HD SNG can benefit from H.264 and DVB-S2, Fujitsu pair says

A little more than a month ago, HD Technology Update learned from Intelsat that while the figure is growing — albeit slowly — only about 10 percent of all satellite newsgathering (SNG) shots are done in HD.
A short time later, Fujitsu announced a new firmware revision for the H.264 HD encoder it first publicly showed in the United States at NAB2007. Because the company has identified SNG as one of the IP-9500 MPEG-4 AVC encoder’s primary markets, it seemed like a good time to discuss where HD SNG stands with Fujitsu’s Dan Dalton and John Pittas.

HD Technology Update: What’s the latest with the H.264 HD encoder you introduced at NAB?

Dan Dalton: What we’ve delivered with the latest firmware revision to the IP-9500 is what we had talked about at NAB, a low-latency mode for HD satellite newsgathering (SNG). This provides the ability to do real-time back and forth interviews for HD SNG, with a low error rate and very high picture quality.

This functionality is key to unlocking the SNG segment for HD, and that’s what we are delivering with the firmware revision we are launching at HD World. The combination of our new low-latency mode and our improved video quality at low data rates are key to SNG. We also have implemented our 720p video for the U.S. and European broadcast market.

HD Technology Update: How will the encoder work in an HD satellite newsgathering scenario?

Dan Dalton: The two key features in SNG are low latency to allow seamless communication between reporters in the field and the network, and good video quality at low data rates to allow for efficient use of satellite uplinks.

The low latency allows the reporter at the news event to communicate with the network anchor without the long pause. With our new low-latency mode, we reduce the encode/decode latency to less than 300ms. When you include satellite uplink delay, the total latency is well less than one second.

Low data rates are important because they allow broadcasters to optimize their use of their satellite feeds. The amount of video that can be passed through the satellite is limited in terms of how much data you can send across it by the channels of the satellite transponder. For instance, if you are using a normal 54MHz transponder, you can divide that up into 10 channels, which allows 5-5.5Mb/s of video using a typical DVB-S modulator. Typical SD MPEG-2 implementations require a single transponder slot or channel. For typical HD video, the bandwidth is up to four times higher using today’s MPEG-2 compression technology. Our IP-9500 MPEG4 encoder can deliver broadcast quality video at as low as 4Mb/s, so an HD video feed can now fit in a typical SD transponder slot.

With DVB-S2, the data rate increases to 6-8Mb/s of video. The IP-9500 can deliver excellent quality HD video at those rates.

HD Technology Update: What’s it going to take to promote the growth of HD in SNG?

John Pittas: There are three axes of value that we are presenting to the client: high HD picture quality, low bit rates and low latency. We believe you need all three of those to jumpstart the HD SNG business.

We’ve implemented a very novel method for our low latency, which is a bit different than the traditional frame-based GOP structures, to get down to 300ms of end-to-end delay through the codecs, and we expect to reduce that even further in the future.

You can use DVB-S2 to get better picture quality through higher bit rates. But even with an existing DVB-S SNG channel, we can get good HD pictures at 5Mb/s, so you can use existing SD-SNG channels to do HD-SNG.

You can roll our encoder into existing SNG trucks today and get HD pictures back to the studio, whether you are doing it for a local or network feed, and whether you are going to tape or straight to air.

As you may know, one of the requirements for high picture quality in SNG is to survive multiple codec concatenation stages. We’ve had a number of clients doing tests in their labs at bit rates down to 7Mb/s with our encoder while going through five, six even seven downstream concatenation stages. They still have very good picture quality at the end, using our encoder for the space segment.

For example, an incoming feed often doesn’t go straight to air. You may have to store it on a server or record it on a VCR for editing later on an NLE system. Between the initial camera encoder and final ATSC transmitter encoder, there are at least four and possibly up to six stages of codec concatenation. Our picture quality still holds up through all those, even down into the 6-7Mb range.

HD Technology Update: Is there room for more efficiency down the road?

Dan Dalton: When we talked at NAB, that product actually started shipping in Japan, so we’ve been in production and shipping since NAB. With our current FW revision, which is the United States introduction, we have improved the video quality and added the low-latency mode.

This is all Fujitsu-developed technology. Fujitsu developed the algorithms. We are going to continue to improve those algorithms and improve their ability to make high quality video so we can make better pictures at lower data rates and reduce the latency to less than 300ms.

So, this is actually the second stage in our evolution. We see a very long road with improvements ahead of us.

HD Technology Update: What does the 300ms delay include exactly?

John Pittas: The 300ms latency is total end-to-end, encoder-to-decoder delay. That is, from the HD-SDI input on the encoder to the HD-SDI output on the decoder, basically it’s the delay between the boxes. But once you put a channel between them, you add more latency. Typically, a channel on a satellite link is going to be 250-300ms for the uplink/downlink. So, including the codecs, we are just more than a half second for end-to-end delay. This compares very favorably with existing MPEG-2 HD systems. In fact, it is lower than existing MPEG-2 HD systems. The result is the kinds of latencies we currently expect with field interviews back to the studio anchors. The IP-9500 now allows you to originate HD from the field camera all the way back to the studio, mix in HD at the studio and output your program to the ATSC transmitter, or whatever distribution platform you are using, completely in HD. No SD involved in the system.

HD Technology Update: Several weeks ago, I spoke with an Intelsat spokesperson who said that currently about 10 percent of all SNG is HD. Is that consistent with your experience?

John Pittas: HD SNG is in the embryonic stages and is exactly where we want it to be as far as a new vendor providing new capabilities to the market. I like that number because it is a huge number for us to grow into. We want to enable the HD-SNG revolution.

HD Technology Update: Why have you identified the SNG market and not the ENG market as your target?

John Pittas: We are working in the ENG part of the market also but don’t yet have an announcement. We are working with folks who do COFDM systems, so you will see the same capabilities on that side of the market eventually.

Tell us what you think!
HDTU invites response from our readers. Please submit your comments to editor@broadcastengineering.com. We'll follow up with your comments in an upcoming issue.