Separate names with a comma.
Discussion in 'TiVo Series3 HDTV DVRs' started by bdraw, Jul 3, 2007.
But for broadcast, this cost can be shared across many nodes.
I don't necessarily know how SDV systems in use today are implemented, but if I were creating one, I'd be encoding the signal once and distributing it to all my nodes; each node would then only need to combine that digital signal with whatever other digital signals are going out in a single QAM channel. The node always has to have the ability to modulate every QAM channel it can use for outgoing channels. This cost is per QAM channel, not per broadcast channel. One of the test systems that put most everything on SDV only needed 8 QAM channels for each node, for instance.
The node itself does not have any modulators. It just converts from optical to electrical.
lrhorer, let's try to get down to the salient points, so that we can tell whether we're actually arguing. The original assertion was:
As I read this, he's stating that one subscriber running a sufficient number of simultaneous tuners to request all offered channels will break SDV, no matter how small the total number of subscribers on his local node. This is intuitively true: in a system using SDV, if all of the channels offered are requested by one or any combination of tuners in a single node then many requests will have to be denied. If they only request services offered as SDV, the total number of tuners necessary to break SDV should be only a fraction of the total number of channels offered. I pointed out a article in a post above (this) wherein TWC Austin was claiming to offer 175 SD channels and 8 HD ones by dynamically switching them onto a total of 8 shared QAMs. On that system, concurrently requesting any combination of 81 or fewer of those 183 channels (presumably a subset of the system total) from any one optical node will "break" SDV.
You eventually state in apparent response to the posing of this scenario (though not in direct response to Luke M's post):
This may also be true, but I'm having a hard time understanding why it's relevant. At a glance, it sounds as though you're disputing that placing such a "super-DVR" on the system will break SDV, but on closer reading you're sayng that it doesn't matter, because all offered channels will be concurrently requested from the headend across the entire system during peak periods. What does the number of channels concurrently requested by all of the subs on the system have to do with the price of peas? SDV uses the fact that no single optical node will face a demand for anything approaching that number. The main bottleneck in these systems is the coax running from those optical nodes into people's homes. No matter how much capacity of that medium might be expanded in the future (and apparently at least one source is offering tech to get 3GHz capacity out of coax), technology currently deployed in the home can only utilitize about 800 MHz of its downstream capacity; I assume that the fiber backbone of the network can carry much more. It's like a bunch of people sipping from a large stream through soda straws.
I have to acknowledge and praise the patience you've shown with the guy (i.e., not plonking him a long time ago . . . ) You are a far better man than I.
Even if that conversion from optical to electrical is a straightforward one (i.e., the individual signals can be captured in one form and re-expressed as the other without demodulating and remodulating), an optical node capable of carrying SDV not only has to be able to demodulate and modulate a signal, it has to be able to compose MPEG Transport Streams. A channel in an SDV group might be present on the fiber in a transport stream in a QAM carrier at 525 MHz (or whatever); the optical node has to strip the packets for the particular requested program from that transport stream, find space in the SDV bandwidth pool for it (which might be in a stream in another carrier at say, 615 MHz), reencode those packets to appear in that stream as a program with a different number and add stuff to tables in the security and information loops on that new stream. (It might not have to re-encode the packets in the stream to have a different program number; it seems to be becoming popular to use very sparse program numbering in transport streams, with direct correspondence ot the tuned channel number). The node further has to be able to modulate all the streams in the SDV bandwidth pool, since its creating those streams dynamically according to demand for channels in the SDV group.
I don't know how these things work precisely, but intuitively it's gotta be something like that. I'm sure that anyone who does know will feel free to correct me .
The node itself, like I said, just converts optical to electrical. It's a dumb device. It doesn't care whether it receives analog, QAM, or whatever.
The point is, there does not necessarily need to be a bank of QAM modulators for each node. They can very easily be combined at the location where the nodes are fed. Since broadcast channels are the same for every node, there's no reason why each node would need a private QAM modulator for broadcast. Hence, switched video does mean more QAM modulators are needed.
Lets put to rest this rich guy with 1000 tuner system that breaks SDV by tuning every channel. First off, they would all need to be digital tuners that utilize the Tuning Resolver. I would think that the cable companies would recognize this quite quickly.
Lets try to stay in the real world.
and what form of electrical does it get converted to?
A: It gets modulated onto QAM channels therefore there must be modulators at the node.
As I explained, this is not the case. Some people just can't understand that optical is not always baseband digital, I guess.
Very simple question for you... How does the optical get to electrical that STBs and TiVos can decode? I say the MPEG streams embedded in multicast streams get modulated onto RF QAM channels but you may have another answer.
The conversion from optical to electrical is signal format independent. If you send QAM channels into a node, you get QAM channels out. If you send analog in, you get analog out.
The Hybrid Fiber-Coax architecture was conceived and implemented before digital cable even existed. It was all analog originally. So naturally, the optical input needed to be an analog (broadband) signal. This is in contrast to the "normal" use of optical signals in telecommunications, which is baseband digital.
Does that clear it up?
So I believe you are saying that the dark fiber is basically carrying multiple frequency shifted chunks of spectra which are already modulated QAM channels. The node 'hetrodynes/frequency translates' them to the appropriate RF frequencies. Right?
If so... this would suggest that the 'node' in SDV terms isnt all in one place... the frequency translation points would be distributed wherever they need to be to serve their assigned 'neighborhood' while the heavy lifting of building the transport streams and QAM modulation is somewhere else (like the headend?).
No, I'm saying that the node is passing the signal unmodified. All the intelligence is upstream of the node.
Hmmm. I'd thought that switching was being done on the level of the optical node at the optical node, but examination of the literature online about specific switched digital video technologies (in particular, BigBand Network's stuff, apparently the most widely deployed at the moment, the switching equipment lives at the distribution hub level (see this popular Wikipedia diagram); each distribution hub sits on an optical transport ring with the headend and has continuous access to all of the offered content, fanning that out over private sets of fibers to each of the optical nodes. Those customized sets of SDV streams (possibly mixed with VOD traffic) are composed in that equipment. Makes sense, and simplifies network maintenance greatly.
From what I just read, every channel starts out in the headend as an RF signal, converted to optical and then converted back by the optical nodes for transmission over "last mile" coax. Some of them are 6 MHz NTSC analog signals, some of them broadband digital streams.
The frequency range and capacity of the fiber is much higher than the freq range of the coax so there must be some conversion going on. The channels need to get mapped into the RF frequencies that the cable STB can understand. Once again, I say that the edge QAMs in the node select the proper multicast stream based on current usage patterns and modulate the selected MPEG stream onto the RF QAM channel. You and I may never see eye to eye on this. Simply saying that fiber is the same as electrical does not make sense to me.
Please don't invent nonsense and attribute it to me. Thanks.
This is what you said...
which implies you can take RF QAM at a central location and jam it into a fiber then suck it back out on the other end and push it back into a coax with no additional modulation at the node.
Anyway, if you have any hope of convincing people that there are no modulators at the node you should be able to answer my question about how you map the fiber optic signals/frequencies to the 800mhz RF spectrum. You keep skirting this issue because you can't answer it and know that you are about as wrong as can be.
Yes, that's how HFC works.
Instead of tossing around insults, why don't you spend a couple minutes reading up on HFC?