| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
The discussion on the call as I understood it was to not specify specific frequencies in the definition, as we couldn’t come up with frequencies that could
be agreed upon for high and top split. I thought we agreed that we would need to specify the frequencies in the actual specification, but did not need to do so in the definitions, which was why it was proposed (and not objected against) to write “generic”
definitions for all that left out specific frequencies. I don’t believe that the action item was recorded incorrectly. I paraphrased this action item again at the end of the meeting and there were no objections;
Marek agreed that my summary was his understanding. There are other action items to work on defining the specific range of spectrum that would be modeled and used. In summary, I think we did agree on the call to make these generic; I don’t think we are in agreement that making these generic helps. It seems to make the
definitions less clear than before. j From: Marek Hajduczenia [mailto:marek.hajduczenia@xxxxxxxxx]
Jorge,
I am a recorder in this case. I am concerned, though, that I have an AI recorded in the minutes from the meeting and now it seems that the AI was recorded incorrectly
… Marek From: Salinger, Jorge
[mailto:Jorge_Salinger@xxxxxxxxxxxxxxxxx]
Marek, From: Matthew Schmitt
[mailto:m.schmitt@xxxxxxxxxxxxx]
Jim, I completely agree. After reading through the meeting notes, I am a bit concerned that people are looking at these as requirements rather than merely as definitions to
ensure we're using consistent terminology. As a definition, they are in no way limiting or even impacting on what numbers we ultimately choose for devices to support. As long as we keep that in mind, including numbers just brings clarity (as opposed to limiting
anything). In fact, without the numbers you can argue that any split fits all definitions, which renders them effectively meaningless. I would encourage the team — like Jim — to accept Jorge's language and move on. :-) We've already spent way too much time writing emails on this topic.
;-) Thanks. Matt From:
Jim Farmer <jfarmer@xxxxxxxxxx> I’m happy with Jorge’s language. This is really not a key element of what we are trying to do; we need to put it to bed and get on with the substantive issues
before us. Thanks, jim Jim Farmer, K4BSE Chief System Architect, FTTP Solutions Aurora Networks 1220 Old Alpharetta Rd. Ste. 370 Alpharetta, GA 30005 USA 678-339-1045 (office) 678-640-0860 (mobile) From: Matthew Schmitt [mailto:m.schmitt@xxxxxxxxxxxxx]
Marek, Unfortunately, I find the "numberless" definitions more confusing. I think this is a case where using specific numbers does not impact our flexibility in any
way, because these are merely numbers and not limitations or performance requirements. So I would vote to use Jorge's language. BTW, it's "mid split" not "middle split". Thanks. Matt From:
Marek Hajduczenia <marek.hajduczenia@xxxxxx> Let’s see if the proposed changes work – I used values as examples only, keeping definitions clean from any numbers.
Let me know if this works.
Also, items marked in yellow highlight need more attention. If I do not hear any proposals for their definitions, I will remove them in the next version – I’d
rather prune than keep items we do not need. Marek From: Marek Hajduczenia [mailto:marek.hajduczenia@xxxxxxxxx]
Jorge,
The impression that I had was that we were to exclude specific numbers from these definitions … making them more flexible.
Marek From: Salinger, Jorge [mailto:Jorge_Salinger@xxxxxxxxxxxxxxxxx]
All, I think that after much discussion on the terms low, mid, high and top split, we ended up with an agreement on what to
define and what to leave out. Based on my interpretation, I modified the definitions as included below. Hopefully I did not miss any suggestions, but please correct me if I am wrong. Regards, Jorge Low split: also known as sub-split, an HFC network requiring a diplex filter, in which the upstream is transported in
spectrum below the downstream, up to 42 MHz in 6 MHz channel plan systems and below 65 MHz in 8 MHz channel plan systems Mid split: also known as extended sub-split, an HFC network requiring a diplex filter, in which the upstream is transported
in spectrum below the downstream, up to 108 MHz High split: an HFC network requiring a diplex filter, in which the upstream is transported in spectrum below the downstream,
above 108 MHz Top split: an HFC network requiring a triplex filter in which there are two upstream bands, one transported in spectrum
below the downstream with the cross-over as per either the low, mid or high splits defined above, and another transported in spectrum above the downstream. From:
Matt Schmitt <m.schmitt@xxxxxxxxxxxxx> Tom, I completely agree that what we are trying to do here is just define terms, not define new splits. I wonder if that's been part of the problem resulting
in a surprising amount of discussion. :-) The splits themselves, if they need to be codified, should be dealt with in the spec, not in a table of definitions. If we can all agree on that, I think we should be able to wrap this up fairly easily. Also, to clarify, I am talking about defining ranges for the definitions of what constitutes a sub, mid, or high split, which is what we (almost) agreed
upon until the discussion kicked back up again. By that definition — and this is the one I'm arguing for as well — your example below of an upstream band that goes up to 168 MHz would be defined as a "high split" option, because it's above 108 MHz but remains
below the forward path. So from a definition POV I think we can cover all of those possibilities. Now, as to what we define in the spec for the capabilities of EPoC devices — how high of an upstream must be supported, the degree of flexibility in
operator with less than that total amount — that's an entirely different story. It's an issue we'll need to tackle for sure, although not one we need to deal with just now. I hope that clarifies things. Thanks! Matt From:
Tom Staniec <staniecjt@xxxxxxxxx> As I understand the point of this exercise, it is not to define new splits for EPoC. It is to establish definitions for sub, European sub, mid and high splits
so everyone in the group is talking the same language when we discuss how channels definitions and operational parameters for EPoC will be defined and applied. Doing so should provide one aspect of the definition of the RF requirements for the EPoC channel which includes overall bandwidth to support 1 Gb/s in the upstream
when added to the 5 to 42 MHz which is already in use. Just assuming 1 Gb/s requires 120 MHz of EPoC channel bandwidth in the upstream adding 42 MHz to 120 MHz ends up being a minimum upstream top end frequency
of 162 MHz. Adding 6 MHz more to control group delay is 168 MHz. By CableLabs definition this doesn’t fit mid, high or for that matter any split. By the attached figure Jim sent it does fit in the high split diagram as it
does with the historic definition I put forward. If Matt Schmitt’s comment on “operator determined splits that are flexible” is true then the new split could be at 85 MHz or 125 MHz or 170.5 MHz or 300 MHz
based on what the operator deems suitable and available. If that is the case, then there is actually no definition of what constitutes a split which accurately profiles EPoC. I agree with Marek and Hal, we should discuss this on the next call.
Tom From: Hal Roberts [mailto:Hal.Roberts@xxxxxxxxx]
On the first two I think that there should be general agreement. On the high split it could be loosely defined like that. I have no opinion one way or another
there. There were also proposed other splits such as going above the downstream etc. to be defined. Perhaps there is need to define these more precisely for EPoC
as others have already proposed below. EPoC opens up an opportunity to define the new splits precisely (including transition bands) and codify them. I sense manufacturers need some limitation on
options to be able to cost effectively define products. I agree with Marek that this is a big issue and the number of emails is hard to follow. It is best to add it to the call agenda for general discussion. Even
better are to have contributions for such discussion. From: Matthew Schmitt
[mailto:m.schmitt@xxxxxxxxxxxxx] Hal, On that basis, I'd suggest we go with the previously proposed definitions, which define sub-split or low-split as going up to 42/65 MHz, mid-split as between
that and 108 MHz, and high split being an upstream that goes beyond 108 but stays below the downstream. Make sense? Thanks. Matt From:
Hal Roberts <Hal.Roberts@xxxxxxxxx> I agree with Victor. This is an opportunity to define the splits in the optimal manner for the evolved HFC platform to address future requirements and to codify
the agreed upon definitions into an IEEE standard. Old definitions should be used only if there are good reasons (i.e. the sub-split must remain at 42MHz/54MHz and the mid-split at 85MHz/108MHz due to the massive embedded base of equipment).
Splits which were never implemented in significant quantities are fair game for redefinition. From: Victor Hou [mailto:vhou@xxxxxxxxxxxx] This discussion has shown that definitions and understanding of low/mid/high split are not totally consistent within all in the industry and perhaps have changed over time. In fact, I just looked at the CableLabs glossary found at
http://www.cablelabs.com/news/glossary/#H Here are the definitions found there: Low Split Mid Split High Split I realize some of these definitions in the glossary have been around for a while and may be “old” and perhaps not exactly the way some people use these terms today. I also do not know when the last update may
have been made. I believe the NCTA site at one time also had a glossary that was derived from the CableLabs glossary but I could not find it anymore on the NCTA website. Given this is 2012 and we have the right industry minds engaging on this topic in this thread, I think we can arrive at consensus definitions for these terms.
Victor From: Matthew Schmitt
[mailto:m.schmitt@xxxxxxxxxxxxx] Jim, Thanks for the additional information — that helps. And FYI, completely understood about the caveat — it is something I am painfully aware of. :-) Because of that, I would propose that our definitions just define the terms based on the top end of the upstream transmit range, and stay silent on the size of
the transition region or the low end of the forward path for each split. Would that work purely from a definition POV? Thanks. Matt From:
Jim Farmer <jfarmer@xxxxxxxxxx> The diagram reflects splits believed to be in use or proposed use at the time of publication. Of course, the industry’s plans change faster than you can publish
a book J, so you can certainly pick apart the frequencies shown. Treat them as examples. The
only caveat, to which both Tom and I alluded, is that the width of the transition region (“no man’s land”) is proportional to the frequency, so the higher you go, the more MHz you lose to the split. And there may be some implications in the complexity of
the CNT if you want it to have flexibility – it will take someone who has looked at that problem recently to answer that question. jim Jim Farmer, K4BSE Chief System Architect, FTTP Solutions Aurora Networks 1220 Old Alpharetta Rd. Ste. 370 Alpharetta, GA 30005 USA 678-339-1045 (office) 678-640-0860 (mobile) From: Matthew Schmitt [mailto:m.schmitt@xxxxxxxxxxxxx]
Jim, FYI, the progressive split move is just one idea — my main point is that not every MSO is necessarily looking at using the same split, and so some degree of flexibility
is going to be needed. And thanks much for the diagram. One question, though: is it defining the splits to be just at very specific frequencies, or defining ranges of frequencies?
The diagram appears to be defining just specific frequencies, whereas I think that ranges make more sense given the desired flexibility. Thoughts? Thanks. Matt From:
Jim Farmer <jfarmer@xxxxxxxxxx> It is certainly possible to move the split from time to time. Each time you move it up, you open more bandwidth for the upstream, and I’m not sure with today’s
DSP and ASIC-based modulators, how the trade-off runs between complexity and the upstream bandwidth you can support. So there may be instances when you move to a higher upstream bandwidth and the older CNTs cannot follow you with more bandwidth. Of course,
presumably you could add new CNTs and take advantage of the higher bandwidth. Older units, if they tune to just above the split, would have to be able to move, but this is probably not a problem. So, yeah, I guess you could do as Matt says. Regrettably, I have to give Mr. Large credit for the figure to which Tom refers, but it is attached. jim Jim Farmer, K4BSE Chief System Architect, FTTP Solutions Aurora Networks 1220 Old Alpharetta Rd. Ste. 370 Alpharetta, GA 30005 USA 678-339-1045 (office) 678-640-0860 (mobile) From: Matthew Schmitt [mailto:m.schmitt@xxxxxxxxxxxxx]
Jim, We've been investigating the topic of moving the split with our MSO members for some time, and one approach we've investigated is a progressive movement of the
split based on market needs,. In this scenario, an MSO would first move to a mid-split (in the range of 85-105, basically as much as you can do without impacting legacy set top boxes), and then potentially follow that by the move to a high-split if demand
warranted (200 MHz is the most commonly used number, although there's been some discussion of going even higher than that). Then again, some MSOs may not move the split at all, and others might jump directly to a high-split. The actual numbers may vary,
and whatever we define will need a good degree of flexibility. But that's separate from the definitions. To me, the simplest approach to the definition —and it matches a lot of recent papers, etc. -- is to use definitions along the line of what Jorge and others discussed
below: "sub-split" is when the upstream goes up to 42/65 MHz, "mid-split" is between 65 and around 108 MHz, and "high-split" is a return path that goes above that (but still stays below the forward path). That may not completely match some historical usage
of the terms, but does seem to match current usage. More detail than that seems to just get us into trouble. :-) Thoughts? Thanks. Matt From:
Jim Farmer <jfarmer@xxxxxxxxxx> With either mid- or high-split, there should be no problem with an upper bound – the downstream band goes as far as you would like it to go. The only thing
that the split limits is the maximum upstream frequency and the minimum downstream frequency. A split in the 200 MHz region would work. Before stating something, I’d like to get inputs from MSOs as to what they want. jim Jim Farmer, K4BSE Chief System Architect, FTTP Solutions Aurora Networks 1220 Old Alpharetta Rd. Ste. 370 Alpharetta, GA 30005 USA 678-339-1045 (office) 678-640-0860 (mobile) From: Matthew Schmitt [mailto:m.schmitt@xxxxxxxxxxxxx]
FYI, we've looked at "high split" options that put the top of the upstream at 200 MHz, with 40-50 MHz of transition (although that's pretty conservative). I
want to make sure we're not excluding that. Maybe the proposed definition is getting lost across emails — would you be willing to restate your proposal completely? Thanks. Matt From:
Jim Farmer <jfarmer@xxxxxxxxxx> I imagine that an 18 MHz transition at those frequencies might be done today, maybe loosing a little to group delay. So I can support Tom’s proposal. I believe that many European systems today do use an upstream band to 65 MHz. ANSI/SCTE 174 (RFoG) acknowledges a 65/85 split as well as an 85/105 split (both
20 MHz wide at lower frequencies). So those two are also known to be used. jim Jim Farmer, K4BSE Chief System Architect, FTTP Solutions Aurora Networks 1220 Old Alpharetta Rd. Ste. 370 Alpharetta, GA 30005 USA 678-339-1045 (office) 678-640-0860 (mobile) From: Tom Staniec [mailto:staniecjt@xxxxxxxxx]
I agree Jim. I expect the crossover to actually start at 168 MHz and go up, inferentially to channel 8 so there is 12 to 18 MHz spacing. On the bottom end,
there would be 6 MHz to play with which should assuage group delay on the top end of the return band. At any rate the downstream band towards the customers should be adjusted upward in frequency depending on the amount of dead space in the crossover of the
diplex filter. Also we need to add a definition for the return in Europe which is 65 MHz if I recall. Someone correct me if I have that wrong.
Thanks Tom From: Jim Farmer [mailto:jfarmer@xxxxxxxxxx]
I tend to agree with Tom, that we can omit the hyper split. There are too many other services contending for that spectrum. MoCA wants it, MSOs are talking
about going up in frequency for downstream signals, and there is a lot of plant out there in unknown condition at those frequencies. I’d be comfortable defining high split so that 174 MHz (bottom of RF channel 7) is the beginning of the downstream band.
I’m not sure that the upstream could extend to 168 MHz – that is kind of a small crossover region. If we were to scale the traditional 54/42 low split up to end at 174 MHz, we’d need a 38 MHz crossover region. This is probably more than we need. But if
we try to tighten the crossover too much, we are either going to get into complex, hard to stabilize filters, and/or we are going to start encountering group delay that makes the frequencies close to the crossover harder to use. The problem is somewhat better
than in the analog era thanks to adaptive equalization, but there are still laws of physics that must be obeyed. Anyway, my idea is no hyper split, and a high split from, say, 144 MHz to 174 MHz. What do the MSOs want? Thanks, jim Jim Farmer, K4BSE Chief System Architect, FTTP Solutions Aurora Networks 1220 Old Alpharetta Rd. Ste. 370 Alpharetta, GA 30005 USA 678-339-1045 (office) 678-640-0860 (mobile) From: Tom Staniec
[mailto:staniecjt@xxxxxxxxx] Jorge I think we can leave hyper split out. I added because what it demonstrates is how the upstream bands were defined along “natural points” of separation from
existing defined channel boundaries. So let’s leave hyper split out.
In regard to high split, it might be wise to move the crossover up to at least the bottom edge of channel 7 (174.00 MHz). I think for EPoC this actually might work nicely for the following reason: if the return is already occupied to 42 MHz then 174MHz – 42MHz = 132 MHz of approximately
“open spectrum.” If a symmetric 1Gb/s of capacity is required for EPoC then this crossover could allow it because we would have 120 MHz available.
That assumes diplex filter band edges extend down into the last channel in the mid-band (CH 22 – band edge 168 MHz) and up to channel 8 lower band edge of
180 Mhz. That will leave 126 MHZ of usable bandwidth for an EPoC carrier of 120 MHz/1 GB/s without touching what exists in the sub-low. Obviously once that bandwidth is in place (5 to 168 MHz), it can be carved up in a myriad of ways. The above also assumes legacy set top boxes are retired so we are not trying to figure out ways to work around OOB (out of band) signals.
I offered my last email that Marek referred to only for consistency of definition. I hope this explanation works for everyone.
Tom From: Salinger, Jorge [mailto:Jorge_Salinger@xxxxxxxxxxxxxxxxx]
Tom, Thanks! I think we agree with the definition of low/sub split and mid split, right? For the definition of high split, do you think we need something other than "the cross-over is above 108 MHz"? And, do you think we need to include the definition of hyper split? I'm not sure this is necessary, but I agree that it would not hurt to include it. If so, would
you mind suggesting a definition? Thanks! Jorge From:
Tom Staniec <staniecjt@xxxxxxxxx> All Let’s try this a little differently. Sub, mid and high splits are all defined as an RF return where the upstream toward the head end or distribution hub and
away from the subscriber. Sub split is fairly clear because the top end frequency for the band pass is always 40 to 42 MHz.
Mid split is actually a little more difficult to define but generally set the top end frequency of the return band pass at 108 MHz where the cross over frequency
was in, what cable referred to as the start of the mid-band (Channel 14 – bottom edge 120 MHz). To Jorge’s point High split is a little more difficult. Also there was “one more split” which we haven’t discussed and frankly can leave alone: super or hyper
or equal (approximately assuming an RF amplifier with a 400 MHz top end). High split was typically considered to have cross over point below VHF channel 7 – 174 MHz hence the reason it started at the bottom of VHF channel 7 in the
high band. Super or Hyper – split was considered to be at or slightly below 220 MHz or above high band VHF channel 13.
As a point of note, the Super/Hyper/Equal roughly divided the upstream and downstream (toward the subscriber – away from the head end/distribution hub) as noted
with the caveat of the 400 MHz top end frequency. It might be easier to craft the definitions we use around the frequency breakdowns shown above. Then the forward frequency use in each HFC network would, in
principle, be defined by the top end frequency of the fiber node or RF amplifier top frequency in conjunction with the coax spacing. Tom From: Salinger, Jorge [mailto:Jorge_Salinger@xxxxxxxxxxxxxxxxx]
NOTE: I'm resending this Email without the diagram that Hesham had sent because the size of the Email exceeds the maximum Email size allowed by the Email reflector. John, From: Hesham ElBakoury [mailto:Hesham.ElBakoury@xxxxxxxxxx]
I found the following illustration for the 4 types of split in a presentation by ARRIS that was presented by Dan Torbit in SCTE seminar on Engineering for All
IP. Jorge was the moderator for this seminar. It seems from this illustration that the high split crossover is 200MHZ instead of 108MHz. John Chapman presentation in the last NCTA shows the following splits: [JDS: I removed a diagram from Hesham's original Email to accommodate the size limit of the EPoC Email reflector] I have seen other presentations from Cisco where high split crossover point is 200MHz (not 400MHz). Is there a universal agreement on the definition of these splits ? Thanks Hesham From: John Ulm [mailto:julm@xxxxxxxxxxxx]
The wording on the High split definition comes across as a bit awkward. I'd suggest we mirror the wording for Mid split:
Does this work?
On Sun, Aug 5, 2012 at 10:17 AM, Salinger, Jorge <Jorge_Salinger@xxxxxxxxxxxxxxxxx> wrote: Marek,
downstream, and where the cross-over between the upstream and downstream and 85 MHz in 8 MHz channel plan systems downstream, and where the cross-over between the upstream and downstream
is transported in spectrum below the downstream in spectrum above 108 MHz
two upstream bands, one transported in spectrum below the downstream with and another transported in spectrum above the downstream. -----Original Message----- Date: Sunday, August 5, 2012 8:45 AM <="" p="">
<="" p=""> <="" p=""> <="" p=""> <="" p=""> <="" p=""> <="" p=""> <="" p=""> <="" p=""> <="" p=""> <="" p=""> <="" p="">
|