Kitz Forum
Broadband Related => FTTC and FTTP Issues => Topic started by: tickmike on September 18, 2014, 08:08:04 PM
-
FTTC Crosstalk.
Can you just explain a bit more about it please.
Eg Does it come from all the pairs of wires in the 'last Mile' from the Fibre cabinet all interference with one another.?
-
Crosstalk is the interference/noise in the DSLAM (Fibre Cabinet). 'Vectoring' aims to reduce this when finally implemented. :)
-
Taken from our 'Field Engineering News' (FEN) ..........
'Vectoring
One of the ways we are looking to improve speed on the broadband network is through ‘vectoring’. Vectoring dynamically calculates a “pre-compensation” factor for each line and the crosstalk can be cancelled in the DSLAM. The best example is the noise cancelling headphones that you see people wearing attached to their mobile device on every train, plane and bus journey'.
-
You can always turn up the volume on the headset to drown out the background noise to the ears the (receiver), but it's different with broadband signals as the background noise is also mixed into the signal, so it's going to need noise reduction algorithm software built into the hardware some how :-\
-
Crosstalk is the interference/noise in the DSLAM (Fibre Cabinet).
It's in the cables :hmm: 'crosstalk interference depends on the position of the pairs carrying VDSL2 signals with respect to one another' I think , this would be the pairs from the cabinet to the end user.
'Vectoring' aims to reduce this when finally implemented. :)
Found this video it explains about it.
http://www.youtube.com/watch?feature=player_embedded&v=zkVmej4urx4
-
Yes, it's the signal in one cable affecting another. This can happen anywhere between the DSLAM and the end users equipment. The proximity of one cable to another alters the effect, as does the strength and type of signal.
-
Vectoring only reduces cross-talk at the DSLAM.
-
Don't think BT will not going to do vectoring nationwide until 2016/17 as more likely BT wait until FTTC/FTTP had completed roll out first.
-
Vectoring only reduces cross-talk at the DSLAM.
But does it not take into account, and counter cross talk affecting all the line?
The OP asked where cross talk occurs, which is anywhere where there is more than one line carrying a signal in close proximity.
-
The information I've personally seen, of which part of it is posted above, states that vectoring of VDSL circuits will only be at the DSLAM. This is called NEXT (Near-end cross-talk).
If the 'noise' is then effectively wiped out at source through the application of pre-compensation factors, maybe the effects of FEXT (Far end) is minimal or negligible ?? I don't know, I'm not clued-up in the slightest with vectoring technology. I'm just re-posting part of what has come from our Chief Engineers office.
-
"An access node of a communication system is configured to control crosstalk between channels of the system.
Vectoring circuitry in the access node estimates crosstalk between channels of the system, generates a compensation matrix based on the crosstalk estimates, and generates compensated signals based on the compensation matrix.
The compensation matrix, which may be a precoder matrix or a postcoder matrix, is generated using a multiplicative update process in which a previous version of the compensation matrix comprising one or more non- zero off-diagonal elements is updated by at least one of pre-multiplying by a first auxiliary matrix and post-multiplying by a second auxiliary matrix, with a given one of the auxiliary matrices also comprising one or more non-zero off-diagonal elements.
The compensated signals may be pre-compensated signals or post-compensated signals."
The above extract is from patent WO 2012102917 A1, available for viewing here:-
http://www.google.com/patents/WO2012102917A1?cl=en
Further reading of the patent suggests that it may well be that vectoring pre-compensation is applied at the DSLAM to deal with the effects of crosstalk from many 'disturber' pairs within the whole length of the binder cable(s) (which I presume is commonly known as the D-side bundle).
-
Thanks for that reference, BE.
I used to have more info on this, even a thermal image of a 'Before and after' vectored DSLAM. I think I may have posted it on here before, but can't fully remember ? The results were dramatic.
-
I presume "access node" is just another term for DSLAM ???
-
Not sure, BE ....... not a term I've heard before ??
Being as the 'Access network nodes' are the cables and associated furniture ............ I can only assume that 'Access node' is indeed the DSLAM ???
-
I must admit when I read the reply earlier, I did wonder if somewhere down the line there may have been some confusion in the term AT the dslam.
For example impulse noise/ noise bursts / low level background noise etc can all occur and be introduced anywhere en route... But is actually corrected AT the router by use of error correction algorithms and interleaving.
-
MSAN -- Multi-Service Access Node
MSAM -- Multi-Service Access Module
DSLAM -- Digital Subscriber Line Access Multiplexer
The current deployment of FTTC with VDSL2 to the end-user uses any one of the above devices which, in the case of the first two, are configured as a basic DSLAM.
-
Useful read I found www.uknof.org.uk/uknof21/Maes-Gfast.pdf
-
Thanks Cat for the clarification. As mooted, vectoring technology only cancels cross-talk at the DSLAM, where it as it's peak.
The theory behind it appears complex and is waaaay above what I need to understand, but if any resident scientists can put it into layman's terms reading from the link tickmike has posted up, I'd be more than happy to try and get my napper round it ? :)
We know that to minimise magnetic-field induction, wires should be as near to 90 degree's to one another as possible. Obviously, that is unachievable in the D-side network, so in my tiny mind I'm wondering if the DSL signals are generated 'out of phase' with each and every other circuit in the bundle ?? Which would be similar to physically running cables at 90 degrees ?
If this is totally wrong go easy with me, it's early morning and I haven't had my brekkie yet. ;D
-
To see the effects of increasing crosstalk on a live connection, see the attached 730 day montage of the stats harvested for my connection.
In a nutshell, DS Attainable rate has reduced from around 35 Mbps to around 21 Mbps.
Actual DS Sync speed has reduced from around 30 Mbps (Interleaved) to around 21 Mbps (whenever DLM decides my connection can handle fastpath - it is usually around 18.5 Mbps these days when Interleaved).
There have also been reductions in US Attainable & actual sync speed rates, but by smaller amounts.
Due to the huge amount of data between October 2012 & the present time (around 1 million samples), the graphs' horizontal scale doesn't depict minor fluctuations too well, but I think the overall picture is quite clear.
Don't be misled by some of the 'sudden' changes in some of the graphs though.
Some of theose sudden changes were due to gradually adding the plotting of more stats data. i.e. the data simply wasn't plotted previously.
Some of the other sudden changes came with the HG612's firmware update in October 2013. i.e. data is now reported differently by the modem (more realistically??)
e.g. DS Bitswapping now seems far more effective & DS Signal attenuation is probably more realistic than as reported by the original firmware version.
Other things to take into consideration when trying to determine the effects of crosstalk are general seasonal changes in line/signal attenuation, particularly evident in the higher frequency D2 & U1 bands (I don't have enough data from a connection capable of using all the frequency bands for comparison purposes, but I imagine a similar or more pronounced effect on attenuation would be seen).
Attenuation on my connection is at its lowest during January/February than it is during August.
It appears that ecah time my connection speeds are reduced by crosstalk, I see a few days of reduced SNRM before the connection resyncs back around the target 6dB (at lower speeds).
We do experience occasional power cuts where I live.
As the HG612 modem resyncs/boots up quite quickly, I do often see higher sync speeds (up to around 25 Mbps) for a few days, but I do also see SNRM reduce dramatically & fairly quickly (down to around 1 dB) before the connection resyncs back at 'usual' speeds.
Presumably the reduced SNRM is the effect of other modems booting up again when power is restored.
DLM does appear to allow a few days of grace in these circumstances, where attainable rate is significantly lower than sync speed.
Some users see much more daily SNRM fluctuation than I see on my connection.
I live in a fairly rural location where electrical 'noise' interference is probably more static than say a heavily populated area where thousands of lights, TVs, central heating etc. will be switched on in the evenings, along with the natural daily SNRM fluctuations.
I live around 1100m from the cabinet & imagine my signal attenuation is much worse than line attenuation due to some sort of power restriction at the DSLAM to avoid my connection swamping other connections with crosstalk.
Likewise, other connections living closer to me than the cabinet will no doubt also have the same power restrictions & increased signal attenuation.
Those living closer to the cabinet won't need as much power per Mbps as I need, so they can still obtain higher speeds, even if their power has been reduced in order to reduce the effects of crosstalk.
I am not aware that our modems can provide a simple measure of crosstalk as such from snapsot data.
My observations are based on really long term monitoring trends which generally even out peaks/troughs from users going on holiday & switching their modems off etc. (to some extent).
I have been able to see this gradual deterioration of attainable & sync speeds as I was possibly the first or one of the first to be connected to FTTC from my cabinet (Plusnet got me connected & working before BT officially reported that FTTC was even available).
In summary then, it appears that we end users have to take the time of year (ambient temperature), time of day/night, possibility of other modems being switched off, rural/urban location etc. into account, using 'reasonably' sized sets of ongoing data before categorically stating how much/little we are being affected by crosstalk.
It may even be worth considering the general age group of residents in any given area as I 'imagine' that younger people are more likely to have FTTC connections than older people, partlly due to cost & partly due to less general internet interest/need from retired people living off their pensions.
As mentioned above, my comments are based on empirical data from my connection, along with more limited data that I have seen from other users' connections.
Please feel free to discuss/discredit any of the above and/or provide any factual technical details accordingly.
-
http://www.thenetworkencyclopedia.com/entry/near-end-crosstalk-next/
Published alcatel test results showed that crosstalk was inconsistent and its effectively a pair lottery however telco's can still mitigate crosstalk by high quality cabling such as bigger diameter copper and tighter twists, avoiding cross/split pairs, using low density cable bundles etc. Of course one mitigation we know openreach are doing is power cutback on the upstream, almost everyone with short FTTC distance has their U1 cutback to aid longer lines who need U1.
Based on that I sort of guess the cabling in my area is poor given the insane levels of possible crosstalk I have seen on my original pair, which went from 110 attainable down to 50. Thats over 50%, although I will never know if its all crosstalk or a fault, as the engineer found it easier to swap the pair than to diagnose further. He only managed to get 50mbit at 120m from the cab. My new pair is syncing at 73mbit roughly in line with my estimate but my belief is BT set the estimates to include predicted worst case crosstalk, which is why in early days many people get speeds of 20+ mbit above their estimate. If we assume my new pair could also get 110mbit without crosstalk then this pair is suffering about 40% loss as well.