Transition to digital
Implementing pass-through
Chris Ward and Ray Lowe
ÿ
10/30/97
Broadcast Engineering
Copyright (c) 1997 Intertec Publishing Corporation. All rights reserved.
ÿ
Sarnoff Corporation, along with Comark, NBC, IBM, MCI, Sun Microsystems, Thomson and Philips have been working to develop a high-definition compressed digital studio. Because major differences exist between compressed and uncompressed video bitstreams, simple things, such as switching, quickly become complicated.
Highly compressed MPEG bitstreams used for transmission are not easily edited. This is due to the predictive nature of the coding schemes. However, these bitstreams can be efficiently stored and transported. Conversely, mezzanine compression levels (an intermediate level between 1.2Gb/s and 19.39Mb/s) that do not use predictive coding are easily edited but are expensive to store and transport. For these reasons, a variety of data formats are expected to co-exist within the studio.
The development effort was divided into three separate phases. Phase 1 concentrated on an architecture suitable for a local station pass-through of a network-produced HDTV compressed signal, allowing for insertion of commercials and providing for switching of MPEG -compressed transport streams at the transmission rate of 19.39Mb/s. Phase 2 added the capability for local origination of material at bit rates greater than 19.39Mb/s, as well as sophisticated routing and connectivity via asynchronous transfer mode (ATM) and high data rate satellite links. Phase 3 provides for full-production capability, with high bit-rate compression, HDTV non-linear editing and archiving, browsing for content retrieval, connectivity with external studios and transcoding between different compression formats.
Encoding, decoding and transcoding The process of transforming uncompressed digital data to a compressed form is called encoding. At present, HDTV encoders are expensive and limited in availability. One effort focused on development of a new generation of encoder to simultaneously accommodate the compressed studio requirements (by permitting frame-accurate splice-point marking) and deliver exceptional picture quality. Encoders are being developed to handle data rates up to 300Mb/s. The converse of encoding is decoding. As with the encoders, the decoders must handle data rates up to 300Mb/s.
Transcoders are used to maintain compatibility among the many different MPEG -2 studio formats, video formats and video frame rates. Within a compressed digital studio, there will be transcoders for audio only, video only or for a combination of audio, video and data. Typical applications include bit-rate transcoding (e.g., from 300Mb/s to 19.4Mb/s), picture size transcoding (e.g., from 1,920x1,080 to 1,280x720), picture format transcoding (e.g., from interlaced to progressive) and picture frame-rate transcoding (e.g., 29.97fps to 30fps). Usually, the term "format converter" is used instead of "transcoder" when the transcoding unit permits picture size, frame rate and interlace to change. Transcoders used to process combination payloads (audio, video and data) must ensure the audio, video and data alignment at the output is the same as at the input.
HDTV transmission transcoders provide over-the-air transmission of audio, video and data from an HDTV studio. These transcoders accept audio, video and data in a studio format and generate transmission bitstream at the output. Typically, the studio format would be MPEG -2 4:2:2 Profile @ High Level using either all I-frames or an alternating I, P format. For transmission, the compressed MPEG -2 bitstream must be compliant with the ATSC digital TV standard (A/53), which has a transmission bit rate of about 19.4Mb/s and may be transmitted over a single 6MHz channel using 8-VSB modulation.
Facility infrastructure Effective cable management is difficult in a large TV station. For most, the existing cable infrastructure has evolved over many years. Traditionally, components have been connected in a single wire per signal, point-to-point manner. The result is multiple cables and cable types, as well as multiple routing types. As facilities become more sophisticated, the sheer number of cables may become a problem. For the HDTV studio project, the desire was to use a single common connector for all studio devices. The ATM short-reachinterface connector supports device I/O. ATM's packetized transport allows multiplexing of many signal types on a common medium. ATM supports a hierarchy of data rates well-suited to the transport of audio and video.
ATM networks within the studio provide the flexibility to handle multiple data types and rates. Additional services may be conveniently added as they are identified, and using the Quality of Service (QoS) parameters: bandwidth, cell loss and jitter, may be controlled. Initially, a hybrid solution using OC-3 and OC-12 will be implemented. This provides maximum flexibility and cost-effectiveness for early systems, while allowing plenty of bandwidth expansion as required.
Unlike a conventional routing switcher, the compressed studio router will be an intelligent device based on commercial off-the-shelf technology. As resources are connected, a configuration dialogue occurs that permits the ATM router to know the capabilities and attributes of the particular device. This approach is the television equivalent of "plug-n-play," and permits an ever-more complex system to be dynamic and easily managed.
Asset management A distributed studio environment that focuses on the delivery of compressed video and audio bitstreams leads naturally to a server-based model. A server-centric view of the studio, rather than a tape-centric view, permits great flexibility in terms of access and cataloging of stored media. Direct-access storage devices permit metadata storage and indexing of all media within the studio. This allows sophisticated content and context-based queries, followed by immediate data retrieval. Server technology also permits the same media source to be efficiently distributed to multiple recipients simultaneously. Finally, a server-centric view of the studio maps conveniently to a networked transport infrastructure. Within the studio, there are likely to be several different classes of servers corresponding to the principal work activities within the studio in particular, a play-to-air server, network servers and production servers.
The play-to-air server would deliver content, encoded in transmission format, to air. It would also store and manipulate program and commercial segments. The server provides this dedicated function to increase system reliability. The network server would provide server support for general studio operations not directly related to production and provide an interface to library services for query and browsing. The production server would support activities associated with the creation, editing and post-production of program material. The high-quality material associated with the production and post-production services dictates that production servers be able to support mezzanine compression levels at approximately 300Mb/s. It is expected that production servers will, in addition to the general server requirements, be able to perform limited off-line transcoding and non-linear edit-type functions.
Archiving to tape storage using automated tape units will be the preferred mechanism for tape storage for many years in the future. The cost/bit of tape storage dictates this. There are, however, many exciting opportunities associated with the automated retrieval and preloading of these tapes that will require innovative solutions.
The ability to be able to browse media based on content and context is considered an essential characteristic of future studio operations. The integration of automated cataloging and image-recognition software allows for the semi-automated and fully automated cataloging and indexing of video data. This is accomplished using off-the-shelf workstation technology and proprietary browsing software.
Challenges remain Today's broadcast chain from the network through the affiliate to the viewer is complex and allows for considerable flexibility. However, moving to compressed HDTV distribution is complicated. For example, insertion of a logo (or "bug") into a program in the uncompressed domain is easily accomplished, however, this becomes complex when the pictureis in the compressed domain. Accomplishing this while still in the compressed domain is one area that deserves immediate attention.
Broadcasters and their affiliates are correctly unwilling to accept even the smallest loss of features in this new compressed digital TV era. Without a compressed solution, affiliates will be required to decode to baseband, insert the logo and re-encode for transmission, presently an expensive proposition.
The distributed network-centric studio architecture presents many opportunities. And, comprehensive compressed domain processing tools need to be developed for most of the operations that presently occur within today's compressed studio. |
