10 Gbps symmetric backhaul services

bealtine

http://www.itnews.com.au/News/292337,csiro-eyes-microwave-backhaul-market.aspx

CSIRO has begun talks with global manufacturers to commercialise microwave technology it says can provide at least 10 Gbps symmetric backhaul services to mobile towers.

The project, funded out of the Science and Industry Endowment Fund and a year in planning, could provide a ten-fold increase in the speed of point-to-point microwave transmission systems within two years, according to project manager, Dr Jay Guo.

Microwave transmission is used to link mobile towers back to a carrier’s network where it is physically difficult or economically unviable to run fibre to the tower.

Where current technology has an upper limit of a gigabit per second to multiple towers over backhaul, the government organisation said it could provide the 10 Gbps symmetric speeds over ranges of up to 50 kilometres.

It uses software-defined radio techniques to combine radio spectrum, along with cognitive radio and “sophisticated baseband coding” to provide the significant boost in speeds.

“For any existing microwave link you can change the electronics - keep the tower, keep the dish - but you can put a lot more bits down,” CSIRO ICT Centre director, Ian Oppermann, told iTnews.

Australian carriers have doubled efforts to fibre up mobile towers wherever possible, with Vodafone Hutchison Australia looking to take advantage of the National Broadband Network rollout to boost the backhaul capacity of its mobile network.

However Dr Guo said demand would remain for microwave links in rural and remote areas, particularly due to the imminent influx of fourth generation mobile technologies.

“The technology would be more expensive [than current products] as some wideband components are needed, but [it is] much cheaper than stacking a number of radios together which seems to become a common industry practice,” he said.

He said the technology would ultimately be able to reach faster speeds, depending on the carrier’s spectrum holdings for microwave links.

Oppermann confirmed commercial talks with manufacturers but declined to name them. He said no talks were being conducted in Australia as there are no large-scale local manufacturers.

Ngara evolved

The microwave backhaul project comes as second phase of CSIRO’s ‘Ngara’ project, which previously aimed to use radio spectrum freed up from the switch to digital television to provide residential fixed wireless broadband connections.

Competing against the Long Term Evolution technology currently being rolled out by many mobile and broadband carriers, it had been tested at speeds of up to 50 Mbps symmetric to 16 receivers in Tasmanian trials.

It could theoretically achieve symmetric 100 Mbps speeds depending on the amount of spectrum used in one instance.

However, CSIRO failed to commercialise the technology in time for key decisions around the fixed wireless component of the National Broadband Network.

Despite commercial discussions with NBN Co last year, the network wholesaler opted ultimately for TD-LTE technology for the wireless component of the NBN, which Oppermann acknowledged was a “much lower risk approach”.

Oppermann said CSIRO still planned to commercialise its wireless technology for end-user access but would look to install the equipment in small purpose-built networks such as mines rather than wide-scale national rollouts.

“Ultimately to get this stuff deployed we’ve got to change both ends of the line - the house terminals and the things sitting in the tower,” he said.

“We’ve had to take a slightly long-term approach on it because it’s a challenge to change both ends of that link.”

While the microwave backhaul industry provided a “much easier space” to sell CSIRO’s proprietary technology, Oppermann said the organisation had so far been unable to prove the Ngara technology for consumers on a sufficient scale.

“We need to go from tens of devices to hundreds of devices to tens of thousands of devices,’ he said.

“Ultimately we need to build credibility up in terms of we can deliver what we say we can deliver on scale.”

watty

I'm sure people have been selling 10Gbps wireless for years. Not cheap and maybe not great distance. Any more than 40km needs pretty high hills/mountains. 20km often needs more than 15m mast (the max height for "easy" PP).

LTE does do 100MBps in one 20MHz channel, or 200Mbps if you bond 40Mhz. But only at small percentage of cell area and only if only one person active. That's physics that can't be broken.

Hard to know exactly what this garbled report is claiming.

oscarBravo

That kind of throughput can only realistically be delivered using hundreds of MHz of radio spectrum. Considering that most backhaul microwave links currently use 28MHz, or sometimes 56MHz, the idea of being allowed to use practically the entire analogue TV spectrum for a backhaul link doesn't sound like the most realistic proposition I've ever heard.

watty

Thousands of MHz...

I thought the article was a bit garbled and talking about two different things?

At 256QAM you need about 1.5GHz of spectrum for 10Gbps. It's the 200GHz bands etc that's used?

With no FEC you might fit 2Gbps into 200MHz at 1024QAM, but it's doubtful.
4096 QAM needs massive power/gain very low noise and only gets to about 2.4Gbps in 200MHz with no overheads or FEC.
10Gbps in 200MHz isn't possible outside a waveguide or fibre and even then very doubtful.

So the 10Gbps must surely be on Microwave and the Article was talking about 100Mbps LTE on UHF (already exists but you only get 3 x 20MHz channels and that needs nearly 150MHz of spectrum and single operator, also of course only one user connected and 256QAM, i.e. "perfect" signal essential).

oscarBravo

watty wrote: »

At 256QAM you need about 1.5GHz of spectrum for 10Gbps. It's the 200GHz bands etc that's used?

And I'm guessing that atmospheric fade characteristics at 200GHz suggest a useful link distance in the hundreds of metres, never mind tens of kilometres.

I think we can agree that there's nothing intelligible or useful in the article from a technical point of view.

bminish

I have the IEEE comsoc mag where this was written up, basically it's done by combining lots of channels across several Microwave Point to point bands. Combined with some rather clever stuff to maintain good spectrum mask performance.

10Gbs is an upper limit based on getting pretty much all of several P2P bands and it will not be cheap to buy or licence.

Summary.
CSIRO NGARA BACKHAUL SYSTEM
Due to the radio propagation characteristics, the
microwave frequency bands at 6, 6.7, and 8 GHz
are suitable for long-range wireless backhaul
links. One way to achieve scalable multi-Gb/s
wireless link is by aggregating a certain number
of RF channels in these frequency bands digitally
to obtain sufficient bandwidth. With 7 RF
channels (each having 29.65 MHz bandwidth) in
the 6 GHz band, 7 RF channels (each having 40
MHz bandwidth) in the 6.7 GHz band, and 8 RF
channels (each having 29.65 MHz bandwidth) in
the 8 GHz band, the total usable bandwidth is
over 724.75 MHz. At spectral efficiency of 7.75
b/s/Hz with 256-quadrature amplitude modulation
(QAM) and RS(247,255) channel coding,
the total achievable data rate is above 5 Gb/s. By
further employing dual polarization transmission,
a 10 Gb/s wireless link can be achieved. If
not all the channels are available, which is most
likely the case, one can reduce the data rate
accordingly.

It's a nice bit of work and certainly has applications but unfortunately it does not rewrite the laws of physics.

This might be viable for reaching very remote bits of Rural Australia but it's a non runner in Ireland, where thanks to our lack of Fibre the viable bits of spectrum are already heavily utilised.

watty

So nothing to see, no cuddly animals hurt or physics abused. Just plain old channel bonding (been available at baseband IP using clever routers on ANY kind of connection for years). So really counting polarisation re-use you need 16 off transeivers on 8 channels to get near the 10Gbps.

Just horrendously expensive in licences and link hardware.

oscarBravo

The ComReg licences (assuming you could get them, which you couldn't) would cost €26,400 per annum.

Ouch.

bminish

watty wrote: »

So nothing to see, no cuddly animals hurt or physics abused. Just plain old channel bonding (been available at baseband IP using clever routers on ANY kind of connection for years).

It's a bit cleverer than channel bonding as there's some very clever stuff being done to do away with the need for guard bands on adjacent channels whilst still maintaining full guard band spec for the channels that have to be avoided because they are in use by others, this maximises the spectral efficiency where you can get a block of channels together

So really counting polarisation re-use you need 16 off transeivers on 8 channels to get near the 10Gbps.
Just horrendously expensive in licences and link hardware.

Not quite, for each band it's one (horrendously complex) IDU /ODU / Dish and it does manage better Bits/Hz than bonding a bunch of singe channel systems.
It's clever, innovative and good work but no laws of physics are harmed here

watty

But not cheap!
It may use one dish but electronics and spectrum make it a very niche solution. Obviously though should be a bit cheaper than the "old fashioned way" (which can use one dish).

It's nice work, but not what the original article was suggesting. Handy for linking two towns where the alligators would eat the fibre?

Still, here when you look at the once off fibre install costs if there was economic access to ESB & Eircom poles and NRA ducts fibre is a lot cheaper than that 20K a year radio licence?