[LINK] The importance of high-speed broadband in regional Australia

[Gerard]

Hello David

 >>It seems unlikely that 5G will ever be much chop outside 
densely-populated areas. Even at te lower frequencies advocated by the 
likes of Telstra, it just doesn' perform over distances and around 
obstructions. Does any Linker know the true potential?<<

Here is what I understand so far.

There is still some uncertainty about what 5G exactly is. I have not 
heard of anything in 5G that is intended to provide broadband to the bush.

The main focus of 5G (in both research and commercialisation) is higher 
data rates and better coverage for mobile users (which are mainly 
indoors), lower latency for self driving cars, VR etc and long range for 
the Internet of Things.

Much of the really broadband comms would be achieved by a move from UHF 
to the mm Wave (30-300 GHz) => much
shorter range.

With this in mind, technologies proposed for 5G include...

* Femto cells are designed to densify the cellular network. These are 
mainly indoor mobile base stations - you can
think of these as WiFi access points for mobile phone. They need 
back-haul and they serve just a small number of users at a time so that 
user experience can be maintained in highly populated areas.

* Cloud RAN = A small outdoor cellular network with cloud based 
processing and fibre front-haul to a radio head located
on a pole in the street.

As for long range, there is IoT ....

* LPWAN  Low power wide area network => long range, very low bandwidth 
and very many devices. Each device
(for example a level indicator on a dam in a remote area) transmits 
occasional short packets so that many devices can be accommodated.

One example of an IoT technology is LoRaWAN which uses spread spectrum 
techniques for better reliability.
The following site describes LoRaWAN and Figure 2 shows the trade-off 
between data rate spreading factor (excess bandwidth consumption) and range.
https://www.digikey.com/en/articles/techzone/2016/nov/lorawan-part-1-15-km-wireless-10-year-battery-life-iot

But LPWAN is not broadband to the bush.

For long range broadband, one option touted is to continue with LTE 
which is already widely deployed. However I dont believe that LTE or any 
cellular mobile technology is designed for regional broadband 
communications. There is however a modification of LTE that is proposed 
for LPWAN.

So are there any possible approaches for regional broadband? - In short, 
yes.

massive MIMO is another technology proposed for 5G.

Massive MIMO = MIMO (Multiple Input Multiple Output) with hundreds 
(thousands?) of antennas on
the base station =>  Much higher spectral efficiency and larger range.

In massive MIMO, the number of antennas on the base station antenna 
array is greater (usually several times) the number of remote clients, 
each having one antenna. Because of the beam-forming process (MIMO), 
massive MIMO can completely reuse the spectrum band so that the remote 
clients gain access to the full band without needing to share it. It is 
as if each client is the only one on the network. For 100 antenna 
elements you could serve about 10 users simultaneously. This is a 
typical number of antennas and the technology at this level has been 
demonstrated - https://arxiv.org/pdf/1701.08818.pdf

Massive MIMO can be viewed as a large-scale version of the CSIRO Ngara 
multi-user MIMO system.
http://www.sief.org.au/Documents/RP/NgaraFinalGeneralReport.pdf

When operated at mm wavelength, Massive MIMO is considered a competitor 
for Cloud RANs.

However the beam-forming process could allow longer range to be achieved 
if longer wavelengths were used.

Massive MIMO has been considered for rural broadband by Facebook who 
have built a system
called 'Aries'
https://code.facebook.com/posts/1072680049445290/introducing-facebook-s-new-terrestrial-connectivity-systems-terragraph-and-project-aries/

...From our recent populationdistribution study 
<https://code.facebook.com/posts/1676452492623525/connecting-the-world-with-better-maps/?__mref=message_bubble>across 
20 countries, we know that nearly 97 percent of the global population 
lives within 40 kilometers of a major city. As such, we are interested 
in developing this technology to harness the incredible gains in 
providing communications to rural communities from city centers. 
Additionally, providing backhaul to rural environments can be 
prohibitively expensive, but the hope with systems such as these is that 
costly rural infrastructure can be avoided while still providing 
high-speed connectivity....


Aries is still a (centralised) cellular network that you would have to 
place for example on Black Mountain in the ACT
to serve the surrounding area.  This means that you would need sub GHz 
frequencies (long wavelengths)  to get over the hills. The antenna 
elements would be in the order of 10s of cms.   Using the digital 
dividend spectrum, a system with 1000 antennas (up to about 100 remote 
clients) on 700 MHz like Aries would require elements on its ring of 
about 20 cm separation and the diameter would be just over 60 m. This is 
a minimum and would be larger if a larger antenna element separation 
were used to minimise inter-element mutual coupling (an RF electrical 
impairment that complicates antenna array design: especially when there 
are large numbers of  elements).

A key property of massive MIMO is that it is VERY POWER EFFICIENT. The 
power required for a given deployment per antenna while serving many 
clients  is much less than would be required by a conventional single 
antenna base station while serving one client. This is how large range 
can be achieved. However this advantage does not necessarily mean that 
we can achieve a MUCH larger range. Centralising the massive MIMO base 
station as in Aries would impose a legal limit on the TOTAL power that 
the base station can emit. Thus the per-antenna power would have to be 
decreased in proportion to the number of antennas in the base station 
array. This critically diminishes the possible range benefits to be 
derived.

The problem with Aries therefore is that it cannot scale. An Aries 
device might work in limited population deployments but would not serve 
Canberra's regional areas (radius 40kms) even if it could be built. If 
you dotted the country side with them they would interfere with each 
other unless they operated on different frequency bands.

For massive MIMO  to scale, we need a distributed network where the 
antenna elements in the array are themselves located in different 
geographical locations. The base station would now have a distributed 
antenna array. Each antenna would be driven by a small radio device that 
uses the existing back-haul to coordinate with thousands of identical 
radios on the base station network. The coordination is vital for MIMO 
beam-forming. Such a distributed massive MIMO system  could be mounted 
on the rooftops of buildings in a city or a regional town centre where 
it could run on much lower frequencies (say 50MHz, 6 metre wavelength=> 
much larger range). Such a distributed  network could scale antenna 
number to meet demand. The legal power limit would now apply to EACH 
antenna in the array, making much more power available to the entire 
array. Right now there is about 21 MHz of spectrum on the band I TV 
channels (6 metre band) that could be deployed for this purpose - the 
most under-used sub-GHz spectrum in Australia. There are a few issues to 
solve before this approach becomes practical.  It is currently a 
research topic  (http://users.cecs.anu.edu.au/~Gerard.Borg/)

Paul Budde states that 5G will drive fibre deeper and deeper into the 
network and benefit the business model for FTTH.
http://paulbudde.com/blog/mobile-communications/mobile-infrastructure-will-ultimately-rely-fibre-broadband/
http://paulbudde.com/blog/mobile-communications/next-development-wireless-broadband/

It should be emphasised that more extensive fibre deployment (FTTH!) 
SHOULD/WILL also motivate further wireless research and lead to better 
use and (hopefully) better regulation of the radio spectrum.