Kitz Forum
Broadband Related => Broadband Technology => Topic started by: kitz on March 20, 2016, 02:14:24 PM
-
Stumbled across this (https://www.alcatel-lucent.com/products/phantom-mode) from Alcatel-Lucent which enables "speeds of 300 Mb/s over two DSL lines at 400 meters".
In just five steps Alcatel-Lucent has been able to create significant bandwidth gains. As illustrated below, starting with two physical twisted pairs, Phantom Mode creates a third virtual pair on top of the two physical pairs. Next, vectoring cancels the noise (i.e., cross-talk) generated between the physical and virtual pairs. Finally, VDSL2 bonding combines the two physical pairs and the virtual pair into a single large pipe.
300 Mb/s over 2 pairs at 400 m – Bell Labs “Phantom Mode” innovative demo 300 Mb/s in 5 steps
(https://www.alcatel-lucent.com/sites/live/files/hln-phantom-figure.png)
-
Nice find, Kitz ........ my own personal opinion is that this technology is probably impractical on a cost basis. It would mean providing more copper in the ground (as most DP's are at, or near, saturation), whereas putting fibre in the ground is a far cheaper solution.
However, it makes one think about what the pioneers of telephony might make of the true capabilities of Cu cable, were they alive today ?? Incredible science. :)
-
I notice they also show it using 0.6mm copper, not the more common 0.5mm.
-
How do I enable "phantom mode" on my VDSL connection? ???
-
How do I enable "phantom mode" on my VDSL connection? ???
you don't. You need 2 fttc lines plus the equipment at both ends needs to support it. Since it's experimental no equipment currently in use will support it.
-
Plus the extra cost for another FTTC line + line rental and me being in a high priced area that would work out at £90 a month for two lines just to get 70Mbps with bonding :o
-
It's nice to see that the techniques first used many years ago with telephony circuits are still found to be useful in the present day. ;)
-
Im sure we all get the bonding bit & vectoring to bring a line up to 200Mbps - its the Phantom Mode that that is new.
There doesnt seem to be much information about how 2 x 100 Mbps lines becomes 3 x 100 Mbps.
However, Ive just found this technical paper which basically states 3 signal paths can be created over 2 pairs.
http://www.hit.bme.hu/~jakab/edu/litr/Access/DSL/IEEE_06108362_phantom_mode.pdf
-
Plus the extra cost for another FTTC line + line rental and me being in a high priced area that would work out at £90 a month for two lines just to get 70Mbps with bonding :o
Im sure weaver would be interested in this if it could be applied to ADSL. ;D
Quite a lot of businesses use bonding, so it has the potential to increase speeds for less £
In new DSL technologies, the phantom-mode signal, defined
between two pairs, will also be exploited in addition to the
conventional differential-mode signal on one pair. In this way,
three signal paths can be defined for a quad. In this paper,
it has been shown that both the two differential-mode signals
and the phantom-mode signal are eigenmodes of the system.
Simulations also show that those three modes do not influence
each other. The measurements confirm that the crosstalk is
indeed very small between the different modes. However, it is
not negligible. Since the crosstalk is very small, this new tech-
nology will strongly increase the achievable capacity and con-
sequently support higher bandwidth applications. This increase
will be even much stronger when combining this technique with
vectoring.
-
It's nice to see that the techniques first used many years ago with telephony circuits are still found to be useful in the present day. ;)
ISDN ?
When looking for info about phantom mode and before finding the above white paper - I did see mention somewhere else saying that the technology wouldn't be wasted on copper and implying it could be used on fibre.... but Ive yet to fathom that one out. ???
----
ETA
Whilst thinking about ISDN, made me ponder on this
It would mean providing more copper in the ground
.
What percentage of lines have 2 pairs going to the premises ? So it would be an issue to the DPs?
-
It's nice to see that the techniques first used many years ago with telephony circuits are still found to be useful in the present day. ;)
ISDN ?
Long, long before ISDN. Go way back to the days when the main (trunk) network consisted of amplified audio pairs (four wire circuits, quads, with a "Go" and a "Return" pair) with repeater stations every N miles. As long as the Go and Return pairs were balanced and equalised then a phantom pair could be implemented for signalling or other purposes.
The best description I can currently lay my paws upon is the Wikipedia entry for a phantom circuit (https://en.wikipedia.org/wiki/Phantom_circuit).
I suspect that both 4candles (http://forum.kitz.co.uk/index.php?action=profile;u=672) and licquorice (http://forum.kitz.co.uk/index.php?action=profile;u=8268) could tell us about PO Telecommunications repeater stations from the more recent past.
-
Ill have 2 lines soon and am under 400m from the cabinet can I break into the cabinet and set this up please...
-
I notice they also show it using 0.6mm copper, not the more common 0.5mm.
Looking at the language used by Alcatel, I think this idea stems from the early days of vectoring: 100Mbps at 400m on 0.6mm copper. Perhaps 2010? When Alcatel pushed on commercially, in 2012, they were happy to suggest 100Mbps at 500m, on 0.5mm.
I found another article, where they update the 2-pair+phantom speed to 390Mbps, and add a 3rd and 4th pair, plus phantom, plus phantom on phantom, to reach 900Mbps on the same 400m!
http://www.ospmag.com/issue/article/The-Copper-Phantom
Yes, 2010. The lack of mention of G.Fast also dates it:
https://www.alcatel-lucent.com/press/2010/002246
Nowadays, G.Fast gets the headlines. Who would bet that Alcatel-Lucent (should I call them Nokia) were using bonding and phantom techniques for XG.Fast?
http://www.zdnet.com/article/bell-labs-achieves-10gbps-broadband-speed-on-copper/
-
Nowadays, G.Fast gets the headlines. Who would bet that Alcatel-Lucent (should I call them Nokia) were using bonding and phantom techniques for XG.Fast?
As you would expect, XG.Fast is also something BT have been playing with (http://www.ispreview.co.uk/index.php/2015/10/bt-and-alcatel-lucent-hit-5600mbps-broadband-using-xg-fast-tech.html).
I believe the attached table dates back to 2014 though as ISP Review say:
NOTE: Take Alcatel-Lucent’s table with a pinch of salt because VDSL2 / FTTC lines can operator at distances of greater than 400 meters, with well above 1,000 meters being seen in the wild.. albeit naturally at much slower speeds.
-
What percentage of lines have 2 pairs going to the premises ? So it would be an issue to the DPs?
This is what I alluded to above, kitz ...... "Nice find, Kitz ........ my own personal opinion is that this technology is probably impractical on a cost basis. It would mean providing more copper in the ground (as most DP's are at, or near, saturation), whereas putting fibre in the ground is a far cheaper solution. :)
I can't see this ever taking off on any kind of large scale, personally ??
[admin - fixed tags]
-
Is it the wires into the DP from the exchange side that are saturated? Or the wires in from the premises side? Or the number of available connection points within the DP?
I assume you mean the first one.
-
I wonder if this is similar to how my cochlear implant was able to give me 200 distinct frequencies despite only having 22 electrodes. They were referred to as virtual frequencies.
-
Is it the wires into the DP from the exchange side that are saturated? Or the wires in from the premises side? Or the number of available connection points within the DP?
I assume you mean the first one.
Your assumption is 100% correct, sir. :) :)
We can quite easily add further cable from 'DP to premises', the shortfall (should bonding become common practice ?) is from the Cab to the DP (D-side pairs) . There is usually ample spare pairs from the Exchange to the Cab (E-side pairs). Of course, there are always the odd Cab that has gone 'Bust' (Run out of pairs). :)
-
So if installing at the DP then it shouldn't be a problem, perhaps that's the way to go in the future, obviously won't be suitable in all cases.
-
So if installing at the DP then it shouldn't be a problem, perhaps that's the way to go in the future, obviously won't be suitable in all cases.
Pardon my ignorance, Ron ...... I'm not sure what you mean there, bud ??
-
So if installing at the DP then it shouldn't be a problem, perhaps that's the way to go in the future, obviously won't be suitable in all cases.
Pardon my ignorance, Ron ...... I'm not sure what you mean there, bud ??
I'm pretty sure he means that if the node is at the end of the drop wire to the house on the DP then there is plenty of copper pairs to be had for bonding, as the drop wires should mostly have 2 pairs.
If they mean running more copper to the DP from the cab, then I'm pretty sure it'd just be cheaper to run Fiber at that point.
-
S.Stephenson beat me to it.
Well most properties have at least two pairs from the DP, so if the box of tricks was located at the DP then it shouldn't be a problem using two bonded lines to do phantom mode. But of course with the diversity in the network this may not always be practical, for example if there is only a few lines at the DP or where the property is already using the two pairs for two seperate lines.
-
I see …… that's what I've been saying all along though, there would not be enough capacity AT the DP from the Cab, in the vast majority of cases. Fibre is cheaper to put in the ground than Copper …….. ergo IMHO this type of technology will not be a mainstream delivery method ? It may have it's part to play in some circumstances though ? :)
PS …… most drop-wires that have been erected in the last couple of years or so are now single pair, as opposed to dual pair. Just for info purposes.
-
S.Stephenson beat me to it.
Well most properties have at least two pairs from the DP, so if the box of tricks was located at the DP then it shouldn't be a problem using two bonded lines to do phantom mode. But of course with the diversity in the network this may not always be practical, for example if there is only a few lines at the DP or where the property is already using the two pairs for two seperate lines.
I'm thinking that if they ever put the node that close then FTTP couldn't really cost more than getting a new drop wire(£129) so I don't see why they'd pair bond at that point.
I honestly don't see the point of XG.Fast surely if someone wants the speeds offered by the spec then they'd pay a few hundred for FTTP.
-
I see …… that's what I've been saying all along though, there would not be enough capacity AT the DP from the Cab, in the vast majority of cases. Fibre is cheaper to put in the ground than Copper …….. ergo IMHO this type of technology will not be a mainstream delivery method ? It may have it's part to play in some circumstances though ? :)
But what I was getting at was from DP to EU, the cab to DP plays no part in the broadband side of things if using remote nodes at the DP ;)
PS …… most drop-wires that have been erected in the last couple of years or so are now single pair, as opposed to dual pair. Just for info purposes.
Well that would certainly cause a complication, on a side note are all new underground installs ducted?
-
I'm thinking that if they ever put the node that close then FTTP couldn't really cost more than getting a new drop wire(£129) so I don't see why they'd pair bond at that point.
I honestly don't see the point of XG.Fast surely if someone wants the speeds offered by the spec then they'd pay a few hundred for FTTP.
I can see the point of XG.Fast and its precisely that customers wouldn't pay more for FTTP if the same service/speed was available for cheaper via XG.Fast - customers are very price sensitive.
Also while replacing a single drop-wire might not be cost that much per se, logistically it would be a nightmare having to replace millions of them - when it would be so much cheaper for Openreach to install equipment at the DP and bond existing copper pairs.
-
They wouldn't have to replace millions of drop wires all at once, I was thinking it would be more based on demand, assuming that 99% of people would be satisfied with G.Fast speeds in 2020.
So at first the small percentage of people who want FTTP speeds will order it as it becomes available and then demand will rise over time until it hits the point where for BT retirement of copper make sense.
-
On our estate all phone lines are buried without ducting, I'm not even sure that there is ducting between the chambers, but there may be. Installing FTTP would be expensive as it would require digging up the whole estate and right up to peoples houses. A lot of people will not want their gardens and driveways digging up, it wouldn't bother me as I ran in a hopefully suitable conduit when I had our extension built and they only need to go across grass:-)
However if G.Fast, XG.Fast or this Phantom system was used they'd only have to install fibre to the DP/chambers and then use 1 or 2 of the existing lines to each house, I have a single buried armoured cable (I know this because it was uncovered when the extension was built) with two pairs in it, and my DP chamber is just across the road about 30 meters away.
If the nodes were close enough I'd expect the speeds available would satisfy the majority well into the next decade, with the others paying for much cheaper than today (if it were available) FTTPod because the fibre runs much closer.
-
A lot of people will not want their gardens and driveways digging up
Alcatel-Lucent regularly put a quote about this in most of their presentations on vectoring and G.Fast: "30% of FTTH subscribers change their minds when the engineer asks where he can drill holes in the wall for the fiber."
eg
http://uppersideconferences.net/g-fast-summit2014/pres-gfast2014/day_0/day_0_1_jochen_maes.pdf
I have no idea how true this is, or where they get the stats from, but I can imagine a higher backlash if gardens had to be dug.
-
I see …… that's what I've been saying all along though, there would not be enough capacity AT the DP from the Cab, in the vast majority of cases. Fibre is cheaper to put in the ground than Copper …….. ergo IMHO this type of technology will not be a mainstream delivery method ? It may have it's part to play in some circumstances though ? :)
But what I was getting at was from DP to EU, the cab to DP plays no part in the broadband side of things if using remote nodes at the DP ;)
PS …… most drop-wires that have been erected in the last couple of years or so are now single pair, as opposed to dual pair. Just for info purposes.
Well that would certainly cause a complication, on a side note are all new underground installs ducted?
Ah right, I think I've got this all wrong. I thought that the two pairs to be bonded were required all the way from the Cab, through to the premises ?? I'm not well-versed in this phantom-thingy at all ........ are you saying only one pair is needed from the Cab to the remote unit sited at the DP, and then two pairs required from there Ron ??
If this is correct, I still can't see any financial gain to be made for the masses using this method, when its competitor G.Fast finally rolls out ??
-
They wouldn't have to replace millions of drop wires all at once, I was thinking it would be more based on demand, assuming that 99% of people would be satisfied with G.Fast speeds in 2020.
That's what FTTPoD version 2 will be for.
But for Openreach to remove copper from the network, they need a mass market solution for everyone (and also to compete with Virgin etc). As such, if Openreach decide not to go to full FTTP, I imagine they will go to XG.Fast or similar in 2025 or so.. after G.Fast.
-
Ah right, I think I've got this all wrong. I thought that the two pairs to be bonded were required all the way from the Cab, through to the premises ??
I think with XG.Fast with/without bonding you would be looking at fibre to the DP/node and then copper to the premises
If this is correct, I still can't see any financial gain to be made for the masses using this method, when its competitor G.Fast finally rolls out ??
Its more of a potential successor to G.Fast if a telco didn't want to go to FTTP..
-
BS Your thinking of a traditional FTTC cab, think along the lines of g.fast which will eventually be installed further out nearer the EU. Take that another step and that equipment could be installed at each DP. I think xg.fast is basically g.fast and phantom combined.
-
Got it now, thank you. :) Technology is a pain in the assss ;) :)
-
@kitz is right, I'll sign up for a dose of phantom technology :-)
-
I think with XG.Fast with/without bonding you would be looking at fibre to the DP/node and then copper to the premises.
Fibre and power to the DP, and active equipment at the DP. Not sure how practical that is, since I'm not sure how a DP is defined - is it the point from which a given cable connects to only a single subscriber?
-
But what I was getting at was from DP to EU, the cab to DP plays no part in the broadband side of things if using remote nodes at the DP ;)
Only just seen this.. I think ron was thinking the same way I was (http://forum.kitz.co.uk/index.php/topic,17300.msg316751.html#msg316751), when I asked what % of lines had 2 pair going to the premises. I too was thinking it could be of benefit with FTTdP. This also could well have been what the other website I found which said it could be of benefit with fibre?
-
Fibre and power to the DP, and active equipment at the DP. Not sure how practical that is, since I'm not sure how a DP is defined - is it the point from which a given cable connects to only a single subscriber?
I imagine Openreach will have a solution before they roll it out. They are already testing forward power for use with G.Fast (i.e. via copper lines if they can't connect it to the grid directly) and I imagine will consider 'reverse power' at some stage..
The telegraph pole is how I usually define the DP though some technologies may have a longer range..
-
I think with XG.Fast with/without bonding you would be looking at fibre to the DP/node and then copper to the premises.
Fibre and power to the DP, and active equipment at the DP. Not sure how practical that is, since I'm not sure how a DP is defined - is it the point from which a given cable connects to only a single subscriber?
There's too many variants to give a definitive answer as to what defines a DP ......... there's overhead, underground, internal and external wall-blocks.
I would say in approx 80-90% of cases, the actual DP (Distribution Point) feeding the EU will be located within 50-80mtrs of the premises. Conversely, it can also be miles away. :)
That is why I feel G.fast will be a major success for the majority of EU's actually interested in paying for ultra-fast speeds. Of course, there will always be losers ....... that's life as they say. :)
-
FWIW this article (http://gfastnews.com/index.php/90-r/189-bonded-g-fast-35b-vplus-from-broadcom) touches on the prospects of bonded G.Fast. Seems Broadcom already has the relevant silicon shipping.
-
Can the cables that feed DPs from the cabinet be intercepted and a G.Fast node be placed as I have shown in my crudely MS Painted diagram?
My diagram isnt 100% accurate in terms of number of DP in that area, but it illustrates my point.
-
The nodes will have to placed at a suitable location where the lines all pass through for the area they want to cover, thats my understanding anyway.
-
Can the cables that feed DPs from the cabinet be intercepted and a G.Fast node be placed as I have shown in my crudely MS Painted diagram?
My diagram isnt 100% accurate in terms of number of DP in that area, but it illustrates my point.
As I understand it, Yes, it is possible. Essentially, it just needs a joint in the copper that could lie anywhere between the PCP and the DP, inclusive.
There are trade-offs, in speed vs money. The closer nodes are placed to the DP, the higher speeds you could get, but the highest number of nodes is needed, and the most effort/cost is needed to get fibre and power to the sites. The closer nodes are placed to the PCP, the lower speeds could be, but the fewest number of nodes is needed, with least cost. Each node would itself need to support more users - which adds to the complexity of supporting vectoring.
Most of the reports about BT's commitment to G.Fast suggest that they find the cost of putting nodes at the DP to be too high, and want to put them at some consolidated location slightly higher in the network, slightly further away from the user. To this end, they are encouraging chipset manufacturers to make nodes for more than the 4, 8 or 16 ports they are focussing on now, and they are taking part in the extension of G.Fast in way that increase the range from the current target of 100-150m.
The current trials seem to be, partially at least, trying to figure out the best target range to use when planning deployment.
-
As mooted above, there's no real guarantee of where the G.fast nodes will be sited ...... it's down to the 'Planners' at the end of the day. I think the key word when planning a DP location/G.Fast DPU is ..... flexibility. It has to capture the maximum amount of EU's on a costing basis.
-
They're certainly not going to be installing 4 million nodes :lol:
-
Well if it is possible then I feel very confident as if a G.Fast cabinet was at the top of my road it would place 70-80 houses within 200m of it, all the DP cables pass down the same path and I assume they have a joint at the top of the road.
G.Fast rolled out that way seems much cheaper and quicker, question is can you power a G.Fast node with a capacity of say 50 with forward power?
-
From my side of the fence, the different powering aspects are still 'In trial', and as such remain a confidential document so I can't personally comment. However there's probably enough stuff on the WWW to find an answer, or some of the resident boffins may be able to provide an answer ?? :)
-
Seems like the idea of G.Fast nodes as shown in my diagram would work perfectly in a village such as mine, it could probably work everywhere if you can tap into the cables in an easy enough way.
Thinking about it if they don't roll it out like I'm thinking then they should hire me to do their planning :P
I'm sure it's not as simple as I'm thinking, but it seems like a great solution to me less nodes and as a result less ducting work, only problem I can see with larger nodes is the powering issue, but if they can forward power 8 small nodes I'm thinking it can't be too hard to do the same for one large one.
-
From my side of the fence, the different powering aspects are still 'In trial', and as such remain a confidential document so I can't personally comment.
I have in my possession some interesting white paper type facts from IEE telecommunications and quite a lot of stuff specifically written by about 20+ bods from BT Exact and other boffins at Adastral Park. Some of the things in there are eye-opening.
Like they were trialling FTTH in Ipswich in something like the late 90's. Way back in 2004 they were already doing designs for FTTC... and wait for this even some FTTdp. My printer/scanner is broken so I cant scan to show (aside from the fact theres so much of it). They were even planning sub-loop unbundling back in 2004 until they hit problems with what was then OFTEL and possible complaints from local councils for all the additional cabs that would be required for sub-loop FTTC. iirc there were also difficulties with OFTEL because their aim of rolling out FTTx was for VoD which they wanted to do over 10 years ago.
They were very aware of the problems with FEXT - in fact they cited it being one of their biggest issues and iirc stated that FEXT in the field was actually far worse than something called the Werner FEXT model.
In there there's also an image of their old 20cn network showing just how meshed it is even before they started adding the 21CN nodes at places like Preston and Sheffield.
It just shows that you cannot compare to a 4 node ring topology like say Skys. Those guys at places like Adastral Park arent dumb and one area where BT shines at is R&D. Their R&D is one of the world's most respected and innovative places when it comes to the telco industry. *
Shame it just doesnt all get rolled out to us, so that we can benefit. I think like that with any corporate business at the end of the day, the bean counters (and OFCOM) have the final say. :(
--------
*its only when you sit down and read some of the stuff that they do, that you realise that behind closed doors, those guys are on the ball. Its one of the prime reasons why Ive always said a total openreach split would be bad for the UK. Its my firm belief that Openreach needs BT for its R&D.
-
Additional info on Adastral Park, for those interested ............
From pioneering work in optical technologies and digital switching through to work in advanced software techniques and protocols, Adastral Park is recognised as one of the leading centres of technical innovation in the communication world.
Adastral Park is BT’s global innovation and development centre. It’s also the home of Innovation Martlesham – a joint initiative by BT and Local Authorities to encourage ICT related companies to Co-locate, Collaborate and Innovate at the Park.
Key Facts
* Long history of innovation dating back to the Aircraft Experimental Unit in 1917
* BT’s global engineering HQ
* Home to 3100 BT people
* The focal point of BT Research and Development
* High tech campus with 70 technology companies
* Employing over 700 partner people
* Over 200 BT research specialists, with 65 PhDs, generating 240 patents a year
* 50 Innovation Hothouses per year with customers, suppliers and developers
* More than 50,000 non-BT visitors each year convening to discuss innovation
* 6 sector specific customer showcases
* Over 1000 VIP customers, civil servants and policy makers hosted in 200 events over the past year
* Largest test & integration facility in Europe
* UK operations centre for managing BT’s UK and global networks
* 4500 students and 920 teachers engaged from 190 schools in 2013/14
-
I think we can all agree, these guys ^^^^ aren't messing around with string and tin-cans. Respect to innovators. :)
-
I had the pleasure of visiting Adastral Park a few times over the years of my career. Always left thinking the same thing... 'Wish I worked for BT' .
A very impressive place, home to some very impressive people. :)
-
To quote the kids ...... 'Well jel' ..... I would love to pay this place a visit. In fact, I might see if there's any scope for engineers to have a look around ??
-
So, I'd guess with relates to XG.fast or anything like phantom mode applied to any of the dsl tech, and any potential deployment will depend on vendors and tech partners getting the technology on silicon and working in a convieinet package to be deployed in the field.
Domestic router and ethernet technology needs to make a leap before we start thinking more than a gigabit on the home front. 10gb ethernet is painfully expensive for most domestic users at the mo.
-
I think we can all agree, these guys ^^^^ aren't messing around with string and tin-cans. Respect to innovators. :)
Indeed, though you can see why the public are largely ignorant given how little of it has found its way to our homes and businesses so far. Outside of small trials we're running on 11+ year old technology for our broadband.
When those holding the purse strings get with the innovation programme I'll be a happy bunny :)
-
Domestic router and ethernet technology needs to make a leap before we start thinking more than a gigabit on the home front. 10gb ethernet is painfully expensive for most domestic users at the mo.
2.5Gb and 5Gb over Cat5e (http://www.ethernetalliance.org/wp-content/uploads/2015/05/Introduction-to-NGEA_FINAL.pdf) coming to a home switch and router near you this year. Access points outgrew GigE :)
-
It's very good to know that there's plenty of life left in all the cat5e cable running around my house.
-
Spotted this today, looks very interesting.
Sckipio proves G.fast is not just downstream only. I've never heard of broadband that is faster on the upstream than the downstream in 17 years reporting. G.fast can swing either way, devoting up to 90% of the total bandwidth (typically, 250-800 megabits) to either upstream or downstream. Current proposals in standards will allow the telcos to switch the ratio on the fly based on traffic demands.
http://gfastnews.com/index.php/90-r/201-750-megabits-upstream-g-fast
-
750mbps upload next to the cabinet :)
-
Interesting but I believe every line on the node has to have the same split, unless they've worked out how to compensate in real time.
-
@Chrysalis it was the ability to vary the DS/US speeds that was interesting rather than the peak speed.
Although if it's a per node basis it won't be very useful, but the article states the telcos could switch the ratio on the fly based on traffic demands, so that sounds more like a per connection basis.