Technologies to Support Collaborative Learning
over the Multimedia Home Platform
ın o a a
Mart´ L´pez-Nores, Yolanda Blanco-Fern´ndez, Ana Fern´ndez-Vilas, Rebeca
ıaz-Redondo, Jos´ J. Pazos-Arias, Alberto Gil-Solla, Jorge Garc´
P. D´ e ıa-Duque,
Bel´n Barrag´ns-Mart´ınez, and Manuel Ramos-Cabrer
Department of Telematic Engineering, University of Vigo, 36200, Spain
Abstract. T-learning —the provision of educational services over Inter-
active Digital TV— is regarded as a complement to e-learning solutions,
whose scope is limited due to the digital divide. The increasing con-
nectivity, together with the fact that IDTV users are no longer passive
spectators, permits to start thinking on highly interactive services that
support ﬂuid communication among users and service providers. This
may enhance the prospects of learning at home, as long as the current
oﬀer is mostly based on broadcast content, with scarce use of return chan-
nels for feedback. This paper proposes an extension to the Multimedia
Home Platform standard, aimed at providing better support for collab-
orative learning services. We describe an architecture for such services
and a selection of freely available technologies for their implementation.
We also discuss the possible market implications of our approach, as the
ideas presented here contribute to openness in the ﬁeld of IDTV services.
Key words: Interactive Digital TV, t-learning, peer-to-peer architectures.
In recent years, there has been growing interest for distance education systems,
in order to overcome the main lacks of traditional learning. This has led to a sig-
niﬁcant development of Web-based learning initiatives (e-learning) . But the
increasing use of the Internet has revealed some shortages, related to diﬃculties
in the use of computers for some social sectors, limited penetration of computers
in homes and uneven presence of broadband infrastructure, among others. Thus,
some initiatives are being taken to port educational services to other mediums
than personal computers. These are t-learning, supported by Interactive Digital
TV (IDTV), and m-learning, based on the use of modern mobile devices .
Progress is being done towards convergence of these mediums, by promoting
interoperability of applications and contents, adaptation to diﬀerent devices and
integration of heterogeneous networks . So, it is not a question of the computer
taking over the television or vice versa: it is a question of how the diﬀerent
services and models converge, increasing value to the consumer . This will set
the foundations for continuous and ubiquitous learning, even though the users’
habits and expectancies vary greatly for each particular medium. As a result, the
situation today suggests that diﬀerent solutions should be crafted to adapt the
characteristics of each medium, while keeping the goal of convergence in mind.
Our working group has been involved with IDTV architectures and services
for several years. Recently, we presented an approach to developing educational
applications, that addressed the speciﬁcs of broadcast services . According
to pjb Associates , such services play a signiﬁcant role at the initial stage
of t-learning in which we are today. However, it is commonly agreed that the
pedagogical value of t-learning will be greatly enhanced by supporting higher
levels of interaction among users and service providers.
The goal of this paper is to anticipate the needs of the future range of services,
introducing a framework for the development and deployment of distributed ed-
ucational services. Our work is based on the Multimedia Home Platform (MHP)
standard , which is the most important normalization initiative in the IDTV
ﬁeld nowadays. In its current version, the mechanisms oﬀered to handle the re-
turn channel are still rather simplistic, only adequate by themselves for limited
feedback from single users. In response to this, we propose an extension to the
standard that improves its support for real interactivity and collaborative work.
In the following section, we comment past and future trends of t-learning,
pointing out its most important peculiarities. Section 3 presents our architec-
ture for distributed t-learning services, with Sect. 4 describing the technologies
we consider most adequate for its implementation and some additional details.
Finally, in Sect. 5, we discuss the relevance of our work regarding learning and
business opportunities, and also suggest future lines of work.
2 T-learning Experiences and Promises
As a learning platform, IDTV is considered a key to reach the widest audiences,
for there is at least one television in nearly every household, while the average
penetration of personal computers is not expected to go beyond 70% in the short
term (see ). Moreover, IDTV is easy to use and well known for everybody,
meeting the socially important need to oﬀer online learning services to people
who cannot aﬀord to buy a computer, do not have Internet access or lack the
knowledge to use such technologies. This is remarkable in the background of a
global economy, where access to knowledge is regarded as the best way to main-
tain a region’s competitiveness. As a result, t-learning is getting the attention of
governments interested in developing nationwide distance education plans .
In the migration from analogical to digital TV, it was soon envisaged that
t-learning could help fulﬁll the emptiness in the set of applications that TV users
are willing to pay for (see ). So, broadcasters started to provide simple, pi-
oneering services as a complement to their channel oﬀerings . At this initial
stage, educational services have been mostly based on broadcast contents (one-
way communication, with little use of feedback mechanisms), providing means
for simulated interactivity, i.e., interaction only with elements available in the
broadcast streams. The emphasis was placed on informal learning through edu-
tainment (education and entertainment), coherent with the consideration of TV
as a medium for entertainment and the passive habits of the users.
The importance of broadcast services is stressed by the fact that their interac-
tive features are driving users towards more active proﬁles in the use of TV, not
being passive spectators anymore. This, together with the increasing availabil-
ity of high-quality bidirectional networks, makes it possible to start thinking on
highly interactive services, based on the profuse exploitation of return channels.
The increased interactivity (referred to as real interactivity) brings up op-
portunities for more engaged learning. By allowing users to actively interact
with others, it contributes to mitigate the feeling of isolation that is typical
in distance education. Moreover, it sets the basis for new ways of collaborative
learning, promoting the creation of virtual learning communities  that gather
together people with common interests and learning needs. As argued in ,
this is more likely to widen participation in learning than just focusing on the
interactive oﬀerings through broadcast-based services.
The Peculiarities of T-learning The extensive research done in the e-learning
ﬁeld since the mid 1990s gives some useful theoretical insight for t-learning ,
as well as powerful standards for content management and student monitoring
(like SCORM or IMS), which are also applicable to t-learning. Despite, there
is unavoidable work needed to accommodate the peculiarities of the IDTV en-
vironment , that range from the limited resources available at the receivers
(set-top boxes) to the complexities of the information distribution mechanisms.
Another major issue is that, while e-learning courses have text and graphics
as a central axis, t-learning ones should be naturally based on audio and video.
This demands great interrelation capabilities between contents and a tempo-
ral reference, and suggests a distinction between user-driven and media-driven
strategies for interactivity . In user-driven strategies, applications respond to
the users’ actions; in media-driven ones, the evolution of a piece of media con-
trols the ﬂow of applications, so that users take part in a reduced number of
decisions. Which approach is most suitable for a given service depends on the
role that users are expected to play: user-driven strategies are recommended
for active roles, and media-driven strategies for more passive ones. Particularly,
media-driven strategies are the best option for edutainment.
3 A Scheme for Distributed T-learning Services
The work we presented in  addresses the needs of broadcast-based t-learning
services, oﬀering mechanisms to eﬃciently exploit simulated interactivity. The
extension to real interactivity and collaborative learning that we propose here
demands additional solutions: services are conceived as distributed ones and, as
commented in Sect. 2, mechanisms are needed to control the way users interact
and self-organize into virtual learning communities. This also requires means to
diﬀerentiate roles among users, group management capabilities, etc.
Fig. 1. The structure of a course and the internals of a pedagogical unit
We have devised a simple and ﬂexible structure for educational services (re-
ferred to as courses). A course is a set of pedagogical units (PUs), which deﬁne
most of the service’s logic. PUs are the primary level for contents organization,
and they can be arbitrarily complex, containing any kind of elements (media
clips, user interface widgets, etc.). As illustrated in Fig. 1, those elements can
be laid out in multiple sceneries and scenes, and several types of interrelations
can be established among them. The sceneries determine how the logic of a PU
is distributed among a number of diﬀerent machines —those of users and ser-
vice providers (spatial organization). On the other hand, scenes provide for the
temporal organization of activities (including content presentation).
Scenes were already introduced in , but sceneries are a speciﬁc need of dis-
tributed courses. Besides supporting distribution, they provide for the deﬁnition
of roles among users. For example, in a remote lecturing service, the lecturer and
his audience run diﬀerent sceneries, so that only the lecturer is given elements to
control the sequencing of the slides. In a given PU, there exist as many types of
agents as sceneries have been deﬁned, and how an user is assigned a given role
is service-dependent: it may be decided by the user himself, by other users, etc.
The composition of a course deﬁnes the ordering of its PUs and the condi-
tional access dependencies among them. These dependencies can be used, for
example, to reject access to one PU until some knowledge has been proved on
the preceding ones. At any moment, a PU can be locked (access forbidden) or
unlocked (access granted), and this state can be changed in response to events of
any kind. Conditional access is controlled by an entity called the Course Man-
ager (CM in Fig. 1). Actually, because of distribution, the CM is not an unique
entity, but a replicated object running on every machine inside a virtual com-
munity where the course is active. This makes it possible for access conditions
to be evaluated for every single user, for all the members of a community as a
whole, for those who form a given subgroup in the community, etc.
Every scenery runs a Scenery Manager (SM in Fig. 1), which centralizes com-
munication with the CM and other machines. It is also in charge of sequencing
the scenes, and controls the synchronization of multiple pieces of information.
For this task, we intentionally avoid standards for multimedia synchronization
available to Web services, like SMIL, because they are too resource-demanding
for set-top boxes and do not ﬁt well within MHP. The alternative we presented
in  solved the coordination of multiple informational sources within a single
set-top box; for the new distributed environment, we have extended it to support
synchronization among diﬀerent sceneries. In the remote lecturing example, this
permits all the members of the audience to follow the explanations of the lecturer
as he browses a slide show, by automatically keeping their screens up-to-date.
4 Technological Details
This section details the technological framework we propose to support the ar-
chitecture presented in Sect. 3, with the foundations of the MHP standard.
MHP deﬁnes two application models: DVB-HTML, based on Internet tech-
nologies, and DVB-J, based on the Java language. As argued in [7, 9], DVB-J has
several advantages, because it is more ﬂexible, extensible and suited for the lim-
ited resources of a set-top box. Thus, we chose DVB-J for our implementation,
and completed the framework with several freely available technologies that ﬁt
well within it: XML, JavaBeans and JXTA.
We use XML for many diﬀerent purposes, such as to communicate with IMS
databases and content repositories, to deﬁne the message types needed for an
application, and to express all those entities with a predeﬁned structure, like
the composition of a course in terms of its PUs. On another hand, we resort
to JavaBeans as the simplest way to construct applications, because it deﬁnes
an architecture of components (beans) that greatly facilitates development. This
can be done in an entirely visual way, provided that a suﬃciently rich set of
beans is available, together with a suitable environment for their composition.
Both XML and JavaBeans were introduced in the context of broadcast-based
t-learning services in , and their use is slightly enhanced for distributed ones.
But the leading role in the extension presented here is played by JXTA, an open-
source initiative aimed at providing a thin layer on top of which peer-to-peer
(P2P) applications and services can be built. This technology fulﬁlls the needs
of the sceneries approach for distributed logic commented in Sect. 3.
4.1 JXTA Virtual Networks as Virtual Communities
P2P technologies are suitable for the purposes of collaborative t-learning, be-
cause they promote decentralization, in a way that end users and devices are
given much more relevance than in classical client-server architectures. In this
regard, JXTA deﬁnes a set of protocols that make it possible to establish virtual
networks on top of Internet and non-IP networks. This technology is network
and language-independent, and can be used on a wide range of devices (including
PersonalJava-compliant ones, such as MHP set-top boxes).
In our approach, a virtual network maps directly to a virtual learning com-
munity, with all the agents in it being treated as peers (Fig. 2). The ways peers
Fig. 2. Mapping physical networks into a virtual one
discover each other, self-organize into virtual communities, ﬁnd resources and
communicate, etc. are all part of JXTA technology.
All the entities in a JXTA virtual network (communities, services...) are pub-
lished by means of advertisements, which are XML descriptors following a given
syntax. The mechanism used for discovery is based on rendezvous super-peers,
i.e., peers in well-known locations designed to cache advertisements. Due to the
limited computing power of set-top boxes, we propose to use service providers’
computers as rendezvous nodes (Fig. 2), making them also responsible for com-
plex tasks such as group management and searching in the virtual network.
JXTA deﬁnes a second type of super-peers —relays— to deal with heteroge-
neous networks and protocols, allowing applications to exchange messages with
no concern about the networks they traverse. We exploit this idea to intro-
duce broadcast relays, which provide for the combined use of broadcast and IP
networks for communication inside a virtual community. Broadcasting is a nat-
ural option for information ﬂows that should be served to all the members of
a community, as it happens with the explanations of the speaker in the remote
lecturing example. This beneﬁts from the robustness and high quality of service
of broadcast networks, and avoids squandering Internet bandwidth. By the way,
remark that most of the current set-top boxes (including MHP ones) do not
support audio or video streaming through the return channel, so any audiovi-
sual content generated by an user should always be sent to a broadcast relay for
others to access it. The broadcast relay forwards the messages it receives onto
the broadcast networks, addressed to the corresponding virtual community.
JXTA over MHP As speciﬁed in , broadcast in MHP is based on MPEG-2
transport streams, and the return channel is mostly operated by means of the
TCP/IP suite of protocols. Following the ideas above, we have implemented a
JXTA communications layer to make transparent use of broadcast networks and
return channels (see Fig. 3, where the gray arrows only exist in broadcast relays).
Regarding the return channel, our communications layer uses the TCP and
UDP protocols. As for broadcast, it uses MPEG-2 sections for streaming audio
and video content, object carousels for ﬁles, and UDP over Multiprotocol En-
capsulation for messages due to bindings between elements in diﬀerent sceneries
JXTA communications layer
MPEG-2 sections Object carousels UDP
Fig. 3. Our JXTA communication layer
(remember Section 3). Thus, all the messages exchanged among peers are sub-
ject to being broadcast —especially when they are targeted to many others—,
taking advantage of the inherent capabilities of broadcast networks for multicas-
ting. Which mechanism to use is decided considering developers’ guidelines and
traﬃc conditions, in order to use the available networks the best possible way.
4.2 Integration in a Development Environment
We have integrated the technologies commented in the preceding sections, com-
ing up with a CASE tool for the development of distributed t-learning services.
This is actually an extension of the tool we presented in  as the ﬁrst envi-
ronment speciﬁc for t-learning services. The new version implements the scener-
ies approach, and provides several new wizards to cope with the needs of dis-
tributed services. One such wizard is used to deﬁne scripts to dynamically access
databases and content repositories. Another one allows deﬁning interaction pat-
terns among the peers in a virtual community, taking into account the diverse
roles they may take up. It is particularly easy to state that some users can see
the eﬀect of actions done by others, or that a particular action can only take
place if all the users with a given role agree on it.
5 Discussion and Future Work
In this paper, we have presented an architecture for distributed IDTV services,
and a selection of free technologies suitable for its implementation. This comes
as an extension to the return channel mechanisms of the MHP standard. Our
proposal is based on peer-to-peer technologies, as they natively provide for the
establishment of virtual communities of people with shared interests. We also
discussed the convenience of establishing a virtual network on top of the physical
ones, where the use of broadcast networks for multicast inside a community is a
very interesting feature, particularly when dealing with multimedia content.
The introduction of P2P technologies may have a signiﬁcant impact on the
IDTV business models. T-learning initiatives have been so far controlled by
mainstream broadcasters, keeping the educative community apart. Because P2P
promotes decentralization in networks and services, our approach contributes to
openness in the educational market, leaving place for private enterprises, and
even individual users, to oﬀer learning services (advertising them is just part
of the P2P framework). This opens several new ways for broadcasters to make
business: hiring their broadcast networks to provide high-quality streaming of
multimedia content; hiring computational power and storage capacity for com-
plex computations and large databases; or giving access to content repositories
that store numerous selected pieces of learning or audiovisual material.
Our future work will deal with personalization and convergence. Personaliza-
tion is a need in order to handle the increasing amount of information available,
and also to target services and products to users who may be interested in them.
In this regard, the TV-Anytime Forum is working on solutions for the deﬁnition
of user proﬁles and mechanisms for the markup, storage and retrieval of multi-
media content. On the other hand, the desired convergence of IDTV with other
mediums requires further advances in interoperability of software and content.
The most relevant roles here will surely be played by the Java language, XML
and the MPEG-4 standard. Particularly, MPEG-4 is called to revolutionize the
creation, distribution and use of multimedia content. Its foreseeable inclusion in
MHP set-top boxes will cause some changes in the way of doing things, extend-
ing the focus in development to media authoring tools. However, frameworks
like ours will still be needed, for example, to deal with distribution, content
management, the deﬁnition of roles among users, and any other additional logic.
1. Alexander, S. E-Learning Developments and Experiences. Education and Training,
2. Atwere, D. et al. Interactive TV: a Learning Platform with Po-
tential. Learning and Skills Development Agency, available at
3. Caro, J. et al. E-learning and Virtual Communities. In Proceedings of the Philip-
pines Computing Science Congress, 2000.
4. Chiariglione, L. Standardisation in a World of Converging Technologies: the Audio-
Visual Case. Audiovisuel et autoroutes, Rennes, France, June 1996.
5. Digital Video Broadcasting Consortium. The Multimedia Home Platform Stan-
dard. http://www.mhp.org, 2001.
6. Gil-Solla, A. et al. Exploring T-learning in the MHP Context. In Proceedings of
the IADIS International Conference WWW/Internet, Lisbon, Portugal, 2002.
7. L´pez-Nores, M. et al. A Mixed XML-JavaBeans Approach to Developing T-
learning Applications for the Multimedia Home Platform. In Proceedings of MIPS
2003, Naples, Italy, November 2003.
8. Lytras, M. et al. Interactive Television and E-learning Convergence: Examining
the Potential of T-learning. In Proceedings of ECEL 2002, Brunel, UK, 2002.
9. Morris, S. et al. The MHP Tutorial. http://www.mhp-interactive.org.
10. pjb Associates. Development of Satellite and Terrestrial Digital Broad-
casting Systems and Services and Implications for Education and Training.
11. pjb Associates. A Study into TV-based Interactive Learning to the Home.