3. Opening thoughts
4G – Too little too soon?
Six concerns about the evolution of wireless:
1. Pursuit of high performance is not keeping up with demand
2. Growing complexity
3. Evidence of diminishing network quality
4. Market and supply chain fragmentation
5. Growing costs
6. Immature market – confusion over role of value, price and cost
Inspired by:
Technology in the Recession: Less is Moore
http://www.economist.com/opinion/displaystory.cfm?story_id=12932356
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 2 28th October 2009
4. Predicting the next winning technology
Often the “best” doesn’t win
• Ethernet vs. Token ring
• 802.11b vs. HiperLAN
• Windows 3.1 vs. Unix
• Iridium vs. GSM
• Esperanto vs. English (or maybe Chinese...)
“Perfection is the enemy
of the good”
Gustave Flaubert
French Novelist 1821 - 1880
• Now it’s LTE vs. GSM, HSPA - and Wi-Fi
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 3 28th October 2009
5. What we would like in wireless…
Simplicity and economies of scale:
- One worldwide wireless standard for everything
- One air interface
- One frequency band
- One core network
- One application development framework
- And no IPR (that we don’t own…)
For 2G, GSM substantially achieved this in 1992
For 3G, UMTS tried to achieve this in 1999 but faltered
For 4G, what is the outlook - will LTE deliver?
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 4 28th October 2009
6. Attractive attributes of LTE
Based on OFDMA giving:
• Scalable bandwidths for deployment flexibility and high data
rates
• Suitable for use with MIMO
• Supports narrowband scheduling to take advantage of fading
• Potential for higher efficiency than CDMA at wide bandwidths
• Supports enhanced broadcast features (MBSFN)
• Support for in-channel relaying (backhaul)
Simpler low latency packet-only core network
Support for QoS (e.g. for VoIP)
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 5 28th October 2009
7. Planned Order
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 6 28th October 2009
8. …and the space (spectrum) to deploy it
Note use
of spacial
diversity!
Denver
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 7 28th October 2009
9. But what we have…
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 8 28th October 2009
10. …is the legacy
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 9 28th October 2009
11. Yuck!
1G
3G
5G 2G
4G
Heathrow
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 10 28th October 2009
12. Wireless evolution 1990 - 2010
Increasing efficiency, bandwidth and data rates
IS-95A IS-136 802.11b
2G cdma GSM TDMA PDC
802.11a
IS-95B HSCSD GPRS iMode
2.5G cdma 802.11g
802.11h
IS-95C E-GPRS W-CDMA W-CDMA TD-SCDMA
3G cdma2000 EDGE FDD TDD LCR-TDD
802.11n
1xEV-DO 1xEV-DO 1xEV-DO HSDPA HSUPA 802.16d
3.5G Fixed
Release 0 Release A Release B FDD & TDD FDD & TDD WiMAXTM
802.16e WiBRO
UMB LTE Edge HSPA+ Mobile
3.9G Rel-8 Evolution WiMAXTM
LTE- In 20 years of evolution
4G Advanced why has so little 802.16m
Rel-10
come off the plate?
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 11 28th October 2009
13. Phases of Technology Adoption
1. Standardization 2. Regulation 3. Physics 4. Commerce
The Committee The Test Maxwell’s The Shopping
Room House Place Mall
And then there was light…
Where engineers Where products are Where electro- Where commercial
consume lots of put on trial to prove magnetic law law determines
coffee while creating their conformance determines if it whether anyone
wireless law actually works actually buys it
Just because the industry invents a new standard does not mean
success is guaranteed. What determines commercial success?
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 12 28th October 2009
14. Technology evolution - Audio
Over the last century, audio has evolved from wax through shellac, vinyl,
metallic tape, to opto-mechanical discs and finally solid state silicon
At each new generation the user perceived benefits have been undeniable
meaning earlier generations have largely been obsoleted.
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 13 28th October 2009
15. Technology evolution – Cellular
Commercial Cellular phone technology started in the 70s with expensive
bulky and heavy products moving through the first true “handsets” and
onto GSM and CDMA
So is the added value of new generations slowing down?
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 14 28th October 2009
16. Performance - Which is the best car?
$2,500 $1,500,000
Answer: Both! It depends on the problem you are trying to solve
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 15 28th October 2009
17. Top end wireless is a bit like the top end automotive
• It is undeniably real
• But is it affordable?
And once you have purchased it, where and how often can
you really experience it? LTE: Long Term Employment or Less Than Expected?
Moray Rumney
28th October 2009
18. 1001 bhp
At top speed, empties its fuel
Top Speed – 253 mph
tank in 12 minutes!
0 – 60 mph in 2.5 seconds
0 – 125 mph in 7.5 sec
0 – 250 mph in 16.7 sec
250 – 0 mph in 9.8 sec
Price: $ 1,500,000 (plus tax!)
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
28th October 2009
19. If your supercar commute
to work looks like this
then you will have paid
for peak performance but
you will experience
average or poor
performance
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
28th October 2009
20. Quiz #1 - High speed vs. high capacity
At what average speed does the capacity (i.e.
cars per hour) of this road reach its peak?
This is counter-intuitive: High
10 mph ? speed drives down capacity!
40 mph ? In wireless, high data rates
consume disproportionately
70 mph ? more resources and provide less
coverage for fewer users –
100 mph ? overall capacity reduces.
1000 mph? Improving capacity through
higher average speeds is plain
hard work
In motoring terms, improving the
average would require a whole
new form of robotic control.
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 19 28th October 2009
21. So what really matters - Peak or Average?
WoW!
Now with predictive
results!
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 20 28th October 2009
22. What is enabling the apparent exponential growth
in wireless communications?
The capacity of a system to deliver services is defined by three
main factors:
• The bandwidth of the available radio spectrum – in MHz
• The efficient use of that spectrum – bits / second / hertz
• The number of cells – this equates to spectrum reuse
Efficiency
Number of cells
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 21 28th October 2009
23. Growth to date dominated by increasing cell count
If we apply Cooper’s law over the last 50 years we are looking at a growth in
wireless capacity of perhaps 1,000,000
Allocating this growth between the axes of capacity looks roughly like this:
10000 2000
1000 Growth has
historically been
Growth factor
100 25 dominated by the
20
increase in the
10 number of cells
1
Efficiency Spectrum No. of cells
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 22 28th October 2009
24. Comparing wireless growth potential for the next
decade
100
100
Growth potential
10
3
2
1
Efficiency Spectrum No. of cells
Using current projections, the increase of cell numbers (spectrum
reuse) remains the dominant means of growing capacity
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 23 28th October 2009
25. Growth in peak / average spectral efficiency by technology
Spectral Efficiency bits / sec / Hz
100
Peak efficiency lies
around this line
LTE
802.16e Average efficiency
10 HSDPA and hence capacity
1xEV-DO growth of deployed
W-CDMA systems lags well
1 IS-95C behind and will level
EDGE LTE target off due to inter-cell
GPRS interference
1xEV-DO(A)
0.1 GSM EGPRS2 1/3
AMPS HSDPA (R7) Peak efficiency drives
HSDPA (R5)
W-CDMA (R99) up air interface cost &
EGPRS 1/3 (R99)
0.01
EGPRS 4/12 (R99) complexity
1980 1985 1990 1995 2000 2005 2010 2015 You pay for the peak
Average efficiency Peak efficiency but experience the
average
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 24 28th October 2009
26. Projecting ahead shows the gap between average
and peak rates in a loaded cell will grow to 90x
Peak rates Average Efficiency Spectrum Capacity
1000000000
Data rates x 100000
100000000 Efficiency x 87
Efficiency, spectrum and capacity are Spectrum x 13
normalized to single-band GSM in 1992
83 users/cell occupying 25 MHz @ 9.6 kbps
10000000 1100x capacity
A 90x gap will exist
1000000 by 2015
The outlook is that
100000
average efficiency
and spectrum will
fall further behind
10000
1985 1990 1995 2000 2005 2010 2015 peak rates
The average efficiency, spectrum and capacity plots are normalized
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 25 28th October 2009
27. And on a linear scale
Peak rates Average Efficiency Spectrum Capacity
1000000000
900000000
800000000
700000000
The capacity Macrocell
crunch reality lies
600000000
somewhere
500000000
below this line
400000000
300000000
200000000
100000000
0
1985 1990 1995 2000 2005 2010 2015
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 26 28th October 2009
28. Setting performance expectations –
150 Mbps is your right!
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 27 28th October 2009
29. Examples of growing complexity:
Voice support
2G & 3G
• Single solution – circuit-switched services
3.9G/4G LTE
1. IMS – preferred 3GPP solution
2. Circuit-Switched Fallback – preferred 3GPP alternative
3. VoLGA – UMA/GAN for LTE
4. Fast Track voice over LTE (NSN – SIP-based)
5. Proprietary solutions (e.g. Skype)
Choice creates unwanted interworking complexity
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 28 28th October 2009
31. Baseline LTE FDD single band architecture
Source: R4-091204 “Study of UE architectures for LTE-A deployment scenarios” Nokia
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 30 28th October 2009
32. 4G: Tri-band aggregation plus
4x20 MHz contiguous operation
The Hedgehog PhoneTM
“An Antenna for every
Occasion”
Supports up to:
• 19 frequency bands
• 8 simultaneous radio
technologies
• 8x8 MIMO
Fan cooled –
Blow dry your
hair while you
chat…
Source: (Excluding the hedgehog)
R4-091204 “Study of UE architectures
for LTE-A deployment scenarios” Nokia
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 31 28th October 2009
33. And for the geeks – A growing gap between
conformance testing and real world operation
Attribute Conformance testing Real world operation
High, medium and zero - Real correlation based on
MIMO Correlation matrix
not linked to real antenna actual antenna pattern
Fading channel Extended PA, VA, TU Channels with dynamic taps
Adaptive Modulation & Off – UE becomes fading On – coding aims for constant
coding channel discriminator symbol to noise at UE receiver
CQI, PMI, RI Separate open loop tests Included as part of throughput
Cell-edge Interference Static wideband Narrowband frequency-
signal Gaussian selective based on loading
Multiple UE, real scheduler with
Scheduling None, Single UE frequency selectivity base on
subband CQI/PMI
Variable based on prevailing
Transmission mode Fixed
conditions
Over The Air (OTA) Closed loop with real loading /
Probably open loop
antenna testing scheduler / interference
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 32 28th October 2009
34. Concern regarding baseline LTE Rel-8
performance
LTE Rel-8 targets were based on 2x to 4x gains over a limited version of Rel-
6 HSPA
• Type 1 receiver – 16QAM Single stream with Rx diversity
A fair comparison would be with Rel-8 HSPA
• Type 3i receiver
• Add dual stream DL 2x2 MIMO
• Add equalizer
• Add interference cancellation
• Add 64QAM
On paper the differences between Rel-8 HSPA and Rel-8 LTE are minimal –
perhaps 20% in favour of LTE
In practise the advantage will be with the more mature HSPA and may stay
that way
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 33 28th October 2009
35. OFDMA – An unproven cellular technology
Many benefits are expected from OFDMA
• Improved spectral efficiency
• Improved cell edge performance
• Better suited to MIMO
To date there are no commercial OFDMA deployments using 1x frequency
reuse
• Clearwire using 3x for suburban, 6x urban @ 2.6 GHz
• Single channel 10 MHz @ 700 MHz is a very different scenario
OFDMA Narrowband frequency-selective scheduling creates non-Gaussian
cell-edge interference
• 3GPP simulations appear to have used the much easier Gaussian
interference model which could inflate expectations
CDMA creates Gaussian interference which can be modelled and cancelled
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 34 28th October 2009
36. MIMO – An unproven cellular technology
Much has been said about MIMO for many years
We are still a long way from proving its effectiveness in typical
cellular environments
In particular the evaluation of MIMO OTA - Over The Air
• This is a radiated measurement that takes into account the all-
important real antenna design
• Expect much angst and gnashing of teeth…
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 35 28th October 2009
37. This is what we
Real life MIMO performance hear about :
2x gains
Variation due to
instantaneous
correlation
Variation in the
frequency
domain not
shown
Most macrocell
activity takes
place in this
region
Variation due to fading and variable interference
LTE: Long Term Employment or Less Than MIMO from
Taking LTE Expected?
Standards to Starbucks
Moray Rumney
Page 36 Moray Rumney th October 2009
28 10th June
38. Managing complexity is an exponential problem
(and Moore is not the solution)
Its not that systems can’t work it is primarily that as complexity rises there
are just so many operational combinations to design and test for
For n elements there
are n2 - n possible
interactions.
E.g. for 50 elements
we have 2450
interactions…
If 4 elements in a Conclusion:
system are capable of Interaction must
interacting either in a be minimized
linear or complex But for a system with 9 since we cannot
fashion there are six elements, there are 36 design or test
bi-directional scenarios bi-directional scenarios to quality into
to design and test for consider complex systems
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 37 28th October 2009
39. Looking ahead to the Cost/benefits of LTE-
Advanced
Bandwidth
Enhanced Uplink Higher order MIMO CoMP Relaying
Aggregation
Peak data
rates
Spectral
efficiency
Cell edge
performance
Coverage
UE cost
Network cost
Complexity (UE) (UE) (Network) (Network)
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 38 28th October 2009
40. The macrocellular dilemma
To deliver true mobile broadband these three attributes are all
required:
1. High data rates with the capacity and density to match
2. Ubiquitous coverage
3. Low or reasonable cost
For macrocellular networks, pick any two!
Conclusion: High efficiency macrocellular can’t do it alone.
Some form of small cells is essential.
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 39 28th October 2009
41. Evidence of growing problems in the delivery of
basic services
No time today for details but:
• Performance of MMS and video calls remain erratic
• SMS becoming less reliable
• Multiple receipts common
• Interworking with CDMA2000 still not figured out after more than a
decade – multiple issues
• Network generated SMS spam
• Welcome to the Netherlands. What? Again!
• Voice connectivity problems between 2G/3G
• Caller ID frequently fails
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 40 28th October 2009
42. In a mature market, Value > price > cost
Price Data rate Price € Cost / Price / Relative
Service / volume MByte MByte price
SMS 160 Bytes €0.15 / €X €1000 400,000
message
Voice 10 kbps €0.05 to €0.5 / € Y € 0.7 - € 7 300 - 3000
minute
Data service 3 GBytes €20 / month € Y/5 € 0.007 3
(capped at 3
GBytes)
Unicast Mobile 50 Hours €5 / month € Y/5 € 0.0023 1
TV (Capped at @ 100
50 hrs) kbps
Cheap capacity generates unsustainable low value traffic
New air interfaces provide a linear solution to an exponential problem
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 41 28th October 2009
43. FCC decision on Net Neutrality
“Broadband providers cannot discriminate against
services or applications by slowing them down”
For wireless, how does it look like from the other end?
Is that the DoD?
I’d like dad’s
newspaper delivered
by helicopter please!
No charge of course.
Stubbed toe of the decade perhaps?
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 42 28th October 2009
44. Key lessons learned from 20 years of cellular
What we learned from GSM (2G)
Scale matters -> 4B devices worldwide
Ubiquitous low-rate services work and are hugely profitable (voice/SMS)
What we learned from UMTS (3G)
A fat circuit-switched data pipe doesn’t cut it
Coverage matters
What we learned from HSPA (3.5G)
Packet-switched data is essential for mobile broadband
When the megabits finally start flowing, data density matters
Pricing data at 1% of voice is not sustainable
What we learned from the iPhone
Usability and apps are vastly more critical to innovation than air interfaces
Wi-Fi is an effective way to offload low-grade traffic from cellular
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 43 28th October 2009
45. What will it take for LTE to displace existing
cellular to deliver on the unified vision?
Sufficient clean spectrum in which to gain a foothold (e.g.
European 800 MHz band)
Network optimization to milk opportunities and overcome
challenges with OFDM
Demonstrated performance to drive obsolescence of legacy and
evolving older technologies
Solution to Voice support caused by the packet-only network
Proof that IMS is a viable solution for mobile systems
Economies of scale to drive down prices to below legacy
systems
Leave high volume / low value data to Wi-Fi and femtocells
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 44 28th October 2009
46. In pursuit of excellence
Could we improve on these?
Surely!
But why don’t we?
The return is not worth the pain
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 45 28th October 2009
47. The future of wireless is bright…
but the future will have to be low cost/complexity
Would you like some nybble and bytes with that Sir?
iFi?
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 46 28th October 2009
48. Thank you for listening!
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 47 28th October 2009
49. Backup
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 48 28th October 2009
50. Geometry factor distribution in urban cells
90% of users 10% of users
This plot shows the
100 % variation in geometry
factor across a typical
outdoor urban cell
Very high spectral
Cumulative distribution
efficiency is only seen
Most new high data when the geometry
rate/MIMO factor is above 15 dB,
performance which is an environment
targets require that 90% of the user
geometry factors population will not
experienced by experience
<10% of the user
population In-building penetration
loss will degrade
performance further
This puts a finite and
0% very low limit on indoor
-30 -20 -10 0 10 20 30 performance when using
outdoor transmission
Geometry factor in dB systems
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 49 28th October 2009
51. In search of 21 Mbps
Source:
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 50 28th October 2009
52. Example of interference-limited throughput*
M M M
HSDPA macrocell, single Rx + equalizer M
M M
15 code 64QAM, 20 Mbps peak M M M M
M M MM M
34 randomly distributed users
M M M
M M M
Quiz #2: What is the combined M M
throughput and why? M M MM M
M M
M M M
680 Mbps
M
20 Mbps
13 Mbps That is a median
throughput of 40 kbps
1.3 Mbps (0.26 b/s/Hz)
* Source: 3GPP RAN WG4 R4-081344
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 51 28th October 2009
53. Impact of femtocell deployment on throughput
M M M
Using the same macrocell add 96 femtocells M
M M
24 users migrate to femtocells M M M M
M M MM M
10 users remain on the macrocell
M M M
M M M
Quiz #3: What is the combined M M
throughput and why? M M MM M
M M
M M M
270 Mbps
M
27 Mbps
That is an median
throughput of 8 Mbps
2.7 Mbps
A 200x improvement!
1.3 Mbps The remaining macrocell users
go from 50 kbps to 170 kbps
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 52 28th October 2009
54. CCDF of throughput with and without femtocells
Co-Channel, Self Calibrated HNB Tx Power Co-Channel, Self Calibrated HNB Tx Power
1 1
0.9 0.9 24 HUEs + 10 MUEs/cell, 1 Rx
34 MUEs/cell, No HNBs, 1 Rx
0.8 0.8
24 HUEs + 10 MUEs/cell, 1 Rx
0.7 0.7
34 MUEs/cell, No HNBs, 1 Rx
0.6 0.6
CDF
CDF
0.5 0.5
0.4 40 kbps @ 0.4
0.3
50 percentile 0.3
0.2 0.2 8 Mbps @
50 percentile
0.1 0.1
0 0
0 1 2 3 4 5 0 2 4 6 8 10 12 14 16 18
All UEs Average Throughput (bps) 5 All UEs Average Throughput (bps) 6
x 10 x 10
Detail showing 40 kbps median for macrocell Full CCDF showing 8 Mbps median for
macrocell plus 24 active femtocells
Projected spectrum and efficiency gains could push the blue trace to
the right by 6x, this femtocell study moved it by 200x
LTE: Long Term Employment or Less Than Expected?
Moray Rumney
Page 53 28th October 2009