1.0 Introduction
-----------------
-----------------
One of the biggest challenges facing this planet is climate change. It is estimated that ICT consumes 5-10% of
global electricity and contributes to 2% of GHG emissions. Greentouch predicts that ICT energy
consumption is going to double in the next decade. As Price Waterhouse Coopers (PwC) stated in a
recent report the current global rate of de-carbonization is .8%. We need to achieve a rate of 5.1% for the next
39 years if we are to keep global average temperatures increasing less than
2C. That means that we have to increase
the rate of de-carbonization by 40 times, for all sectors of society including
ICT, starting right now.
Greentouch claims that already ICT equipment is achieving 10-20%
increase in energy efficiency per year. But
given the growth of ICT of approximately 10% per year, overall efficiency is
probably neutral.
Increased energy efficiency does not have a one to one
correspondence to reduction in GHG emissions.
The reduction in GHG emissions depends on the local energy mix. So for
example in the US you need 30% increased energy efficiency to achieve 10%
reduction in GHG emissions, as coal powered energy is 30-40% of the energy mix. Overall globally coal powered electricity is
typically 50% of the electrical energy supply which means that you need a 20%
increase in energy efficiency to achieve 10% reduction in GHG emissions.
To meet the PwC target therefore would require the ICT sector would
have to achieve overall energy efficiency of 8000- 10000% per year!! Clearly to achieve this greater rate of de-carbonization
we will need much more radical measures to reduce the contribution of ICT to
GHG emissions. Efficiency on this scale
essentially means we must build zero carbon ICT solutions.
Although ICGT
contribution to GHG emissions is relatively small at 2%, it is one of the
fastest growing sectors in terms of GHG emissions. For example, new studies suggest that the growth in wireless networks could be the single largest component of that growth in CO2 emissions from the ICT sector. In a recent report by the Centre for Energy-Efficient Communications, at the University of Melbourne-based research centre claimed that by 2015, the energy used to run data centres will be a "drop in the ocean", compared to the wireless networks used to access cloud services. The report predicts that by 2015 energy consumption associated with 'wireless cloud' will reach 43 terawatt-hours, compared to 9.2 terawatt-hours in 2012. This is an increase in carbon footprint from 6 megatonnes of CO2 in 2012, up to 30 megatonnes of CO2 in 2015, which is the equivalent of an additional 4.9 million cars on the road, the report states.
More worrisome is another report from Sweden KTH that predicts will need to increase the density of wireless base stations by 1000 times to meet the insatiable demand for the “wireless cloud”. If this came to fruition, it would be incredibly huge jump in the demand of electricity by the ICT sector.
The wireless industry in particular is an ideal sector to be powered by local renewable energy sources such as solar panels and windmills. Already many wireless towers in the developing world are powered by renewable energy (but unfortunately often with diesel backup). Because of it is inherently distributed, lower power architecture the wireless industry is ideally suited to be powered by local renewable energy.
Achieving GHG reductions as suggested by PwC forecasts is not uniformly attainable across all sectors of society. Some sectors such as airplane travel will have extreme difficulty achieving any meaningful reduction. Fortunately the ICT sector, with the right architecture can possibly achieve the required GHG reductions and more, thereby compensating for those sectors. As well, the lifecycle of ICT products and services is very short, typically 5 years. By designing zero carbon solutions today, it is conceivable that within 5 years we will be able to achieve significantly better overall energy efficiency for the sector thereby helping other sectors less amenable to GHG reductions.
More worrisome is another report from Sweden KTH that predicts will need to increase the density of wireless base stations by 1000 times to meet the insatiable demand for the “wireless cloud”. If this came to fruition, it would be incredibly huge jump in the demand of electricity by the ICT sector.
The wireless industry in particular is an ideal sector to be powered by local renewable energy sources such as solar panels and windmills. Already many wireless towers in the developing world are powered by renewable energy (but unfortunately often with diesel backup). Because of it is inherently distributed, lower power architecture the wireless industry is ideally suited to be powered by local renewable energy.
Achieving GHG reductions as suggested by PwC forecasts is not uniformly attainable across all sectors of society. Some sectors such as airplane travel will have extreme difficulty achieving any meaningful reduction. Fortunately the ICT sector, with the right architecture can possibly achieve the required GHG reductions and more, thereby compensating for those sectors. As well, the lifecycle of ICT products and services is very short, typically 5 years. By designing zero carbon solutions today, it is conceivable that within 5 years we will be able to achieve significantly better overall energy efficiency for the sector thereby helping other sectors less amenable to GHG reductions.
2.0 Purpose of this paper
-------------------------
-------------------------
The purpose of this paper is to lay out a possible zero
carbon architecture for networks and ICT equipment. It is predicated on the following
assumptions:
(a) Solar power will become cheaper
than power from the grid within 2 years in Europe and 5 years in North America
(b) Most of the growth and
availability in solar power will be from off grid, highly distributed and
relatively small sources such as roof top and mobile systems.
(c) Even with low cost solar, for
the foreseeable future utilities will continue to supply a major portion of
their power from fossil fuel resources, because of sunk capital cost and
reliability of such power.
(d) Renewable energy will constitute
only a small portion of the overall energy mix and large scale storage is still a pipe
dream.
(e) The use of Renewable Energy
Credits (RECs) from the utility will not reduce meaningful dependence on fossil
generated fuel because of need for dependable and reliable power.
(f) In jurisdictions that have
mandated the utilities to use renewable energy such as California at 30%, the
utilities will be motivated to find customers who can use the unpredictable
power feed represented by this renewable power.
The network architecture also assumes that small scale, low power
computing devices will almost be universal and in the same cost range as a
single solar chip e.g. Raspberry PI. In
fact it is conceivable that some solar arrays will be manufactured with integrated
computing, network processors and memory storage. As a result we may well end
up with a surplus of distributed computing and storage whose availability will
wax and wane depending on the time of day.
It is clear that any dependence on the electrical grid,
which will still incorporate fossil fuels for the foreseeable future will make
it impossible to achieve 8000-10000% energy efficiency never mind zero carbon solutions. Based on
these assumptions this paper proposes a zero carbon ICT architecture that will
primarily be based on a highly distributed, off grid, small scale solar power,
sometimes supplemented by off loaded grid power in those jurisdictions that
mandate the use of significant percentage of renewable power.
The architecture will involve different approaches for
different ICT devices broken down as follows:
(a) Access Networks – Optical and
wireless - Reverse passive optical networks (RPON) and wireless mesh networks
with software defined radios
(b) Computing – Infrastructure powered solely by renewable energy and highly
distributed computing as above
(c) Consumer Devices – Multiplex AC
power systems and PoE solar powered charging stations
(d) Backbone networks, Base
stations and routers– Solar panel arrays
supplemented with autonomous eVehicle mobile storage systems
An overview of the architecture of each of the above
categories is provided in the following sections.
3.0 Zero Carbon Access Networks
-------------------------------------
-------------------------------------
With zero carbon access networks, whether wired or wireless,
the fundamental assumption is that all active network devices will have to move
to the edge of the network, where the power is located. The core of the network,
as much as possible, needs to be made up of passive devices. The best analogy is to apply the concept of
BitTorrent discovery and routing to the physical infrastructure.
Considerable progress has been made in this space with
wireless mesh networks. For optical networks RPON (Reverse Passive Optical
Networking) is the proposed solution where all active lasers and DWDM equipment
is located at the edge of the network on the customer’s premises integrated
with, or powered by their solar panels.
Not unlike wireless networks, multiple independent optical paths are
maintained with nearby neighbors and central core optical switches. In
fact, integration of both wireless and wired networks would make sense to
achieve greater density and throughput.
All forwarding and routing would be done at the edge in the CPE equipment with multiple
overlapping distributed forwarding tables using a Hadoop based routing table.
Solar powered, low powered GSM and Wifi is a well established
technology. Solar powered Raspberry PI
GSM,FTP and routing engines are also freely available.
Ultimately higher aggregation and routing would make sense
at traditional routing nodes as described in section 6.0
This architecture can also be used on backbone networks as
well, where the core backbone is only made up of optical amplifiers and wave
selection switches.
4.0 Computing
Infrastructure
------------------------------
------------------------------
A zero carbon computing infrastructure could be made up of two
components – large scale infrastructure powered solely by renewable energy such
as GreenQCloud and highly distributed computing and memory at the edge as
described in the previous section.
Considerable research has already gone into these highly
distributed computing architectures that need not be replicated here. The only significant difference is that in this
situation is that storage, memory and computing is not persistent. For example see Green Hadoop and Greenstar.
5.0 Consumer Devices
-------------------------
Most consumer ICT devices, except perhaps for TVs and
printers have internal battery storage.
These devices could easily be charged with special charging systems that
only renewable energy from rooftop solar panels such as Power over Ethernet
(PoE) or multiplex power systems. TVs
and printers would require some external storage device. Multiplex power systems provide both 400 HZ
and 60 (or 50 Hz) power over the same copper wiring. They are currently used in aircraft and
military systems. The 400 Hz power is only visible to special plug in adaptors 9 (or perhaps USB connectors) and is for the most part invisible to regular AC devices because of reactive
filtering from transformers and motors.
For consumer devices the biggest challenge is not
technology, but getting standards bodies to agree that all consumer devices
should use charging adaptors that work only with PoE or Multiplex AC.
Universities and other public institutions could set up
charging stations with multiple outlets that are powered solely by roof top
solar panels and supplemented by autonomous eVehicle mobile storage, with
powered delivered from garage or parking lot over AC multiplex systems or PoE.
Surplus grid power from renewable sources, made available because
of mandatory requirements to carry renewable power could also be used. However, complex signaling and contract negotiation
between grid operator, utility and
customer to make this happen.
6.0 Backbone networks, base stations and core routers
-------------------------------------------------------
-------------------------------------------------------
These devices are generally the most energy intensive in the
ICT sector and because of their design require high reliability. Moving as much computing and routing to the
edge, as described previously will help alleviate some of their power
requirements but not all.
Thousands of solar and wind powered radio base stations and
network nodes are already deployed. But
diesel generators are the usual backup power supply in case no renewable power
is available. An alternate approach is
to use autonomous eVehicles for mobile storage.
The vehicles could be moved from site to site based on predicted power
load etc. For the most part autonomous vehicles
would be parked at base station or network node and charged from surplus power
generated locally by solar panels or windmills.
But in the event of predicted severe climate or long periods of cloudy,
windless days, autonomous eVehicles could be directed to drive to nearby
roadside solar arrays or energy routing exchanges to pick up additional
supplemental power.
PWC study on de-carbonization
Greentouch ICT energy statistics
UBS study that solar will be cheaper than grid power
Reverse Passive Optical Networking (RPON)
Green Hadoop