Kitz Forum

Broadband Related => Broadband Technology => Topic started by: Weaver on September 20, 2019, 10:37:52 PM

Title: Long distance fibre
Post by: Weaver on September 20, 2019, 10:37:52 PM
[Apologies sincerely if I have asked this before; it’s been bugging me for a long time but I can’t remember whether or not I have already actually come out and asked anyone ]

How does a fibre link get to wherever - either Inverness or Fort William - civilisation anyway - from NSBFD the Broadford or Kyle exchange ? It’s 82 or 87 miles respectively by road. Can it be done in one hop ?

And if not what do they do then ? Have to find a location somewhere where there is power available locally, for active repeater kit ? Or send power down the cable ? Or even generate power at some intermediate node, by solar and wind combined, or something ?
Title: Re: Long distance fibre
Post by: burakkucat on September 21, 2019, 12:45:39 AM
Interesting question.

Obviously I can't "speak" for Beattie Bellman but I wouldn't be surprised if one of her "escapees" could provide some distance figures, for single mode fibre.

Then, of course, we have CarlT who, I'm sure, would also be able to give some typical distances for the various types of fibres.

A quick search finds a table (https://www.universalnetworks.co.uk/faq/fibre-optic/what-are-achievable-distances-single-mode-vs-multi-mode-fibre) by the supplier Universal Networks (https://www.universalnetworks.co.uk/).

Looking at the Submarine Cable Map (https://www.submarinecablemap.com/) and considering, say, TAT-14 (https://www.tat-14.com/tat14/) I wonder how that 15,295 km is covered? I read --

Quote
The cable system is comprised of four fiber (sic) pairs and traffic can be configured point to point or in (protected) ring configuration.

TAT-14 has a total design capacity of 9.38Tb/s.
TAT-14 has a total system (lit) capacity of 3.15Tb/s.

(https://www.tat-14.com/tat14/images/SprintMap1.jpeg)
Title: Re: Long distance fibre
Post by: Weaver on September 21, 2019, 01:02:07 AM
That is an active cable ? It has power sent down it to active inline repeaters ?

I wonder what the limit is without any repeaters.

And I wonder which type is used on land in the highlands. In the BT undersea cable laying orgy between islands a couple of years back, the chap from BT who gave a lecture about the project said that it was exclusively passive cable. The reason behind that decision was he said safety; they were worried about a fisherman hooking the cable, lifting it up and getting a jolt of god knows how many volts for his trouble iirc. I presume passive cable is cheaper too, obviously, but I don’t know how much. Presumably they could get away with it in that project because none of the inter-island distances are that great. It’s not that far from Skye to Uibhist or to Harris, and that has to be the longest hop that they had to deal with in that project. Mind you I don’t know about Orkney to Shetland which might well be via Fair Isle which would halve the link and thus halve the problem. I don’t know how undersea cables get to Shetland; I don’t even know if the cable path comes from Orkney, which would be the same route, or whether it comes directly from mainland Britain.
Title: Re: Long distance fibre
Post by: burakkucat on September 21, 2019, 01:18:26 AM
Looking at the table, mentioned above, we see --

(https://www.universalnetworks.co.uk/media/wysiwyg/images/Fibre_Distances_1.png)

I think 100 km is about 62.5 miles and 40 km is about 25 miles. Hmm . . . I'm intrigued.

Title: Re: Long distance fibre
Post by: Weaver on September 21, 2019, 01:36:13 AM
Distances:

Orkney to Shetland :
49 mi direct from Orkney (North Ronaldsay) to Shetland
27 mi from Orkney (North Ronaldsay) to Fair Isle mid-point
24 mi from mid-point (Fair Isle) to Shetland

15 mi from Skye to Western Isles (to either Harris or to Uibhist)

So 15 mi must be ok for passive but I don’t know about the 27 mi hop to Fair Isle or the 49 mi hop Orkney to Shetland direct. From Burakkucat’s table 40 km is 24.85 mi and so even the shortest of the latter hops is out. It seems that Fair Isle even will not be enough to help. And Skye to Inverness or Fort William means big trouble too.



As for Skye to Inverness or Fort William :

Broadford (Skye) to Inbhir Garadh in the Great Glen would be one possible midpoint but that is still 57 mi and so no good for a 40km limit. Would need additional hops. And there’s the rest of the run, either 23 mi up the Great Glen from Inbhir Mhoireastain by Loch Ness into Inverness itself or 25 mi down the Great Glen into Fort William. So these two latter ‘halves’ might be just about ok.

The solution would have to be a node at the Cluanaidh Inn, which is in the middle of nowhere, on the main A87 road, has mains of course and is about 21 mi from the Great Glen as the crow flies so even that is pushing it for a 40km limit as the road distance will be somewhat greater.
Title: Re: Long distance fibre
Post by: burakkucat on September 21, 2019, 02:09:25 AM
Seeing "Inbhir Garadh" had me scratching my head. Then I had a sudden thought - Google Maps will know. And it showed me Invergarry (https://www.google.com/maps/place/Invergarry+PH35+4HB/@57.0709505,-4.800261,9z/data=!4m5!3m4!1s0x488ebc9847ccf159:0xfdbef5bb0e390898!8m2!3d57.070951!4d-4.800261).  :)
Title: Re: Long distance fibre
Post by: Weaver on September 21, 2019, 02:31:17 AM
That place has various forms and I’m unsure a to which is to be preferred as I haven’t met any local speakers from the area. Inbhir Gàraidh or Inbhir Gearraidh and a medley of possibilities derived from these provide yet more alternatives. Speakers could have been confused by similar words gàradh ‘dike’, ‘garden’ and gearradh ‘cut’, ‘cutting’ from the verb for cut, also ‘summer grazing place for cattle’. The former has a long vowel. The latter has an initial /gj/-like sound. There is also a possible choice between genitive singular and genitive plural forms. Combine all the lot together and the range of possibilities is daunting.
Title: Re: Long distance fibre
Post by: Weaver on September 21, 2019, 03:00:09 AM
For some reason, unlike in Wales which has a much better track record with maps, maps of Scotland tend to still show anglicised forms on the mainland whose mangling makes placenames hard to decipher. Some maps are doing rather better in the Western Isles but have yet to get their act together on the mainland and in Skye. Many many obvious mistakes, misspellings and grammatical errors abound which would not be found in Wales and no maps of England would dare exhibit misspellings such as ‘*Burminghum’. Recently the authorities aided by the Scottish Placename Society, of which I am a member, and other bodies have done a good job getting all the road signs fixed but trying to sort out all the various maps is a Herculean task.

In many parts of the country the placenames are Norse (Old Norwegian) mangled by Gaelic speakers or from British (ie Welsh), not Gaelic and some have been ‘gaelicised’ in spelling. For these names it seems anything goes but now the tendency is to increasingly spell these names with phonetic Gaelic spelling based on the pronunciation of local Gaelic speakers, which is a sensible choice.

English speaking readers have to cope with the complex coding system which Scottish Gaelic and Irish use to encode far too many consonants into the few symbols provided by the Latin alphabet. Welsh does not have this problem because it has only approximately half as many consonants. The letter h after a consonant can be present or absent and one of two sets of extra vowels are used to bracket consonants - the sets are a/o/u and e/i, one or the other - and this choice between the two sets, combined with the +/- h option gives four possible consonant sounds per core consonant letter. To prevent ambiguity the vowels on either side of any consonant must be from the same set.
Vowels can be short or long and in traditional Gaelic spelling there can be à ì ù and è ò | é ó, the accents are used to mark long vowels and the choice of grave vs acute for e or o gives different sounds. So the number is sounds is massive, Gaelic spelling is a code, and like utf-8 it needs decoding first; in the code, multiple letters convert into single sounds. Compare English mac and mace where the addition of the silent e changes the consonant and the vowel. The difficulty for an English-speaking audience perhaps explains some of the differences in historical map-making practice between Wales and Scotland.
Title: Re: Long distance fibre
Post by: niemand on September 21, 2019, 08:57:23 AM
You can get between 80 and 120 km out of a 10GBASE-ZX optical link.

That's just a basic SFP you plug into a router or switch. If you put 'real' transmission equipment either side you can get reach of over 400 km without amplification, regeneration or using anything experimental depending how many wavelengths you need the fibre to carry.
Title: Re: Long distance fibre
Post by: Weaver on September 21, 2019, 01:31:23 PM
That answers my questions then. With 400km there’s no issue anywhere. Thank you for that Carl.

I wonder if there are any maps around showing large network operators’ geography in Britain ?
Title: Re: Long distance fibre
Post by: niemand on September 21, 2019, 02:32:25 PM
https://www.telecomramblings.com/network-maps/europe/

Out of date. BT's full network map is I suspect not just a company confidential thing but also a matter of national security.
Title: Re: Long distance fibre
Post by: j0hn on September 21, 2019, 02:44:28 PM
Is this the sort of thing you're looking for?
Title: Re: Long distance fibre
Post by: Weaver on September 21, 2019, 03:21:30 PM
@CarlT good point. A shame though as academic researchers could benefit from this kind of info for research purposes; would give them a concrete basis for modelling.

@j0hn brilliant thank you very much indeed. Was that from the time of the undersea cabling project a couple of years ago? I should have remembered that that might be a likely source of maps but I forgot that the maps included on-land links, which they very likely upgraded hugely; the run right up the west coast including the likes of Geàrrloch in Wester Ross and settlements much further north up towards cape wrath and along the north coast even, to complete a ring heading towards hops across to Orkney.

BT did a superb job, including a hop across the sea from Skye to Malaig which has the benefit of a very short run to Fort William, ensures Malaig isn’t left out, and improves triangulation redundancy for Skye. My missus suggested that Malaig might be a route for BT from Skye to Fort William rather than trying to go from the Skye Bridge to the Great Glen.



Any one have a similar map for England ?
Title: Re: Long distance fibre
Post by: Weaver on January 07, 2020, 12:07:39 AM
A general question though. If you are BT and you want to run a new link from say London to Birmingham or to Sheffield or to Edinburgh, where physically do you run it?

Do you use canals, sewers? Follow roads? You don’t just follow main roads up to those places because it would mean disruption at every side junction / slip road ?

Do you find existing long distance large ducting and add to that, hoping it isn’t already full? What if ducts are completely full though and then you do need new additional ducts over distances of several hundred miles?
Title: Re: Long distance fibre
Post by: niemand on January 07, 2020, 12:44:24 AM
You use existing ducts, repairing where necessary.

You can get a lot of data down the fibre you can fit in ducts that used to have huge, larger core means lower attenuation, coaxial cables in them.

https://en.wikipedia.org/wiki/Coaxial_cable for discussion on coaxial cable properties. Long distance ones were huge.

As far as main roads go BT can and do work down A roads; motorways and the odd trunk AxM roads that should be motorways not so much, however the roads that linked our towns and cities before those were built certainly, alongside ring roads and the like.

Worst case roads can have a lane closed, and at junctions they can be closed overnight, or can be directionally drilled under.

Here in Leeds CityFibre are doing some directional drilling under the A653, BT did repeated lane closure to work on their chambers on that same road: as the road was built so the BT network was built underneath it and that's fairly normal.

While I'm at it don't forget the train lines, too. CenturyLink I believe have the rights on that one and fibre can be leased from them.
Title: Re: Long distance fibre
Post by: niemand on January 07, 2020, 01:00:37 AM
Sorry, last thought.

Remember also that the fibre runs will be very indirect. They'll stop off at a bunch of places to drop off traffic or be regenerated.

The coaxial stuff needed frequent amplification and fibre following those routes will stop off plenty, even if it's just to go through a patch panel rather than being regenerated.

Plenty of options to take diversions along duct paths, too. For historic reasons there is a lot of resilience on the BT network: UKWMO used it as did the WBxxxx siren warning transmission systems.
Title: Re: Long distance fibre
Post by: Weaver on January 07, 2020, 06:43:52 AM
Would you always use roads or railways rather than going through the middle of nowhere? Because of the hassle of securing permission from farmers so very many times? Roads can be very wiggly, so a long distance route could get longer than the straight line route.

What else is there ? Canals? Long distance big pipelines - is that correct?

I thought I remembered something about B4RN and a pipeline giving them a huge backbone when they got started, but I can’t find it so I must be imagining things.

If you’re in the middle of nowhere though like B4RN you might get lucky and they have the M6 so they’re extremely lucky.

I was reading about Bogons the ISP and their Cold War bunker in Perthshire (https://www.bbc.co.uk/news/uk-scotland-tayside-central-27501689). They needed a fibre optic massive link to Edinburgh (http://comriedevelopmenttrust.org.uk/big-bandwidth-at-the-bunker) and Openreach hooked Bogons up with a 100Gbps link. I would love to know just exactly how that was physically done. Existing ducts? Upgrades of existing paths?


The train lines I suppose can be a great asset, accessible and straight, so the shortest distance, and therefore they minimise latency.

If you need to switch sides, or you have a problem because there is a line branching off on your side, then you have to go under the track.

In the latter case, will there often be an existing (speculative) duct in place ready, just for this purpose ?

If no existing duct, can we easily just sneak under, by scooping out a little material under the rail and then replacing some of it as needed?

Can you buy the rights to put your own fibre into the railway line?
Title: Re: Long distance fibre
Post by: niemand on January 07, 2020, 10:36:40 AM
Would you always use roads or railways rather than going through the middle of nowhere? Because of the hassle of securing permission from farmers so very many times? Roads can be very wiggly, so a long distance route could get longer than the straight line route.

What else is there ? Canals? Long distance big pipelines - is that correct?

Canals are largely already done and Sky now own that network via Easynet, who in turn purchased Fibreway (https://www.independent.co.uk/news/business/fibreway-opens-new-trade-on-canals-1428006.html).

Openreach usually avoid going across private land like the plague - wayleaves are an ass unless you're B4RN and can blag them for free :)

A reminder also you'll need to get access both to install and to repair where necessary, so keeping this stuff reasonably close to roads is useful. There will be occasions where the fibre goes on poles and takes a bit of a weird route.

I was reading about Bogons the ISP and their Cold War bunker in Perthshire (https://www.bbc.co.uk/news/uk-scotland-tayside-central-27501689). They needed a [url[http://comriedevelopmenttrust.org.uk/big-bandwidth-at-the-bunker]fibre optic massive link to Edinburgh[/url] and Openreach hooked Bogons up with a 100Gbps link. I would love to know just exactly how that was physically done. Existing ducts? Upgrades of existing paths?

Existing duct / paths, some new fibre and some new digging. Any and all of the above - Excess Construction Charges are there for a reason  :)

The train lines I suppose can be a great asset, accessible and straight, so the shortest distance, and therefore they minimise latency.

If you need to switch sides, or you have a problem because there is a line branching off on your side, then you have to go under the track.

In the latter case, will there often be an existing (speculative) duct in place ready, just for this purpose ?

If no existing duct, can we easily just sneak under, by scooping out a little material under the rail and then replacing some of it as needed?

Can you buy the rights to put your own fibre into the railway line?

I'm not that familiar with the train lines beyond that Network Rail can be a real pain to deal with and there is an extensive fibre network running alongside them. This was sold off alongside the railways. I was, however, wrong about who owned them - it's actually Level 3 on the core and Thales on the other stuff.

The terms and conditions of this arrangement I'm not aware of, neither am I aware of what would be required to cross them. There are certainly ducts there though you'd need to lease space in them. To go underneath them you'd be looking at directional drilling and it would be incredibly expensive.
Title: Re: Long distance fibre
Post by: Black Sheep on January 07, 2020, 05:38:29 PM
Planning FTTP,  for the vast majority of the time, follows the same route its Cu/Ali cousins take.

As Carl points out, there are a plethora of innovations available to use to avoid disruption and also mentions the retrospective access to the network for repairs, which is why Aggs/Spliiters/CBT are planned the best they can be to be sited in easily accessible boxes (rather than carriageway boxes, or anywhere TM - traffic Management - may be needed).

Please also bear in mind, the books are constantly being re-written with regard to FTTP deployment as innovations get better all the time, government backing also gets bigger (especially where wayleaves and MDU's are concerned), and of course its is  a relatively 'new product' for all concerned ....

 

 
Title: Re: Long distance fibre
Post by: Weaver on January 07, 2020, 10:42:55 PM
I’ve been reading and horizontal drilling and trenchless in general. Fascinating stuff. I’m very interesting in the drillhead location tech stuff. I don’t know though how you close the last bit of distance when you know roughly where your two ends are but you’re not close enough. And I don’t know how you know when your ends have met up. I’d like to know a lot more about exactly how all that is done.

Actually, I’ve just realised: perhaps you only have one end, and you locate that then drill straight down to it, but that means that you have not halved the maximum distance that you can cross, you don’t have two mutually meeting halves, ie two mutually meeting drill lengths, meeting under the road or whatever it is you’re trying to pass underneath of. That solves the problem of locating two ends, by only having one ‘end’.


A thought: Carl was talking about Fibreway on canal towpaths. That ought to make them exceptionally reliable, no? Because there will be pretty much zero problems with other minidigger men and women wrecking your fibre with their alien works. Does that sound plausible?

I read an article about Fibreway from 1994 [!] - pretty much the Big Bang as far as internet history is concerned.
Title: Re: Long distance fibre
Post by: licquorice on January 08, 2020, 08:46:20 AM
You have to remember that FTTP distribution fibre and core network fibre are 2 entirely different animals.

In days gone by, there were effectively 3 (or 4 if you differentiate between cabinet E side and D side cables) levels of cable distribution.

Local network from the exchange to individual premises via cabinets and DPs, Junction cables linking local exchanges to a larger switching centre and Trunk cables linking major switching centres.

The Trunk and Junction network would have been ducted and followed roads closely and that duct network will still be used today for the fibre core network.
Title: Re: Long distance fibre
Post by: Weaver on January 08, 2020, 09:18:53 AM
If you have an existing duct, how do you upgrade link capacity? You pray to the gods that you can cram more fibres into it, but the gods don’t hear you and the duct is really really full. Can you try and cram more wavelengths down the fibres that you already have?

Does it ever happen that you remove some rubbish fibre to free up the duct and put in better fibre instead?
Title: Re: Long distance fibre
Post by: licquorice on January 08, 2020, 02:19:31 PM
For ex Trunk and Junction routes, if the old copper cables have been removed, there will be ample duct room. The cables will be multi fibre which may not all be lit and extra wavelengths can be utilised in DWDM systems. I'm not sure what current technology offers (possibly 16Tb/s) but when I retired 12 years ago the core network cables were capable of carrying around 2Tb/s per fibre.
Title: Re: Long distance fibre
Post by: Black Sheep on January 08, 2020, 05:43:06 PM
For ex Trunk and Junction routes, if the old copper cables have been removed, there will be ample duct room. The cables will be multi fibre which may not all be lit and extra wavelengths can be utilised in DWDM systems. I'm not sure what current technology offers (possibly 16Tb/s) but when I retired 12 years ago the core network cables were capable of carrying around 2Tb/s per fibre.

I've had a very quick search that mentions in June 2017 they managed 13 terabits per second over a single fibre, on a live 250mile round trip circuit.

It could well be more now, as you say Licq ??  :)
Title: Re: Long distance fibre
Post by: Weaver on January 08, 2020, 05:50:46 PM
Woa - 2Tb/s will keep you going. Isn’t that roughly about the half of LINX or LONAP peak traffic (https://portal.linx.net/)?
Title: Re: Long distance fibre
Post by: Weaver on January 09, 2020, 04:23:48 AM
So capacity isn’t the problem, it’s about reaching people. And these service providers have no excuse for congestion unless the nodes are the bottlenecks not the links.

Thanks to Licquorice for that. And hello my friend, nice to hear from you, feels like I haven’t heard from you in a while. Hope you are doing well.
Title: Re: Long distance fibre
Post by: niemand on January 11, 2020, 11:34:27 AM
The access fibre is something of a PITA. Often you're using a single fibre to serve a single premises. This was part of the issue when Virgin Media's high fibre count access network cables were pile driven outside their New Malden site.

The trunks you can push more down the same wavelengths by using analogue transmission, and can get better efficiency out of the fibre by using different Wavelength Selectable Switches. Remember these wavelengths have to go to something to terminate them, prisms can bounce them around but they have to hit something to decode them. WSS select wavelengths and deliver them to receivers. Can only demodulate so much via a single transceiver so have to split the 'big' signals for optical to electrical conversion and combine them for electrical to optical. :)

Older school 'on or off' optics are old news. Optics now run with 2 sine waves in quadrature, one in phase one 90 degrees out of phase (https://en.wikipedia.org/wiki/Phase-shift_keying#Quadrature_phase-shift_keying_(QPSK)), and by measuring power variance between them Quadrature Amplitude Modulation (https://en.wikipedia.org/wiki/Quadrature_amplitude_modulation) can be used to further increase density. They emulate the electrical version of combining two electrical signals using optical inferometers (https://www.britannica.com/technology/optical-interferometer).

We can now get 400 Gb/s out of 75 GHz of optical spectrum via 64QAM. 256QAM, increasing throughput in the same optical spectrum by another third, is being tested.
Title: Re: Long distance fibre
Post by: Weaver on January 12, 2020, 12:07:13 PM
I don’t know anything at all about optical comms, I would like to read up on it. I’m familiar with all the standard electrical modulation schemes and the maths behind them. I didn’t know that optics had now reached the sophistication of going beyond 1-bit/off-on as you said and going to quadrature and QAMn. Of course electrical does more still for the moment with QAM1024 and even QAM4096 I saw mentioned in connection with microwave comes (how on Earth? Never mind.).

400Gb/s out of 75GHz, so what 5.3 bits/Hz ? Is that good these days by RF standards?
Title: Re: Long distance fibre
Post by: niemand on January 12, 2020, 02:09:21 PM
It's equivalent of 64 QAM. Nothing special, your ADSL will be managing more that that on some tones.

More impressive is it being emulated by optical inferometers.

EDIT: A reminder that you can find 15 bits per tone on DSL - 32768 QAM.
Title: Re: Long distance fibre
Post by: Weaver on January 13, 2020, 08:07:45 PM
Good point, I had forgotten to count my DSL configurations.