W3C Workshop "Television and the Web"
A View to the Broadcasting and the Internet
leading the Broadcasting to its full potential...
Information Broadcasting Laboratories, Inc.
1-1-1, Nishi-Asakusa, Taito-ku, Tokyo, 111-0035, Japan
*partly on loan from Sony Corporation, Tokyo, Japan
The rapid transition of broadcasting systems from analog to digital has been predicting the possibility of integrated services with the digital broadcasting and the Internet. In order to make the most use of the digital information infrastructure, a variety of forms of information must be well-structured from every point of view such as authoring, delivery, browsing, searching and retrieval. In this sense we define Information Broadcasting as a broadcasting service provided with highly structured information. In this paper we propose a view to the future of information distribution infrastructure and technical issues for the architecture of global and seamless information networks by means of metadata structuring technology between the Broadcasting and the Internet.
increasing number of services leads to metadata structured directory
As infrastructures of digital information network are expanding and improving rapidly, numerous number of services are going to be provided on the Broadcasting and the Internet. Millions of Web sites are already provided on the Internet. Digital satellite broadcasting systems are now about to provide hundreds of TV and Radio channels. People are loosing ways to remember channel numbers of favorite TV programs and URLs of favorite Web applications. The majority of TV users and novice PC users will be paralyzed in floods of TV programs and Web applications. Under the circumstances people will need to be well-navigated by sophisticated directory services to find and access the objects which are not only actually wanted but also potentially desired. Performing such navigation will require to disseminate appropriate information about any kinds of contents efficiently and seamlessly between the Broadcasting and the Internet, which we call metadata structured directory service.
increasing number of users leads to broadcast-based information distribution
In the meantime reducing the cost of terminals like set top boxes (STBs) and personal computers (PCs) is accelerating the growth of the number of users to millions and zillions. These users are expected to have different terminal resources from small STBs to high-performance PCs, and are supposed to use in different manners such as with his family or by himself, on the couch or at the desk. Broadcasting systems, taking the topological advantage, have been providing TV and Radio contents distribution for millions of people. Cutting-edge multimedia Web applications are also intending to be provided with multicast capability on the Internet. In order to provide directory services and contents distribution for millions of people efficiently, broadcast-based information distribution mechanisms are expected to provide a lightweight protocol framework for metadata distribution and content distribution.
learning from the Broadcasting
It is important to learn from the nature of information on TV and Radio.
* streaming information media
Historically broadcasting contents such as TV and Radio program has been implemented in streaming-based protocols, which are optimized only for transportation and presentation of audiovisual contents. Taking an advantage of connectionless communication, broadcasting systems can perform congestion-free information transportation. This is one of major advantages of broadcast-based transportation.
* poorly-structured information transport
In order to transport the information just for presentation of audiovisual streams, there has been no need to structure the information to high extent during transportation. Information on the broadcasting has been mapped only on the frequency bandwidth called 'channel'. A directory service on the broadcasting system named EPG, electronic program guide, is now expected to provide schematic structures for the information on the broadcasting.
* topological advantage of on-air broadcasting
It is impossible to know how many people are about to receive broadcasting contents and where they are located in advance of transmission. Under the circumstances broadcasting on the air can disseminate information from a few servers to millions of public clients effectively by the nature of its propagation mechanism in the air. Clients just need to filter and extract preferred information out of the stream transmitted from servers, which performance is independent of the number of clients. Note that this is one of the greatest advantages of on-air broadcasting.
learning from the Internet
There are also things to learn from the nature of the Internet.
* interactive communication media
This aspect is an essential property, which provides negotiation opportunity for communication between servers and clients. While this allows us to deliver the information on demand, it is potentially inevitable that rush of clients may cause congestion of servers. This can be a fatal problem particularly in case of live applications.
* highly-structured information transport
Such interactivity allows us to implement a various kinds of protocols, such as FTP, HTTP or RTSP, depending on the application on the IP-based single network namely the Internet. These protocols can be optimized for the corresponding applications, which appears to be well-structured on the Internet protocol layers.
* routing-based information transportation
In contrast with the on-air broadcasting, the IP-based wired network needs to perform packet routing in order to deliver specific information to designated clients. This will potentially cause network contention against unexpected peak concentration even if routing process gets faster and network bandwidth grows wider.
Keeping these aspects in mind, following points should be considered toward the unification of the Broadcasting and the Internet.
generality of description
Any kind of information should be described about any type of resources of information by descriptors. A descriptor is a representation of metadata. Metadata allows us to search desired information without parsing whole information. While a descriptor may have its own format or syntax optimized for transport systems or storage systems, it should be transformed into each other schematically. Some might be optimized for the IP system, the other might for the MPEG-2 system and the other might for file systems or database systems.
As linking information is also a type of resources of information, a descriptor may represent a link to other resources of information. Some might be linked to a web object represented by a URL, the other might to a broadcasting object represented by a program identifier. Some might be linked to a remaining object like a static file, the other to a streaming object like an audiovisual live program. Any type of linking should be described in universal expression.
Metadata has a logical or semantic structure, which is called metadata schema. With different metadata schemas, distributed web and broadcast objects can be organized into multiple metastructures. Metastructures can be formed not only between distributed objects apart from each other but also inside individual objects. This allows not only static documents but also continuous audiovisual streams to be segmented and to be restructured into virtual structures namely metastructures.
metadata transport protocol
Not only descriptors of metadata but also metadata schemas must be transported to millions of clients efficiently. Since these auxiliary data are anticipated running to a considerable amount, metadata transport protocols should be optimized for transport systems. This function should be associated with data filtering and data updating mechanisms on individual terminals.
filterable metadata coding
As a way to optimize metadata transport protocols, metadata can be compressed and coded into binary data suitable for broadcast-based systems. Highspeed and lightweight filtering schemes should be implemented with a sophisticated coding system. This aspect is practically important especially for low price STBs to filter preferred data out of highspeed broadcast streams efficiently.
distributed object management
Broadcast-based protocols can be applied for distributed object management systems, which are expected not only to perform lightweight metadata structured directory services but also to provide a distributed database management framework.
All these aspects expected to be standardized making the most use of de facto technologies. Too much heavy standards will not gain in popularity because of difficulty of implementation.
metadata description and metadata structured directory service
On the Internet there has been DNS as a directory service, which is provided by poorly-structured metadata about the domain name system. Now we are going to be able to describe highly-structured metadata about general type of web objects by means of RDF, which not only allows search engines to surely hit desired objects, but is also capable of providing directory services in a broad sense. RDF can potentially describe any kind of metadata about any type of information including TV and Radio contents.
On the digital broadcasting MPEG-2 provide a data coding system for audiovisual presentation, which is optimized for transport audiovisual streams. DVB is proposing a framework to describe and transport metadata called service information including EPG on the MPEG-2 system. MPEG-4 will provide effective compression and transport mechanisms for digital data streams by describing individual components of audiovisual data streams. MPEG-7 is intending to generalize such kinds of descriptions and to provide a metadata description standard for digital broadcasting systems. First CFP is scheduled on this October and there is no full-scale activity yet.
These separated activities are expected to be intercommunicated in order to establish a standard fit for practical use. W3C workshop "Television and the Web" is expected to be a good chance to bring up a new activity at W3C toward the unification of the Broadcasting and the Internet.
leading the Broadcasting to its Full Potential...
From every point of view proposed in this paper, we will contribute to and affect the standardization activities in collaboration with our capital companies. The key idea of metadata broadcasting framework was proposed as "Requirements for Metadata Handling" from Sony to DAVIC 11th CFP on interactive multimedia services based on in-home storage. DAVIC established the ad-hoc groups related this technology called Meta-Data ad-hoc in INR-TC and Local Storage based Multimedia Application ad-hoc in SII-TC. Our proposal will soon be focused by corresponding WGs at ARIB, Association for Radio Industries and Businesses, which will affect on Japanese broadcasting standard. The activity of MPEG-7, which will focus on metadata description for information search, is also our target. Most of the statements proposed in this paper are not contradictory to the objectives of MPEG-7 except that we are aware of the importance of interconnectivity between the Broadcasting and the Internet. Boundary province among these activities is expected to be focused by W3C. Furthermore, extending the idea of metadata broadcasting framework, we will propose a framework for distributed information management applicable to distributed directory services and distributed database services.
Yoshihisa Gonno, Research Scientist
Information Broadcasting Laboratories, Inc. #temporary affiliate
tel:+81-3-5826-7383 fax:+81-3-5826-7382 email:email@example.com
Network Technology Lab, I T Labs, Sony Corp. #permanent affiliate
tel:+81-3-5448-4423 fax:+81-3-5448-4327 email:firstname.lastname@example.org