IEEE Circuits and Systems Magazine - Q1 2023 - 63

Table 1.
Exemplary modeling parameters for various smartphones. The first device was measured during local playback and
capture, the second device and the third device were connected using a Wi-Fi connection. If parameter ranges are
displayed, the true value depends on video parameters such as the bitrate. The asterisk is a replacement for the index
* ˆ= " s,UT, d, r " .
p∗,0
p∗,Rx
p∗,Tx
Samsung Galaxy S III [43], 2013
805 mW
279 - 1,524 mW
min. 1.809 W
Fairphone 2 [16], 2020
900 mW
up to 2.21 W
-
HTC G1 [57], 2011
319 mW
-
min. 1.05 W
where we multiply the total power consumption of the
device with the duration of the requested streaming
event tr. The total power consumption is the sum of the
constant offset power consumption ps,UT,d,r,0, the receiving
and playback power ps,UT,d,r,Rx, and the capturing and
upload power consumption ps,UT,d,r,Tx. If only one direction
(Rx or Tx) is active, the power for transmission in
the opposite direction is set to zero.
First, we discuss the constant offset power ps,UT,d,r,0
for a single streaming request. This power mainly
depends on the type of device and its components.
For example, a smartphone usually demands only a
few Watts [16], [43] due to its energy efficient architecture.
Modern tablet PCs require about 10 to 30 W
[1], [42], the power consumption of TV sets mainly
depends on the display technology (liquid-crystal
display (LCD) or organic light-emitting diode (OLED)
display) and their screen size (50W to 200W) [41], and
the power consumption of desktop PCs depends on
their hardware configuration and the size and type
of the monitor (power consumption values similar
to TVs) [44].
For the receiving and playback power ps,UT,d,r,Rx,
the effect of certain video parameters on the power
was studied in various papers. For example, a very
detailed study was performed on the influence of
video parameters such as bitrate, resolution, frame
rate, and video codec for smartphones [16]. First, it
was found that the power consumption depends on
the smartphone model and the chosen application,
which indicates that also the software implementation
(in terms of the written code) affects the power
consumption. Moreover, the choice of the network
connection (wireless, wired) was shown to influence
the power consumption of the device. In general, it
was found that during online video streaming, up to
50% of the power consumption (including the constant
offset power) can be controlled by manipulating
the aforementioned parameters [16]. Similar results
could be observed for laptops and desktop PCs
[42], however, the constant offset power had a higher
FIRST QUARTER 2023
TABLE 2.
Rough modeling parameters for further enduser
devices as reported in [1]. The asterisk is a
replacement for the index * ˆ= " s,UT, d, r " .
p∗,0
Tablet
10 W
Laptop
20 W
TV
100 W
PC
100 W
influence in general. Unfortunately, to the best of the
authors' knowledge, similar studies on power consumption
are not available for TVs. However, due to
the similarity of the technologies used, we expect
that the same kind of dependencies can be observed.
A detailed analysis of this conjecture is an interesting
topic for future research.
Other studies focused on specific subprocesses in
end-user devices such as video decoding in hardware
and software [45], [50], where also devices such as desktop
PCs were considered. The influence of streaming
protocols and different kinds of networks (wired/wireless)
was studied in [51] and [52]. Furthermore, it was
shown that for LCDs, the backlight brightness can be
adjusted to the brightness of the video content without
sacrificing image quality, which can save power, especially
for large screens [53].
The capturing and uploading power consumption
ps,UT,d,r,Tx was discussed in detail in [54], [55], [56], [57],
and [58]. However, to the best of the authors' knowledge,
no dedicated power or energy models were developed
yet, which is another interesting topic for future
research.
Table 1 shows specific power values for different
smartphones from three different studies [16], [43],
[57]. Furthermore, Table 2 lists some rough power estimates
for other devices (tablet PCs, laptops, TVs, and
desktop PCs) that were reported in [1], where we could
only find data for the constant offset power ps,UT,d,r,0.
Values for the other parameters are not given because
they were not provided in the respective literature. It
is further important to note that some of the values
IEEE CIRCUITS AND SYSTEMS MAGAZINE
63
IEEE Circuits and Systems Magazine - Q1 2023

Table of Contents for the Digital Edition of IEEE Circuits and Systems Magazine - Q1 2023
