R..L. Lagendijk
This chapter outlines the main features of the SMASH storage system (COMBO)
and introduces the role of visual search and security. It also outlines the
structure of the deliverable.
The overall objective of the SMASH project is to develop and to show the technical feasibility of an integrated storage unit for multimedia applications in a domestic environment. Integral part of this storage unit are provisions for searching and protection of the stored information, and in particular for the visual information. This deliverable describes the technical possibilities and methods for these functions in the SMASH project.
Figure 1.1 shows the project's conceptual view on the place that the SMASH
storage unit has in the consumer's home. The storage system is externally
connected to the network through either a set top unit (DVB input) or through
a PC (internet connection). In the domestic environment the SMASH storage
unit serves as a remote server for multimedia applications. For this reason
the unit itself does not have a user interface, but is accessed through either
the PC application with Internet/Java interface or through the set top unit
(STU) application running for instance an MHEG engine.
Figure 1.1: SMASH storage unit in the domestic environment
The architecture of the storage unit is built around two tightly coupled but technologically different storage media, namely a linear tape drive (LTD) and a hard disk drive (HDD). The linear tape drive is a cheap (in terms of Ecu/Mbyte) high capacity storage medium (at least 50 Gbyte) with relatively slow access time, while the hard disk drive provides a relatively expensive smaller capacity storage (e.g. 2 Gbyte) with fast random access. The two components of the SMASH storage unit are managed by the operating system of the unit itself, while PC/STU applications can access the data residing on LTD or HDD through the Application Programming Interface (API).
The multimedia data used in applications will be dynamically stored in a distributed way, depending on the character of the multimedia data and on the application. The main distinction in storage will be between (audio-visual) stream data (to be stored on the LTD) and random access data files (to be stored on the HDD). To the user, however, the storage unit will be transparent, i.e. the user is not aware of how the data is distributed between the LTD and HDD. The chosen architecture of the storage unit combines the advantages of hard disk technology (fast and random access) with those of linear tape technology (huge storage capacity). The MultiMedia DataBase (MMDB), to be stored on the HDD, plays a crucial role in storing various kinds of service information, in relating information stored on the LTD and the HDD, and in providing efficient user access to the stored information.
The development of a storage system as outlined above requires the solving of many technical and conceptual challenges. In the SMASH project many of the architecture and application problems and choices have been discussed. Of course, these issues cannot and have not been discussed in isolation, since there are many interrelated issues in architecture, applications, and the systems special functionalities. Taking into account the developments in other branches of the project, this deliverable concentrates on specific software/hardware functionalities for managing the stored data, among which are searching of stored video streams, copy protection and encryption of private data, and the multimedia database (MMDB) for organizing and efficient retrieval of stored information.
Of particular interest in this deliverable is the DVB-VCR, which is one of the SMASH project's application studies: a recorder and player for satellite and cable DVB services, especially digital MPEG video with enclosed service information, which will use the storage unit for additional functionalities (break button, recording and playback of multiple stream) and increased user friendliness (browsing and selection of services using the program guides and visual contents of the services, management of tape collections).
With the development of mass storage devices such as the SMASH storage system, also comes the need for (i) efficient management and (ii) protection of the locally stored multimedia information. The importance of these two functions and the place where and how they are discussed in this deliverable are addressed in this section.
There is a growing world-wide interest in the development of browsing methodologies and tools tailored to the storage system and applications that accommodate for easily locating specific pieces of information in huge volumes of information. Existing navigation systems are all based on prior annotation of the information or search processes on textual information. It is widely recognized that there is a need for intelligent management and search methods in particular for the visual information in multimedia documents and in digital video and image libraries. The problems at hand are firstly that textual annotation can usually not be made beforehand because of the subjectivity and complexity of this task, and secondly that visual information will be transmitted and stored in a compressed format complicating even the most simple image analysis operations.
In one of the applications selected by the SMASH project, digital video broadcasting (DVB) services will be recorded. To accommodate for an efficient navigation of the stored information, some sort of contents-based summary of the stored video streams needs to be made as the streams are received and stored. In Section 2 of this deliverable, the state-of-the-art in making such a summary of video information is described.
In the recent past we have experienced that the success of digital video broadcasting was not only determined by the existence of technical solutions (e.g. MPEG-2), but also by the ways in which service providers would be able to make money. The latter has led to a complex system of enciphering MPEG transport streams, deciphering, and billing. On similar grounds one can state that the success of multimedia storage systems not only depends on the technological advances, but also the existence of adequate copy protection methods. Service providers will not be willing to offer services in digital form without copy protection mechanism that limits duplication of the digital multimedia data.
Copy protection systems exist in which the protection signals are stored separately from the data. Since these protection signals can easily be removed without affecting the data quality, this system is insufficiently strong. More recent approaches embed the protection signals into the data, which is called labelling or watermarking. In Section 3, the state-of-the-art in copy protection of analogue and digital audio and video information will be presented.
On the basis of state-of-the-art in searching, protection, and combo architecture, and taking into account the application requirements, the SMASH project has developed requirements and a solution concept for visual search, copy protection, and storage of the (summarized) information in the MultiMedia DataBase (MMDB). In Section 4 these requirements and the solution concepts will be outlined.