UK Spectrum Policy Forum Cluster 2 Meeting – 25 September 2014 Stephen Temple, 5G Innovation Centre (5GIC) Where is the spectrum for a “small-cell” 5G mobile revolution? More information at: http://www.techuk.org/about/uk-spectrum-policy-forum All rights reserved

1. Critical Issues for 5G
Where is the spectrum for a “small-cell” 5G mobile revolution?
Prof Stephen Temple CBE
Visiting Professor, 5G IC

2. Summary of where we are with the 5G initiative
There is a growing international consensus that 5G will be introduced around 2020
To meet this time-scale standardisation will need to be underway in earnest by early 2016
We are roughly half way through a two year debate on concepts…a time to get ideas on
the table for open discussion
Somewhere in the 5G mix will be at least one new Radio Access Technology (RAT) and it
seems highly likely that one such new RAT will be applied to small cells
It will only be possible to bring such a new 5G RAT to market if suitable spectrum is
available on a 2020 time-frame…it is a critical success factor!
This can only come about through a timely dialogue between the research community,
standardisation body and spectrum managers
…the Spectrum Policy Forum has a role to play in the 5G spectrum dialogue -

3. Anatomy of global mobile technology revolutions – The Essential Dialogue
RESEARCH
Early Spectrum Issues:
affecting spectrum choice/use?
regulatory issues to resolve?
SPECTRUM
PLANNING
STANDARDS
MAKING
5G
1. What is 5G in network terms?
2. Where is the spectrum for it?
3. What will be different about 5G
4. What are the most likely
Market
A proactive EU & global
dialogue between
research, spectrum
planning and standards
making is essential
To get this dialogue going the first “Working Assumptions” about 5G need to be
emerging now…so 4 working assumptions are tabled that address the above 4 questions

4. 5G
WIRELESS
FIBRE
5 –10 Gb/s
20-60 GHz
SUPER-FAST“MOBILE”
Working Assumption 1: – The
5G Network-Spectrum Vision
High-density urban small-cell clusters
500 –1000 Mb/s
2 - 4 GHz
“DAN enabled” so “Always sufficient” Quality of Service matches demand
Fibre Highway Lattice
SATELLITE
MASSIVE DATA ENGINES powering the Digital Economy & Social Space
5G IC
SUPER-EFFICIENT “UNIVERSAL” MOBILE DATA ACCESS
5 –10 Mb/s 700 MHz – 1900 MHz
Surrey University
CELLULAR
Ultra LP
I o T

5. What is the problem we are trying to solve: finding spectrum for 5G small-cell
clusters to provide economic (<4GHz) super-fast (>2 GHz) urban wide-area coverage?
Working Assumptions:
Working Assumption 2 – Clean spectrum between 2 – 4 GHz will not be reserved for 5G
small cells & they will have to share spectrum with what may exist by 2020. This looks feasible.
Working Assumption 3 – 5G RF channel widths between 100 and 200 MHz will be needed
(Compare: GSM 200 kHz, 3G 5 MHz, 4G LTE 20 MHz so 100-200 MHz looks about right for 5G)
Working Assumption 4 – the big 5G regulatory issue will be getting large RF channel widths in
bands sliced-up into exclusive allocations for competition reasons

6. Feasibility of 5G small cells sharing spectrum in bands between 2 – 4 GHZ
 Ofcom data suggests it likely that over ~ 90% of the UK land mass some or all of
any newly released spectrum above 2GHz will lay unused for a long time.
Spectrum Range Area of UK where some of the
spectrum is unused (1)
Area of UK where all of the
spectrum is unused (1)
GSM spectrum at 900MHz or
1.8 GHz
34% 6%
3G spectrum at 2.1 GHz 87% 24%
5G at >2 & <4 GHz >>87% >>24%
 Sharing of spectrum in the 2 – 4 GHz range looks feasible with
“statistical” large areas of stable use of otherwise idle spectrum
Note 1: Ofcom UK communications infrastructure report 2011 figure 12. Numbers have been rounded. 3G figures today are likely to be lower with improved 3G
coverage but the point remains that the 2.1 GHz numbers above were for 10 years after the spectrum release in 2000.

7. Exclusive spectrum allocations will remain essential in the 5G era
In our purely illustrative example:
1. Whilst each MNO has exclusive
spectrum, the RF channel width
each has is significantly less than
the band’s full potential
Illustrative
Exclusive
A B
100
Mb/s
225
Mb/s
200
Mb/s
300
Mb/s
225
Mb/s
D C
Spectrum
Allocations
2. There is also massive
geographic spectrum
inefficiency above 2GHz
eg 10 years after
release of 2.1 GHz 3G
spectrum::
• All of the latent
spectrum capacity
rested idle over ~ 25%
of the UK
• Some of the latent
spectrum capacity
rested idle over ~ 87%
of the UK
Hierarchy key:
Exclusive
spectrum cell
Indoor cell
Expanded
spectrum cell
Powered-down
cell
Surrey University
5G IC

8. Illustrative
Exclusive
100
Mb/s
225
Mb/s
200
Mb/s
300
Mb/s
225
Mb/s
Spectrum
Allocations
Dynamic
Spectrum
expansion
100% of allocated
bandwidth at unique
locations. 1
Gb/s
A
A
B
D C
“Simple” dynamic spectrum expansion
1. Operator A has a
small cell at a unique
location
2. 5G technology
switches RF bandwidth
to maximum at this
location
3. User data rate
jumps from 225 Mb/s to
1 Gb/s
4. Operator A has full freedom to re-adjust up-path
and down- path capacities at will (if TDD)
Model 1

9. Illustrative
A
100
Mb/s
400
Mb/s
225
Mb/s
200
Mb/s
300
Mb/s
225
Mb/s
600
Mb/s
Exclusive
Spectrum
Allocations
B
B
C
C
D
Dynamic
Spectrum
expansion
Dynamic spectrum expansion that copes with multiple operators
1. Operator B & C
both locate small cells
at the same specific
location
2. The spare spectrum
could be shared in
proportion to size of
exclusive spectrum
allocations
Model 2

10. Dynamic spectrum expansion for reliable rural mobile base station back-haul
Illustrative
A B
100
Mb/s
225
Mb/s
200
Mb/s
300
Mb/s
225
Mb/s
1
Gb/s
Exclusive
D C
Spectrum
Allocations
C
Dynamic
Spectrum
expansion
Model 3
1. Operator C has a
rural 4G site capacity
limited by absence of
reliable longer range
wide bandwidth back-haul
2. The 5G mobile
technology is re-purposed
for back-haul
3. It uses the entire
bandwidth to match
capacity of mmWave
systems but much
longer range and
more reliable in all
weathers

11. Dynamic spectrum expansion with building penetration loss achieving in-building re-use
1. Operator C allows
its customer at a
unique location to use
some of its spectrum
for purely indoor use
2. The 5G technology
in the user terminal is
under control of the
MNO to manage the
bandwidth &
interference
200
Mb/s
800
Mb/s
Spectrum
expansion
3. An unlicensed
channel could be part
of the mix (100 Mb/s
in this example ie 10%
of the band) which is
essentially a pooled
use but usable by
operators outdoors
Model 4
Illustrative
Exclusive
A B
100
Mb/s
225
Mb/s
200
Mb/s
300
Mb/s
225
Mb/s
D C
Spectrum
Allocations
D

12. Regulatory implications -
These depend upon the spectrum in question:
For existing mobile spectrum :
It is a matter entirely for (two or more) mobile network operators to decide for
themselves whether “pooling” their localised scraps of unused spectrum offers “a
bigger spectrum cake” with reciprocal benefits. It is a form of spectrum trading.
The regulatory task is simply - to enable it.
For spectrum yet to be released: There is a public interest:
a) In finding ways to reduce the massive geographic spectrum inefficiency for
spectrum above 2 GHz that is unlikely ever to be used for “national” coverage
b) In creating an industry consensus for a spectrum approach to 5G that allows
the UK to be an early beneficiary of a potentially huge advance in UK mobile
infrastructure performance.

13. Conclusions:
Dynamic Spectrum Expansion preserves all the up-sides of exclusive spectrum allocations
whilst capturing substantial benefits from a form of dynamic spectrum access to otherwise
wasted spectrum at many locations
It offers a possible solution to the 5G spectrum conundrum of needing much wider radio
channels in a regulatory environment where spectrum in sliced into narrower exclusive
allocations for competition reasons.
Many versions are possible ranging from simple bilateral “punch-through” agreements
with simple data base access to usage to complex automated Dynamic Spectrum
Access requiring specially developed 5G technology to make implementation practical
… and Questions:
1. Is there a consensus on the problem to be solved - that in the 5G era we want the
benefits of exclusive spectrum allocations, we want allocations sliced for network
competition reasons but we also want a substantial leap in RF channel bandwidth
2. Is Dynamic Spectrum Expansion of interest “in principle” as a potential solution and worth
further study in a 5G context and time-scale?
3. What are the other options for solving the 5G spectrum requirements that are timely,
economic and technically sound?