed as XML documents. XML documents consist of both parsed and unparsed data. Parsed data is character data, but unparsed data has no limitations on its structure. Thus we can used unparsed data to carry the multimedia content of a TAO if it is useful to pack it into the XML document. Otherwise, the multimedia data can be represented in the XML document by Uniform Resource Locators (URLs) and can be either on the local machine, or on a remote server, and they can be combined in a multimedia presentation upon demand.

    Since TAOs are multimedia objects with a dual representation (physical structure and knowledge structure), the TAOML format must reflect this by including the knowledge structure of the TAO (the set of index cells) in the XML documents. Index cells (ICs) are currently implemented as CGI programs (compiled C programs) and are therefore unsuitable for a heterogeneous computing environment. In order to meet this requirement, ICs can be implemented as servlets in Java, which is platform neutral. Using Remote Method Invocation (RMI), ICs can be distributed, as well. ICs implemented in Java can even adapt to changes in the environment by downloading new ICs from an Index Cell Base using JDBC assuming the ICs are stored in a relational database on some remote server.

    The multimedia components of TAOs, whether embedded in the XML document or not, need to have a standard format for exchange as well. Fortunately, a number of international standards exist in the area of multimedia. MPEG can be used for video data, either with or without synchronized audio. JPEG can be used for image data. MPEG-A or some de facto commercial standard (e.g. WAV) can be used for audio data. Streaming audio and video can use the format of some widely available browser plug-in or a newly emerging international standard (perhaps some future MPEG). Standard are also needed for vector graphics (e.g. WMF) as well as synthesized music (MIDI). Both XML and Java support UNICODE which is fortunate since it allows the text content of TAOs to be in both Latin and non-Latin character sets, depending on local student needs.

    The MMUs are responsible for the creation, exchange, management and updating of the TAOs. The creation functionality is already present in the MICE (Multimedia IC Developer's Environment). Exchange of TAOs can be implemented as a push technology, whenever new courseware needs to be distributed, it is sent to the Virtual Universities participating in the given course. Documents for specific purposes (e.g. communication with course instructors) can also be defined. The MMU may batch together multiple messages for the Virtual University originating the the course. Asynchronous communications and discussion groups are also coordinated by the MMU's communication component. Management of the TAOs involves resource allocation on the local machines (e.g. use of local filesystems). The update component of the MMU is responsible for incremental update of TAOs based, once again, on exchange of XML-based documents. In order to interact with Java XML parsers and Java ICs, the MMU itself should be implemented in Java.

    6. Evaluation and Assessment by Computation on TAOs

    Is distance learning REALLY as effective as face to face instruction? If so, what types of delivery systems or students react the most favorably to distance learning? Dr. Thomas Russell has been researching these issues for the past several years. His work has been posted online at the "No Significant Differences" Web site [5]. The site is so named because after reviewing over 300 studies on the effectiveness of all types of distance systems Russell concluded that their generally is no significant difference in learning outcomes when face to face and distance learning options are compared for the same populations. Now, Dr. Russell has expanded his research and has compiled a printed book that details his findings [6].

    Although the "no significant differences" phenomenon is reassuring, it is still necessary to provide measurable and quantitative analysis of the effectiveness of distance learning sessions. A formal framework for evaluation and assessment can be formulated, based upon on computation on media objects. A distance learning session consists of time-sequenced slices of collection of multimedia objects (TAOs). Given a time sequence of collections of multimedia objects (TAOs), an evaluation function be defined. For example, the distance learning session consists of a collection of e-mail messages and video clips. The computed output is the session summary. We plan to use adlets to dynamically compute and propagate evaluation functions.

    7. Discussion

    We need to consider the following main points for obtaining good results in distance learning:

    a) a fast and cheap interaction with a standard software multiplatform package;

    b) the possibility of sharing working contexts (writing, listening and drawing together);

    c) the access to centers of excellence in the subjects taught;

    d) the encouragement of "lateral teaching", i.e. learning from other students;

    e) the production of a list of urls containing papers, proceedings, books, etc correlated with the lectures/experiments;

    f) as in Shneiderman's book his work can be turned "active" via an url which is made available by his publishing company, in the same way, any subject, task, exercise should also have a web site for finding out "more about it" and "who is who in that subject";

    g) to provide a "social environment" for the real students so that they may immediately become aware of other, similar, students doing their same work.

    References:

    [1] H. Chang, S. K. Chang, T. Hou and A. Hsu, "The Management and Applications of Tele-Action Objects", ACM Journal of Multimedia Systems, Springer Verlag, Volume 3, Issue 5-6, 1995, 204-216.

    [2] S. K. Chang, "Visual Languages for Tele-Action Objects", in Artificial Vision: Image Description, Recognition and Communication, V. Cantoni, S. Levialdi, V. Roberto (eds.), Academic Press, 1996, 281-301.

    [3] S. K. Chang, E. Hassanein and C. Y. Hsieh, "A Multimedia Micro-University", IEEE Multimedia Magazine, Vol. 5, No. 3, July-September 1998, 60-68.

    [4] Illinois Virtual Campus, see http://www.ivc.illinois.edu.

    [5] T. L. Russell, "The No Significant Difference Phenomenon", see http://teleeducation.nb.ca/nosignificantdifference/.

    [6] T. L. Russell, "The No Significant Difference Phenomenon", published by Instructional Telecommunications, North Carolina State University, Raleigh, NC, USA, 1998. (see http://www2.ncsu.edu/oit/nsxflyer.jpg)

    Return to Macro University Home Page