The importance of monitoring throughout the MPEG signal chain
NEW YORK
Once a baseband HD or SD signal plus associated audio and data get converted into an ASI stream, it becomes necessary to use specialized measurement and analysis tools to ensure that the MPEG-2 stream and all its elements are present, in the correct location, and with the correct timing.
"With the set-top box being cheaper and cheaper and from an increasing number of manufacturers from all over the world, there will be some interoperability issues," said Stephane Billat, communication division application engineer for Sioux Falls, S.D.-based Sencore. "The only way for a broadcaster to feel confident about its DTV delivery is to provide a compliant transport stream." And that means testing.
Billat said that MPEG monitoring can be separated into three areas: syntax compliance of the MPEG-2 transport stream, template checking for the presence of audio and video services, PID (packet identifier), and customer-specific parameters like minimum bit rate for minimum quality purposes.
Observing an MPEG signal with a particular set-top box on a particular monitor or receiver doesn't guarantee that the viewer at home will have trouble-free reception. And observing only at the end of the signal chain doesn't guarantee success either. It's important to monitor at all points in the MPEG signal chain. These include satellite or fiber contribution feeds, video servers or tape-based devices producing MPEG streams, encoders, IRDs, any re-multiplexers, PSIP generators, data generators, modulation outputs, and cable TV output to the set-top box.
DIGITAL DEFECTS
What are some of the more common DTV defects?
Rich Chernock, director of technology for Triveni Digital in Princeton Junction, N.J., listed the ones he's observed the most: PSI/SI/PSIP tables missing, incorrectly formatted, incomplete and/or inconsistent; excessive jitter in PCR values; audio or video buffer underflow or overflow; audio or video program elements missing; and incorrect audio/video synchronization.
The causes of these defects range from initial setup and configuration; equipment drift; equipment failures; communication link failures; and loss of synchronization.
"Having the buffer off just a little bit is fairly common, and often a TV station won't know this is happening until they do the analysis," Chernock said. "This is usually related to the setup of the encoder."
What happens if these defects are present? Chernock said that DTV receivers could have trouble tuning. There could be no information in the on-screen program guide. The picture or sound could break up or be absent, and there could be noticeable lip sync errors.
So to prevent these defects from occurring, what needs to be monitored?
"The simplest answer is to follow the ETSI TR 101 290 Measurement Guidelines that originated from DVB," said Dennis W Kucera, an MPEG applications engineer for Beaverton, Ore.-based Tektronix. These measure various parameters of the transport stream.
"Compliance to this guideline helps to ensure that set-top boxes will be able to receive the transmitted signal, accurately decode the video and audio, and access and make use of the embedded electronic program guide," Kucera said. "Set-top boxes are designed under the premise that these guidelines will be met and failure to comply could mean that a set-top box will fail to operate."
As TR 101 290 was developed for DVB, there are some ATSC-specific tests that this document doesn't address. To rectify this situation, ATSC recently established two sub-groups that will provide recommendations on what to monitor in the ATSC realm, in a similar way as TR 101 290. The DVB document will be used as a starting point, with ATSC specific tests added.
The SEWG (The ATSC Implementation Subcommittee Systems Evaluation Working Group) has created a sub group called DTV Stream Monitoring Points and Parameters chaired by Billat. In addition, the TSG Ad Hoc Committee on ATSC Bitstream Verification, chaired by Chernock, "should explicitly describe the elements and parameters of A/53 and A/65 that must be verified in an ATSC Transport Stream for it to be considered a proper emission," according to its mission statement.
PRIORITY LEVELS
TR 101 290 is divided into three priority levels or groups: 1, 2, and 3.
The first priority includes parameters that are essential for the MPEG stream to be decoded properly.
These include transport stream sync loss, errors in sync byte, PAT (Program Association Table), PMT (Program Map Table), continuity count, and PID (packet identifier). Proposed ATSC parameter checks for priority 1 include MGT (Master Guide Table) and VCT (Virtual Channel Table) errors as well as PAT, PMT, MGT and VCT consistency errors. (The list of parameters for all three levels was supplied by Triveni Digital and Sencore.)
"The most important parameter of all is the MPEG-2 0x47 Packet Sync Byte which should always be present without error," Kucera said. "Secondly, the PAT and PMT tables must be present, because without these tables, most set-top boxes will fail to find anything to decode. ATSC adds to these tables with Program and System Information Protocol, which is a major requirement for ATSC set-top boxes, allowing them to identify and decode video and audio."
Kucera further explained the function and importance of the PAT and PID.
An MPEG-2 program uses time-division multiplexing to deliver video and audio over the same link. MPEG-2 also supports multiple programs on the same link. Both cases require the use of a PAT for describing the number of programs carried within the link, such as an ASI stream. Every MPEG-2 transport packet header carries a 13-bit parameter called a Packet Identifier. This PID parameter allows the receiver to differentiate all of the tables and elements. The PID allows for the unique identification of the PAT, each of the different video and audio elements, and the other MPEG-2 PSI tables, like PMT, as well as ATSC PSIP tables.
Level 2 priority lists recommended parameters that need to be monitored continuously or periodically. These are the ones that can affect overall program quality.
Repetition rates for many of these parameters must be within certain tolerances. For example, "if the repetition rate for PSIP is too slow, the tuning and channel surfing will be slow," Billat said.
Level 2 includes tests for errors in transport, CRC (Cyclic Redundancy Check), PCR (Program Clock Reference) jitter and discontinuity check, PTS (Presentation Time-Stamp), and CAT (Conditional Access Table). Proposed for ATSC are detection of the presence of EIT-0 to EIT-3 (Event Information Table) and checks missing required descriptors, RRT (Rating Region Table) if referred in the MGT, errors in the SST (System Time Table), and descriptor consistency error in the PMT and EIT-0 (for the current event), table syntax, PID, table type and version number.
"The CAT is the pointer to enable the IRD to find the EMMs (Entitlement Management Message) associated with the conditional access system that it uses," Billat said. "If the CAT is not present, the receiver is not able to receive management messages."
The third level of priority deals with application-specific data, such as audio and video buffer level and SI tables, Chernock said.
These include: buffer error, audio and video buffer underflow or overflow; un-referenced PIDs, and proposed for ATSC, errors in optional EITs from EIT-4 to EIT-127 if referred in MGT; plus errors in ETT if referred in MGT, A/90 datacasting table, as well as consistency errors among MGT, VCT, EIT and ETT.
BIZARRE BEHAVIOR
Inconsistency between tables can cause bizarre behavior in receivers. "Some will work fine, some will tune to somewhere else, and others don't work at all," Chernock said. These problems can be difficult to trace, which is why it's important to perform cross-table analysis.
In ATSC, some parameters need to appear in different tables. For example, the captioning service descriptor tells the set-top box what kind of captioning information is available to make decoding easier. This data is located in two places, the PMT and the EIT. If the two tables disagree, it's uncertain how a set-top box will react.
The tables also use pointers from one to another. For example, the Event Information Table, which carries the schedule, has a pointer from each event to the Extended Text Table that carries a description of the show. If the ETT doesn't exist, or if the information in it doesn't correspond to the event itself, that's a problem.
There's a wide variety of equipment available for performing ASI bitstream monitoring and analysis. These include the Streamscope analyzer (including cross-table analysis) from Triveni Digital, the TSM 1770 (monitor) and MSA 1850 (analyzer) from Sencore, the MTM400 and MTS400 from Tektronix, the Thales Mercury MPEG Analyzer and Garnet MPEG-2 handheld analyzer, DVStation from Pixelmetrix, and DVM 400 from Rohde & Schwarz.
A good tip is to actually try out these analyzers. They make good learning tools on the inner workings of ATSC. Chernock, who travels around the country teaching seminars and tutorials, finds that when his students start to use the measurement tools, they begin to understand what all the various tables do and how they interrelate to each other.
The nomenclature and all the acronyms can be confusing, so a visit to the ATSC Web site at www.atsc.org to download and read the standards, would be time well spent.
And if there is still any doubt about the need to monitor, Chernock can dispel that. "Why monitor? Because you really want to find out the problem before the viewers do." And before they start calling with complaints.
