IEEE Power Electronics Magazine - September 2021 - 23

for each person on Earth, but will have many hundreds of
networked sensors. If looking to optimize energy efficiency
(EE), we should focus on device energy consumption rather
than increasing bandwidth. First, we focus on the savings
of mitigating power generation and transmission, then we
elaborate on the savings gained by emphasizing the importance
and opportunities delivered by edge computing.
Specifically, one must separate the application from the
power consumption (better perceived as energy utilization)
and focus on the Power Value Chain (PVC) between original
energy source (i.e. - a power plant or a battery) and endload
(i.e. - a smart phone or a wireless sensor network or
WSN). Figure 2 shows what a PVC looks like with the unitless
Power Cost Factor (PCF) metric, a relative measure of
the load's energy expressed as a multiplier of the overhead
to get energy to the consumption point. A self-powered
device has PCF=1x, but if derived from a power
plant transporting only 1% to the end-load,
becomes 100x.
System Level Impact of 5G IoT
Devices on Systems
Can a Microwatt-Scale-Device Have
a Significant Impact on Gigawatt-Scale
Utility Distributions?
When it comes to a massive amount of tiny, networked
devices serviced by very few base-stations
or small cells, then the answer to the
question is a resounding " Potentially! " Energy
for edge data already has a PCF = 103 - 106x
purely due to wireless transmission inefficiencies.
A base-station is ~10% efficient (PCF =
102x) and even ~10 - 15% is lost in utility transmission
(PCF=1.1 - 2x). Multiplying PCFs of
black boxes in the PVC yields 108x or more.
This may seem like an unreasonably large multiplier
but a 5G-NR-compliant radio (sensitivity
of -110 dBm/10−14 W) can still utilize a 0 dBm
(10 mW) signal from a base station so technically
we can have a PCF = 1012x in the worst
case for transmission alone. Each milliwatt
transferred ultimately draws 10s of watts from the power
plant, i.e. an edge-load creates 10+ orders of magnitude
demand on the power source.
For IoT, device density is hugely dependent on the application.
Though not unique to 5G this does explain the potential
for highly disproportionate growth of network energy
consumption relative to network traffic. The massive scaling
of " things " served by a single access point enabled by IoT,
coupled with 5G, is the game-changing-scenario here. If this
disproportionate demand is neither buffered (i.e. - energy
storage) nor mitigated (i.e. - edge processing, TinyML, MEC,
etc.), then all networks and systems deployed are at risk if
unable to fully utilize their potential. One thing we know
for sure is that the smaller the cell, the more opportunity
for increased energy optimization (mostly achieved by the
ease of quickly turning off stuff not in use from the μs to the
min/hr timebase). Utility grids will potentially be unable to
support the demand of a huge increase in access points, and
therefore ultimately support all the edge devices.
The 5G Energy Gap
Having demonstrated how 5G and massive scale deployment
can have a highly disproportionate cumulative draw
on networks and utility distributions, consider some additional
implications. Energy is not unlimited so either system
limitations (i.e. - over-current protection, over-temperature
protection), upstream components (i.e. - breakers, grid frequency
monitoring), or physics (i.e. - utility transformer
blows up) will eventually instantiate a hard limit. Most systems
are designed to restrict/foldback energy to a point of
failure to maximize application utilization. This means the
doomsday scenario described above is much less likely to
result in blowing transformers on streets and more likely to
result in derating (see " Derating the 5G Network " [3]). This
is articulated as the 5G Energy Gap (5GEG) [4], and further
elaborated as the 5G Derate Factor (5GDF) metric.
Moving outward from direct system and network reliability
impacts of tiny-device demand at scale, we take an
opposite approach in our analysis. What benefits are gained
by mitigating this highly-disproportionate power transmission?
Note more power is spent moving data than doing anything
else with it so the less data moved the more energyefficient
we become. We still require computation, but a
focus on increasing the edge-system budget with low-PCF
energy sources (i.e. - energy harvesting or EH) enables us to
keep those processing watts at the edge and gain the benefit
of lower PCF and data transmission mitigation concurrently.
User experience is typically top priority to any service
provider. If this suffers the network will be unattractive to
end-users and fail to drive the necessary subscriptions (or
however monetized) making the effort for naught. Analogous
to descriptions of PVC and use of PCF, most successes
depend on the edge so if the revenue is not there the entire
ecosystem suffers and is unattractive for economic and
technical resource investment. [5]
The deployment of cellular infrastructure requires massive
capital investment and therefore has critical payback
calculation associated with determining ROI. Careful
attention to EE plays a large role in the realization of payback
calculations. This is more obvious in things like OPEX
of energy, but less obvious opportunities also exist and are
not of equal weighting.
Now we see where the worlds of 5G and EH/micropower
management overlap. Specifically, a major valueadd
of EH is to provide/supplement system energy at the
consumption point (again, PCF of 1x) by capturing ambient
energies. Just as cellular network economics are complicated,
the justification and success of EH implementations
(particularly in terms of TCO) are highly-dependent on the
payback calculation methodology. For instance, adding
September 2021 z IEEE POWER ELECTRONICS MAGAZINE 23
WIND TURBINES-©SHUTTERSTOCK.COM/METAMORWORKS, 5G-©SHUTTERSTOCK.COM/THE MASTERPLAN STD
IEEE Power Electronics Magazine - September 2021

Table of Contents for the Digital Edition of IEEE Power Electronics Magazine - September 2021
