Future Networking v Energy Limits ICTON 2024 Bari Italy

In the global energy equation, the IT industry is not yet a major contributor to global warming, but it is increasingly significant. From an engineering standpoint we can achieve huge energy saving by replacing electronic signal processing with optical techniques for routing and switching, whilst longer fibre spans in the local loop offer further reductions.  The mobile industry on the other hand has engineered 5G systems demanding ~10kW/tower due to signal processing and beam steering technologies. This sees some countries (i.e. China) closing cell sites at night to save money.  So, what of 6G?  The assumption that all surfaces can be smart signal regenerators with beam steering looks be a step too far and it may be time for a rethink!

On the extreme end of the scale we have AWS planning to colocate their latest AI data centre (at 1GW power consumption) along side two nuclear reactors because it needs 40% of their joint output. Google and Microsoft are following the AWS approach and reportedly in negotiation with nuclear plant owners. Needless to say that AI train ing sessions and usage have risen to dominate the top of the IT demand curve. At this time, there appears to be no limits to the projected energy demands of AI, but there is a further contender in this technology race, and that is the IoT. In order to satisfy the ecological demands of Industry 4.0/Society 5.0 we need to instrument and tag ‘Things’ by the Trillion, and not ~100 Billion as previously thought!

Now let’s see, Trillions of devices connected to the internet with 5G, 4G, WiFi, BlueTooth, LoRaWan et al using >100mW demands more power plants…

1. Future Networking
V
Energy Limits
Felix Ngobigha & Peter Cochrane
https://petercochrane.com

2. HERE we are?
A species of infinite aspiration
in a world of finite resources!
Growing ‘Crisis Catalogue’:
Food
Water
Energy
Climate
Refugee
Pollution
Materials
Migration
Demographic
Social Division
Global Warming
++++

3. HERE we are?
A species of infinite aspiration
in a world of finite resources!
Human Demands
~1.3 - 1.5 Planets

4. Human
Machine
Symbiosis
Internet
Cyber
Systems
Computers
Automation
Mass
Production
Electricity
Water
Steam
Power
Global Growth of
Energy Demand
Evolution is unidirectional
~35k Power Plants Today
(~10,000 Nuclear Plant Equivalent)
Club of Rome Report 1972
“There are limits to, and catastrophic
consequences, from unbridled growth”

5. A ‘ManageMEnt’
BASED Crisis
Wireless, Satellite, Computing
It is not a lack of: Technology
Bandwidth
Products
Access
Speed
Reach
Et al
It is the Energy Cost: To-Environments
Of -Distribution
Construction
Maintenance
Generation
Storage
IT Causality Thread
And NEW CHALLENGES
Increasingly chaotic demand/growth
Unpredictable: Weather Systems
Communication
Computing
Industry
Society
Thermodynamics: There is no free lunch
There is always a cost

6. A ‘ManageMEnt’
BASED Crisis
Wireless, Satellite, Computing
It is not a lack of: Technology
Bandwidth
Products
Access
Speed
Reach
Et al
It is the Energy Cost: To-Environments
Of -Distribution
Construction
Maintenance
Generation
Storage

7. A ‘ManageMEnt’
BASED Crisis
Wireless, Satellite, Computing
It is not a lack of: Technology
Bandwidth
Products
Access
Speed
Reach
Et al
It is the Energy Cost: To-Environments
Of -Distribution
Construction
Maintenance
Generation
Storage
POOR DESIGN/ENGINEERING/
COMMERCIAL/OPERATIONAL
CHOICES? KEY DECISIONS
ARE AT THE CORE
LINE OF LEAST RESISTANCE
CHEAPEST IS BEST
GO WITH WHAT YOU’VE GOT
MAXIMISE ROI

8. 6G
5G
4G
3G
2G
~9 years
~9 years
~9 years
~9 years
OLD Models
New Bounds

9. 6G
5G
4G
3G
2G
1G 2G 3G
2.5G
Voice centric
Same genome
3.5G
~9 years
4G 5G
Data centric
New genome
6G
Smarts centric
Radical genome !
7G?
~9 years
~9 years
~9 years
OLD Models
New Bounds

10. 6G
5G
4G
3G
2G
1G 2G 3G
2.5G
Voice centric
Same genome
3.5G
~9 years
4G 5G
Data centric
New genome
6G
Smarts centric
Radical genome !
7G?
~3kW
~1kW
Greater BW
More Data
Lower Latency
More Towers
Denser Net
Higher Power
>20kW?
~9 years
~9 years
~9 years
~0.5kW
~10-20kW
OLD Models
New Bounds

11. 6G
5G
4G
3G
2G
1G 2G 3G
2.5G
Voice centric
Same genome
3.5G
~9 years
4G 5G
Data centric
New genome
6G
Smarts centric
Radical genome !
7G?
~3kW
~1kW
Greater BW
More Data
Lower Latency
More Towers
Denser Net
Higher Power
>20kW?
~9 years
~9 years
~9 years
~0.5kW
~10-20kW
OLD Models
New BoundsCHINA 5G ALONE IS NOW
PROJECTING A NEED FOR
~40 NUCLEAR POWER
STATIONS FOR 3.4M CELLS
5G ENERGY MANAGEMENT
MAY GET THIS DOWN TO
~10 NUCLEAR PLANTS
BEST ESTIMATE FOR PLANT
WIDE 5G COMMUNICATION
~50 NUCLEAR PLANTS
AND THEN COMES 6G
TIME TO RETHINK ?

12. AI LLM Training
ENERGY DEMAND
Google’s Greenhouse Gas Emissions Expand
by 48% Since 2019, Due to AI July 5, 2024
“All 3 technology powerhouses need to find
new sources of zero-carbon energy that can
operate reliably around the clock”
“50MW Data Centres to be overtaken by new
1GW replacements on nuclear plant campus”

13. AI MAY Cost
THE EARTH ?
Duke Energy announced agreements yesterday with Amazon,
Google, Microsoft, and Nucor to accelerate Carolina clean
energy deployments through new rate structures.
The company signed memoranda of understanding this month
that include proposed Accelerating Clean Energy (ACE)
tariffs that would help offset the long-term costs of investing
in clean energy technologies, such as new nuclear and energy
storage, through early commitments. Duke announced its
partnership with the tech and utility giants at this week’s
White House Summit on Domestic Nuclear Deployment
aimed at strengthening the U.S. nuclear industry.

14. AWS Ahead
of the Game

15. Greenhouse Gas
Emissions up 48% from
2019, due to AI
July 5, 2024
$1Bn
$20Bn
$100Bn/1GW
Projected

16. LLM AI Training
FLOP DEMAND

17. Energy Demand
Incomplete - PCs & devices

18. Energy Demand
Incomplete - PCs & devices
NOTICE THAT NETWORKING
(IE - OUR SECTOR)
IS THE LEAST DAMAGING
CONTRIBUTOR
WE OUGHT NOT TO BE
COMPLACENT, BUT START
TO LOOK FOR THE NEXT BIG
ENERGY DEMAND/SURPRISE

19. LAST MILE
TRAVESTY
Conceived in the
1980s when fibre was
expensive, but is now
the cheapest item and
we can ‘waste’ it to
eradicate electronics,
provide full bandwidth,
and improve reliability!

20. Data Centres
ENERGY DEMAND
Could use 13% of global electricity supply
and create a 6% carbon footprint by 2030

21. Data Centres
ENERGY DEMAND
Could use 13% of global electricity supply
and create a 6% carbon footprint by 2030
TO ME THIS LOOKS VERY
OPTIMISTIC AS THE RATE
OF AI GROWTH IS
ACCELERATING RAPIDLY

22. ESTIMATED GLOBAL
ENERGY DEMAND TWH
Best Case
Expected
Worst Case
Worst Case
Expected
Best Case
Data Centres
Consumer Devices

23. Worst Case
Expected
Best Case
Data CentresProduction
All consumer devices, networks,
and data centre equipment!
Worst Case
Expected

24. Fundamental
ComputE Limits
This is for digital Silicon,
but we also have:-
-Analogue Computing
-Quantum/Hybrid
-Optical Processing
-AI Generated Code
-AI Friendly Configs
And new materials:-
-Carbon NanoTubes
-Gallium Nitride
-Silicon Carbide
-++++
Ultimately this all comes down to the energy
density, thermal & interconnect management
-and we have some way yet to go with 3D
integration and multi-layering

25. 1.0
Hunter
Gatherer
2.0
Agrarian
3.0
Industrial
4.0
Information
5.0
Symbiotic
Augmentation
1.0
Artisans
Automation
2.0
Mass
Production
3.0
www
Computing
4.0
AI-Cyber
Robotics
~13,000 BC
~300-40k
years
~14k years
~200 years
~70 years
~25 years
Society
Industry
~1780 ~1950 ~2000
10
8
6
4
2
SO, Here WE ARE !
Driven by science & technology

26. 1.0
Hunter
Gatherer
2.0
Agrarian
3.0
Industrial
4.0
Information
5.0
Symbiotic
Augmentation
1.0
Artisans
Automation
2.0
Mass
Production
3.0
www
Computing
4.0
AI-Cyber
Robotics
~13,000 BC
~300-40k
years
~14k years
~200 years
~70 years
~25 years
Society
Industry
~1780 ~1950 ~2000
10
8
6
4
2
SO, Here WE ARE !
Driven by science & technology
WHAT IS THE NEXT BIG
DISRUPTER/CHALLENGE
IN THE IT SPACE?

27. IoT
At the core: greater connectivity
…and ‘The IoT’ as the nervous
system of societies and planet
wide resources
BIO-TECH nano-TECH
Lower energy
Lower waste
Less friction
New materials
New industries
New processes
New capabilities
I4.0 catalytic
starting point
AI
AL
QC
Robotics

28. IoT
At the core: greater connectivity
…and ‘The IoT’ as the nervous
system of societies and planet
wide resources
BIO-TECH nano-TECH
Lower energy
Lower waste
Less friction
New materials
New industries
New processes
New capabilities
I4.0 catalytic
starting point
AI
AL
QC
RoboticsTHE CURRENT APPROACH
TO ENGINEERING THE
IOT IS UNTENABLE ON
EVERY LEVEL
ENERGY HUNGRY
MATERIAL WASTEFUL
A SECURITY LIABILITY
UNTENABLE NETWORKS

29. LORA BASED
IoT MESHNET
It works, but cannot scale to satisfy
our ‘realistic’ IoT projections

30. Bluetooth Low
ENERGY meshnet
It works, but cannot scale to
satisfy realistic IoT projections

31. Networked for LIFE
This redefines the expression - ‘Always On-Line’

32. SWEden - HUMAN
IMPLANT PROGRAM

33. SMART HIGHWAYS
AND LOGISTICS

34. DUMB to SMART
Wired and wireless vehicle IoT
Dozens of sensors,
recorders, control
units, aggregated by
on-board analysis with
IoT/Internet connect
via multiple channels
WiFi,3, 4, 5G Direct
to Service Centre
Communication
Direct Car-to-Car
Avoiding 3, 4, 5G Congestion

35. Everything
Connected
Taking an integrated view of the prime
IoT objective, and biggest challenges!
To save the planet and humanity by
powering sustainability!
We have no suitable networks or wireless
systems & technologies!
We are amplify the Cyber Attack surface
and have no effective defences !
The energy costs outweigh any
advantage and only add to the problem!

36. Categories
Devices
Sensors
Vehicles
Systems
Buildings
Wearables
Appliances
Components
Infrastructure
Function
Safety
Security
Logistics
Inventories
Monitoring Use
Industrial Control
Monitoring Health
Historical Records
Environmental Control
Purpose
Repair
ReUse
Recycle
RePurpose
Materials Recovery
Environmental Policing
Performance Assurance
Maintenance Scheduling
++++
THE Categorising of
THINGS is Endless
Lists of this kind see population
estimates of ~100 Bn Things

37. 2000 04 08 12 16 2020
Initial Over
Optimism
Corrective
Pessimism
50
10
20
30
40
Actuality
Commercial IoT
‘Things’ Forecast
2030 - 50 looks to BE
a different picture

38. OLD Mindsets
Wrong era technologies
Wireless thinking and practice has
been stuck in a rut since 1920 !

39. LoRaWan 4/5G/LTE WiFi BlueTooth ZigNee
Range ~20km ~10km ~100m ~50m ~50m
BitRate ~50kb/s ~1Mbit/s ~500Mbit/s ~1Mbit/s ~200kb/s
Power
<25mW up
<500mW down
<1W up
<120W down
<0.5W
<100mW
>0.01mW
<100mW
>1mW
Topology Star Star Star-Mesh
P2P, Star,
Mesh, Broadcast
Star
Cost/Module ~$10 ~$20 ~$5 ~$3
~$15
sample IoT wireless
Adoptions to Date
All energy & space hungry: far too
expensive, complex, and unfit for
our future projected) needs

40. Recorded power
received over a
24 hour period
Mehdawi (2013). Spectrum
Occupancy Survey HULL
Artificial spectrum crowding
DUE To old wireless model

41. Allocated Bandwidth
Active Bandwidth
Band Occupancy =
VHF & UHF Seldom > 20% Occupancy
So FULL YET
So EMPTY !

42. O2
O2
H2O
H2O
<1.0% <0.01%
A natural filter for high
density short distance
networking
50
10
dB/km 1
0.1
0.01
0.0010
10 20 30 GHz 100 200 400
Occupancy <10%
NO spectrum Shortage
Above or Below 10GHZ

43. Degrees of
freedom
S/N
Frequency
Time
I = k.B.T.log2(1+kS/N)
Think information volume
and not signal space !
The same information conveyed by different
combinations of S/N , BW and Time
100% digital realisation, ultra low energy (µW- pW)
signal << thermal noise over very short distances
with little/no control of frequency occupancy

44. Degrees of
freedom
S/N
Frequency
Time
I = k.B.T.log2(1+kS/N)
Invisible @ >10dB below the thermal noise
Think information volume
and not signal space !
The same information conveyed by different
combinations of S/N , BW and Time
100% digital realisation, ultra low energy (µW- pW)
signal << thermal noise over very short distances
with little/no control of frequency occupancy

45. Wireless future ?
No bands, channels, regulation…
dBm
-40
-60
-80
-100
-120
-30
-50
-70
-90
-110
We can continue on
this path but failure
is certain!
Or we can adopt ‘massive spreads’ and
operate below the thermal noise level

46. Active Node
Repeater Node
Last Hop Node
NOTE 1: A simplified 3 hop
network is shown for clarity -
5 hop limits may be ideal…
Last Hop Vector sees
no further propagation
Aggregating Access Node
Common practice depicts nets as 2D, static,
and planar, which they are not! The grey
layer indicates many hidden dimensions.
CoNceptual
IOT MESHNET

47. IOT RICH
ENVIRONS
Meshnet configurations appear to have
predilection for dynamic clustering even when
starting from a random distribution

48. CoNceptual
SIMULATION
Meshnet configurations appear to have
predilection for dynamic clustering even when
starting from a random distribution

49. Prediction
/Surprise
“The IoT will see the spontaneous emergence
of new forms of Artificial Intelligence”
Things that think want to link
and
Things that link want to think

50. THE Last BIG
Question ?
“Will we be smart enough to recognise
new intelligences when they
spontaneously appear”

51. !e "ture belongs # $e
most adaptable and $ose
who dare !
!ank You

petercochrane.com