This chapter is from the book
This chapter is from the book
This chapter is from the book
1.3 MPEG Modus Operandi
MPEG is open to experts duly accredited by an appropriate National Standards Body. On average, a meeting is attended by more than 300 experts representing more than 200 companies, universities, and research centers spanning all industry domains with a stake in digital audio, video, and multimedia. On average, more than 20 countries are represented. The group meets three to five times a year. Participants in MPEG-4 include broadcasters, equipment and software manufacturers, digital content creators and managers, telecommunication service providers, and publishers and intellectual property rights managers, as well as university researchers.
MPEG held its first meeting in May 1988 in Ottawa (Canada) and reached its 58th meeting in December 2001 in Pattaya (Thailand). A Convener chairs the MPEG activities: Dr. Leonardo Chiariglione has been the MPEG Convener since the start of MPEG.
Because of its size, an MPEG plenary meeting can be a difficult experience for a newcomer if he or she has no guidance. MPEG meetings are well known for their openness, lively discussions, and detailed dissection of issues. During the meetings, MPEG is organized in several subgroups, each one with a chairman, notably these [N4500]:
Specification development subgroups: Systems, Video, Audio, Multimedia Description Schemes (MDS), and Synthetic Natural Hybrid Coding (SNHC)
Auxiliary subgroups: Requirements, Test, Implementation Studies, and Liaison
To better coordinate the work between the MPEG plenary meetings, MPEG may decide to establish ad hoc groups (AHG). AHGs are established with mandate, membership, chairman, duration, and meeting schedule at the end of an MPEG meeting, for the sole purpose of continuing work between MPEG meetings. They are established by MPEG and report to it; the task of an AHG may cover just preparation of recommendations to be submitted to MPEG, with no decisions made. The duration of an AHG usually is limited to the period between two successive MPEG meetings; AHGs always cease to exist at the start of an MPEG meeting. Participation in AHGs is not restricted to the delegates present at the meeting.
1.3.1 Mission
MPEG's area of work is the "development of international standards for compression, decompression, processing, and coded representation of moving pictures, audio, and their combination, in order to satisfy a wide variety of applications." [N4500]. According to the MPEG Terms of Reference [N4500], the MPEG Programme of Work is as follows:
Serve as responsible body within ISO/IEC for recommending a set of standards consistent with the area of work.
Cooperate with other standardization bodies dealing with similar applications. For this, MPEG creates liaisons with other standardization bodies as well as with other relevant duly constituted organizations (e.g., industrial consortia), exchanging requirements and technical specifications with the aim of reaching the largest possible use of standards.
Consider requirements for interworking with other applications, such as telecommunications and broadcasting, with other image coding algorithms defined by other SC29 Working Groups and with other picture and audio coding algorithms defined by other standardization bodies.
Define methods for the subjective quality assessment of audio, moving pictures, and their combination for the purpose of the area of work.
Assess characteristics of implementation technologies realizing coding algorithms of audio, moving pictures, and their combination.
Assess characteristics of digital storage and other delivery media, targets of the standards developed by WG11.
Develop standards for coding of moving pictures, audio, and their combination, taking into account the quality of coded media, effective implementation, and constraints from delivery media.
Propose standards for the coded representation of moving picture information.
Propose standards for the coded representation of audio information.
Propose standards for the coded representation of information consisting of moving pictures and audio in combination.
Propose standards for protocols associated with the coded representation of moving pictures, audio, and their combination.
1.3.2 Principles
Because the technological landscape changed from analog to digital, with all the associated implications, it was essential that standard makers acknowledged this change by modifying the way they create standards. Standards must offer interoperability across countries, services, and applications, and not just a system-driven approach by which the value of a standard is limited to a specific, vertically integrated system. This brings us to the toolbox approach by which a standard must provide a minimum set of relevant tools which (after they are assembled according to industry needs) provide the maximum interoperability at a minimum complexity—and very likely cost [Chia97]. The success of MPEG standards is based on this toolbox approach, bounded by the one functionality, one tool principle. In summary, MPEG wants to offer users interoperability and flexibility at the lowest complexity and cost.
To develop its standards, MPEG has been following a few principles. The most important of these principles, many of them very well known, are these [Chia97]:
Stick to the deadline: Because of the importance of allowing industries to make serious planning and investments based on timely standards, MPEG rigorously follows the workplan set at the beginning of each standardization project. Never has an MPEG standard been delayed in reaching IS status compared to the planned dates.
A priori standardization: To avoid becoming a standardization body endorsing industry-developed standards, MPEG identifies the maturity of technologies for standardization before industries make commitments. This approach allows the standard development process to be essentially technical and not biased by specific company interests.
Specify the minimum: In order that a standard might be useful to several industries (and not especially tuned for any of them), will technically evolve (increasing its lifetime), and will keep space for industrial competition, it is essential that only the minimum set of tools essential for interoperability is specified.
Not systems but tools: Because MPEG wants to specify standards that are useful for the various industries using the same technology, it is essential that tools rather than systems are specified, leaving the various industries the task of shaping those tools for their systems. To guide these industries, however, some major combinations of tools (profiling concept) are normatively specified so that functionality-driven and not application-driven combinations are adopted; this helps to increase interoperability and share costs.
One functionality, one tool: To reach interoperability with an acceptable level of complexity justifying industry investments, it is essential that clear choices be made regarding the best tool to provide a certain capability. Past experience has shown that the options epidemic (often engaged to satisfy companies' wishes) prevents interoperability, many times preventing the standards from flying and thus destroying everybody's work.
Relocation of tools/algorithms: So that the tools/algorithms are as useful as possible, not only must they be generic in the sense that they are not shaped to any specific application environment, it must also be possible to locate them at different positions in the final system, as different industries will likely position the same tool/algorithm differently (e.g., depending on the business model they follow).
Verification of the standard: To check that a standard delivers what it should deliver, it must be evaluated and checked against the identified requirements in the same way a product is checked against the product specification. This verification process typically involves the performance of subjective or task-based tests.
These principles have guided the development of the MPEG-1, MPEG-2, MPEG-4, and MPEG-7 standards with the same success, and they will be applied to the MPEG-21 standard development process.
1.3.3 Standards Development Approach
In MPEG, any standardization project is the result of an exploration attitude by which MPEG proactively looks for the relevant problems to be addressed—for example, emerging applications, functionalities, or even technologies. This proactive exploration attitude allows the standards to be produced in a timely way (and not too late) when applications and technology are fully mature as, by that time, proprietary solutions will be conquering the markets. For this purpose, MPEG holds open seminars to discuss relevant topics with people from industry and academia outside MPEG and creates ad hoc groups with the mandate to study the relevance in terms of standardization of specific topics.
After identifying a relevant area of interest, and in order to fulfill its objectives guided by the preceding principles, MPEG follows a standards development process with the following major steps [Chia97]:
Applications: Identify relevant applications using input from MPEG members.
Functionalities: Identify the functionalities needed by the applications.
Requirements: Describe the requirements following from the functionalities so that common requirements can be identified for different applications.
Common requirements: Identify which requirements are common across the areas of interest, and which are not common but still relevant.
Specification: Specify the tools supporting the requirements in three phases.
Call for proposals: A public call for proposals is issued, asking all interested parties to submit technology that is relevant to the identified requirements and functionalities.
Proposal evaluation: The proposals are evaluated in a well-defined, adequate, and fair evaluation process, which is published with the call for proposals. The process can entail subjective testing, objective comparison, and evaluation by experts.
Technical specification: As a result of the evaluation, the technology best addressing the requirements is selected; this technology typically does not correspond to a single proposal but to the set of the best tools extracted from all proposals. This is the start of a collaborative process to draft and improve the standard. The collaboration includes the definition and improvement of a working model, which embodies early versions of the standard and can include non-normative tools to better and more completely test the normative tools. The working model evolves through comparing different alternative tools with those already in the working model, through the so-called core experiments (CE).
Verification: Verify that the tools developed can be used to assemble the target systems and to provide the desired functionalities with an adequate level of performance. This is done by means of verification tests. Until MPEG-4, the verification tests consisted of formal subjective tests aimed at evaluating the quality of either audio or video signals processed using specific MPEG algorithms; in MPEG-4, new types of tests have been performed [N999]. In order to obtain reliable and representative results, the tests are performed using optimized assessment methods and suitable panels of subjects.
This process is not rigid; some steps may be taken more than once and iterations are sometimes needed (as happened in MPEG-4). The time schedule, however, is always closely observed by MPEG. Although all decisions are made by consensus, the process keeps a fast pace, allowing MPEG to provide good technical solutions in a timely manner.
While the period until the proposals are evaluated is called the competitive phase, the period after the evaluation is the collaborative phase. During the collaborative phase, all the MPEG members collectively improve and complete the most promising tools identified at the proposals' evaluation. The collaborative phase is the major strength of the MPEG process, as hundreds of the top experts in the world, from many companies and universities, work together for a common goal. In this context, it does not come as a surprise that this superteam traditionally achieves excellent technical results, justifying the need for most companies to at least follow the process, if direct involvement is not possible.
Two working tools play a major role in the collaborative development phase that follows the initial competitive phase: the working model and core experiments [N1375]. In MPEG-1, the (video) working model was called Simulation Model (SM); in MPEG-2, the (video) working model was called Test Model (TM); and in MPEG-4, the various working models were called Verification Models (VMs).9 In MPEG-4, there were independent VMs for the video, audio, synthetic and natural hybrid coding (SNHC), and systems developments. Regarding the MPEG-4 VMs and CEs, it is important to highlight a few points [Pere99].
1.3.3.1 Verification Models
A VM is a complete framework defined in text and with a corresponding software implementation, such that an experiment performed by multiple independent parties will produce essentially identical results. VMs are enabled to check the relative performance of different tools, as well as to improve the performance of selected tools. The MPEG-4 VMs were built after screening the proposals. The first VM (for each technical area, e.g., video, audio, and SNHC) was not the best proposal but a combination of the best tools, independent of the proposal to which they belonged. Each VM included normative and non-normative tools to create the common framework that allowed performing adequate evaluation and comparison of tools targeting the continuous improvement of the technology included in the VM. After the first VMs were established, new tools were brought to MPEG-4 and evaluated within the VMs following a core experiment procedure. The VMs evolved through versions as CEs verified the inclusion of new techniques or proved that included techniques should be substituted. At each VM version, only the best performing tools were part of the VM. If any part of a proposal was selected for inclusion in the VM, the proposer had to provide the corresponding source code for integration into the VM software in the conditions specified by MPEG.
1.3.3.2 Core Experiments
The improvement of the VMs started with a first set of CEs defined at the conclusion of proposal evaluation [N998]. The CE process allowed for multiple, independent, directly comparable experiments to determine whether a proposed tool had merit. Proposed tools targeted the substitution of a tool in one of the VMs or the direct inclusion in the VM to provide a new relevant functionality. Improvements and additions to the VMs were decided based on the results of CEs.
A CE must be completely and uniquely defined, so that the results are unambiguous. In addition to the specification of the tool to be evaluated, a CE also specifies the conditions to be used, again so the results can be compared. A CE is proposed by one or more MPEG experts and is accepted by consensus, providing that two or more independent experts agreed to perform the experiment.
It is important to realize that neither the text of the VMs nor any of the CEs ended (or will end) up in the standard itself, as they were just working tools to ease the development process. However the VMs' software is the basis for MPEG-4 Part 5: Reference Software [MPEG4-5] presented in the next section. Although it is not easy at this stage to tell how many CEs have been performed in MPEG-4, it is possible to state that they reached their goal by continuously improving and completing the technology to be included in the standard.