Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Bowdon]

BT Test Uses G.fast Broadband to Supply New C-RAN Mobile Network

BT’s forthcoming 300-500Mbps G.fast broadband technology won’t just be used to upgrade homes. The operator has also claimed a “world first” by using the G.fast to supply a cellular Cloud Radio Access Network (C-RAN) over copper lines, which could be used to support 4G and future 5G mobile services.

BT: G.Fast Copper Can Deliver 5G Networks

BT says it has conducted the “world’s first” trial of Cloud Radio Access Network (C-RAN) technology over copper in a development it claims could make the deployment of 4G and 5G networks cheaper for mobile operators by eliminating the need to lay additional fibre to base stations.

BT Trials Cloud Ran Over G.fast In Step Towards 5G

C-RAN is a new network architecture used to connect cellular base-stations to mobile operators' core networks. A traditional approach to C-RAN requires a dedicated fibre link to connect transmitters at the top of a cell tower to complex signal-processing equipment deeper in the network. This can involve complex and costly engineering work if no fibre is present in the ground to carry the signal.

BT brings G.fast to cloud RAN

Researchers at the operator's Adastral Park R&D lab worked with semiconductor company Cavium on the trial, which delivered speeds of up to 200MBps.

BT uses G.fast to deliver C-RAN cellular service over copper

BT has successfully used G.fast technology to deliver a 'Cloud Radio Access Network' (C-RAN) cellular network service over copper. C-RAN is a new network architecture used to connect cellular base-stations to mobile operators' core networks. Working with US-based semiconductor manufacturer Cavium, researchers at BT's Adastral Park Labs in Ipswich demonstrated that they can use G.fast technology to deliver cellular data over copper lines at speeds of 150-200 Mbps.

BT Trials C-RAN Over G.fast

LONDON -- BT announced today that it has successfully used G.fast technology to deliver a ‘Cloud Radio Access Network’ (C-RAN) cellular network service over copper, in an experiment believed to be a world first.

BT tests G.fast technology to underpin future 5G mobile services

BT has announced that trials of its G.fast technology for mobile networks have shown that the technology can allow copper lines to handle the speeds necessary for mobile 'fronthaul' networks.

[kitz]

Thanks Bowden.

There seems quite a lot of reading to do there, however I think the basics for anyone who doesnt have time to read it all is that research at BT's Adastral Park has shown that G.fast technology can be used to provide dedicated backhauls for mobile networks.

BT said that it managed to send cellular data connections at speeds of 150Mbps to 200Mbps over G.fast connections in a C-RAN environment, more than adequate for mobile data requirements.

The research took place at BT’s Adastral Park labs in Ipswich, and Dr Tim Whitley, managing director for research and innovation at BT, described the results as a major breakthrough for the future of mobile data networks.

“These technologies will play a key role in 4G networks and will be fundamental to 5G architectures. The trials are another step towards a fixed and mobile network which will support customers’ increasing demands for data,” he said.

I think Mark from ISPr has summed it up rather nicely when he says

The news is particularly poignant given that BT has just started its merger with UK mobile operator EE and will no doubt be seeking to harness the power of both their fixed and mobile networks, which could results in various efficiency and network related cost savings.

[Bowdon]

I will say that after the recent appointments to top level positions BT seems to be more alert and on the ball. They appear to have multiple projects going and keeping themselves in the headlines. I'm sure this isn't coincidence

It is good that BT are becoming more pro-active and pushing technology. I'm sure there isn't that long to go until some of this new stuff is released to the public.

I was wondering before, will we need a new modem, at least, to be able to access G.fast?

[burakkucat]

I was wondering before, will we need a new modem, at least, to be able to access G.fast?

Yes. I believe WWWombat mentioned that in another thread, recently.

My understanding is that rather than use the frequency division multiplexing (as used by G.Dmt, ADSL2, ADSL2+ & VDSL2, to name the modes with which we are most familiar), G.Fast uses the entire allocated bandwidth in time division multiplexing.

Here's a link to the Wikipedia page.

[gt94sss2]

Simply put, the more uses they find for G.Fast - the easier it will be to make a business case to roll it out.

As such, I'd be interested to know what “further functionality” they expect to deliver in their Summer 2016 pilot (as they have already indicated)

I recall from my dealings with BT when ADSL was being trailed, one of the internal battles the broadband side had to fight was from the voice side - who were worried it would decimate the market for voice minutes/second lines etc.

At least now, BT appears to recognise 'fibre'/broadband as the way forward for future growth and know that politicians/the public are much more interested in this than 15-20 years ago.

 

[WWWombat]

I haven't followed all those links to check if they make the point I'm about to, but I thought I should add this ...

In olden days (the nineties), the traffic from the core network out to a base-station was mostly user traffic - voice and data - with a little control signalling. Carried over leased lines or dedicated microwave links. The basestation did all the digital processing work, and the analog RF work, and sent a strong RF signal up to the antenna at the top of the tower.

In our networking world, we would perceive this as the voice and data traffic being sent out to the remote site, across a "fibre" access connection: FTTP, FTTC, G.fast, etc.

C-RAN doesn't work like this. It creates a weird, almost illogical, split.

Basestation architecture has, relatively recently, split the digital "baseband" processing from the analog RF processing; at first this split resulted in separate units at the top and bottom of the masts. The backhaul wasn't affected, and still largely carried voice and data traffic as we understand it.

C-RAN extends this split further. It will leave the RF processing at the tower, but place the digital "basedband" processing back in a central location, where it can share computational power with many other towers. The data being sent between the two is no longer plain voice and data traffic, but will have been encoded, ciphered, separated for frequency hopping or spread for WCDMA, and separated for MIMO. I've seen this described as "fronthaul", rather than backhaul.

(*) I have an analogy with s VDSL2 modem; see below if you are interested.

The bandwidth that needs to go out from the core baseband units (BBU) and the tower-based "remote radio heads" (RRH) gets multiplied heavily; it is no longer just voice and data traffic, but is something that (up to 15 years ago) would have been found going through the backplane of a 2m rack-sized basestation. It is the kind of bandwidth where you'd expect fibre.

More importantly than mere bandwidth is latency. The RF units will have to be synchronised perfectly, with little room for delay. I've seen numbers specified in microseconds, not milliseconds.

https://en.wikipedia.org/wiki/C-RAN
(*) Analogy with a VDSL2 modem

This is akin to splitting the chipset in your VDSL2 modem into two new chipsets.

Have one half do all the work at determining how to split your data across the 4096 tones; what FEC, interleaving and retransmission is needed; what QAM constellation the data should be transmitted with; what vectoring adjustments are needed. Turn that (probably) into 4096 separate streams of data describing the QAM constellations.

Have the other half take those 4096 streams, and turn it into RF. Amplify as necessary. Put that data out on the line.

[Some of the 4096 tones will, of course, be working in the opposite direction. Adapt your thinking to cope as necessary]

Then separate those two chips by 10km, and use fibre between them.

C-RAN is as weird as that! It splits something you just wouldn't think of as splittable.

[WWWombat]

I was wondering before, will we need a new modem, at least, to be able to access G.fast?

Yes. I believe WWWombat mentioned that in another thread, recently.

My understanding is that rather than use the frequency division multiplexing (as used by G.Dmt, ADSL2, ADSL2+ & VDSL2, to name the modes with which we are most familiar), G.Fast uses the entire allocated bandwidth in time division multiplexing.

Yes, you'll need a new modem. TDM offers one reason, but perhaps it is better considered like this:

VDSL2 needed whole magnitudes more processing power than ADSL - which places more demands on the chips at the DSLAM and CPE. These chips are being made to a price, and while they offer *some* flexibility, they just can't cope with differences in complexity of whole magnitudes. Equipment designed for ADSL, and built at the time, couldn't make the jump to handling VDSL2. The first ADSL modems ran hot, and came expensively. Within 3-4 years, they were embedded with routers and firewalls, and ran only warm. But they still couldn't cope with running VDSL on 4x - 8x the frequency range.

Alcatel produce a great graph that shows where different telecom standards fit, in terms of "complexity", at introduction and how that complexity scales downwards over the following few years as Moore's Law takes effect.

Simply put: Right now, chipset manufacturers can make chips for VDSL2 very cheaply ... because the complexity is well within their means of manufacturing. But they cannot make G.Fast chips cheap enough, because G.fast is more complex - 8x frequency, which perhaps amounts to 64x vectoring. As we work out how to make complex chips cheaper, then the hardware becomes feasible.

The graph shows that there looks to be a gap of 2 years between chips that can cope with G.fast at 106MHz and ones that can handle 212MHz.

[WWWombat]

More importantly than mere bandwidth is latency. The RF units will have to be synchronised perfectly, with little room for delay. I've seen numbers specified in microseconds, not milliseconds.

The more I think about this, the more it seems the timing issues come to the fore, not the bandwidth.

Thinking a little further on the implications of this...

It probably means that the data to the remote radio heads can be sent over a G.fast link from a fibre-head end exchange, but probably don't have enough time to transit the national backhaul or core networks.

Would this mean positioning BBU units in the exchange building? Or having a farm of BBU's in an exchange area, with a leased-line direct to the exchange, with connections direct into the Openreach L2S (a dedicated cablelink, perhaps)?