Link Budget and Radio Channel

Properties
Roger Birkeland

June 2018

Note
This is a part of my thesis, titled On the Use of Micro Satellites as Communication Nodes
in an Arctic Sensor Network, please do not distribute (yet).

1 Link Budget and Link Budget Parameters

A link budget is used to assess and evaluate the quality of communication links, and it
expresses the expected signal-to-noise ratio at the receiver. The result of the link budget
can be expressed as a required carrier-to-noise ratio ( N C
), the carrier to spectral noise
density ( N0 ), plus a margin, or as a sensitivity value.
C

The basic link budget equation is:

C Eb Gt P t Gr 1 1
= · Rb = · · · (1)
N0 N0 L0 Tsys k La

N0 is the spectral density of the noise, and Eb is the energy per bit. Gt and Pt are the
antenna gain for the transmitting antenna and the transmit power, respectively. This
term is often denoted Effective Isotropic Radiated Power (EIRP). Gr is the receiver

2
antenna gain. L0 is the free space loss defined as 4πdλ , with λ as the wavelength and d
is the distance between the transmitter and receiver. Tsys is the equivalent system noise
temperature, k is Boltzman’s constant and Rb is the data rate of the system. La is the
additional losses accounted for, and the individual factors will be identified and discussed
in the following.
To fully comprehend the link budget, different parameters and their effects should
at least be sized and estimated. Here, ionospheric scintillation, polar cap absorption,
Faraday rotation, polarization loss, multipath and dispersion will be considered.
Different effects inflict the signal in different manners, both with respect to fading
depth and the duration of the fading period. Short fading, scintillations and fading
due to for example shallow multipath, stemming from reflections from the sea surface,

1
can according to [1] be modeled as Additive White Gaussian Noise (AWGN). Common
Forward Error-correction Code (FEC) codes can mitigate this fading and increase the
margin. However, deeper fading will have a longer duration and cause more loss of
data than the codes can handle. To mitigate such effects, signal processing techniques
such as equalization, frequency and/or spatial diversity or packet retransmission could
be considered. The complexity of the mitigation techniques must be traded against the
gain in reliability each method gives.

1.1 Polarization Loss

Due to Faraday rotation, the polarization of a signal will change on its way to or from
a satellite. From the International Telecommunications Union (ITU) recommendation
ITU-R P.618-12 [2] we find that a 100 MHz and a 500 MHz signal will experience Faraday
rotation in the order of 30 and 1.2 rotations respectively. In order to counteract the
Faraday rotation, either the node or the satellite (or both) should have a circular polarized
antenna. This is typically done for frequencies below 6 GHz [1, Chapter 2.5]. For the
link budgets between sensor nodes and satellite(s), it is assumed that the sensor node has
a vertical linearly polarized antenna and the satellite has a circular polarized antenna.
The cost of this setup is 3 dB polarization loss and it is added to the link budget.

1.2 Dispersion
Different frequencies will experience different propagation delays through the ionosphere.
Due to this effect, the signal will experience dispersion. According to ITU-R P.531-12 [3],
these effects must be taken into account for wide band systems at VHF and UHF. An
example shown in [3] is that a 1 µs signal can experience a differential delay of 0.02
µs, which is 2% of the pulse duration. For a narrowband application, this is of less
importance.

1.3 Ionospheric Scintillation

The dynamics of small scale structures in the ionosphere can cause rapid changes in the
amplitude and phase of a signal traveling between a satellite and a ground node. These
irregular structures are due to local variations of electron density. This in turn causes the
refractive index to change, which will influence the signal [1, Chapter 2.3]. The fading
due to ionospheric scintillation can vary a lot; from small variations to deep fades that
could cause link outages. Also, the properties of this fade vary with the time of day, the
time of year, the geographical location and the sun activity.
The scintillation effects typically occur after local ionospheric sunset, meaning that it
is a phenomenon that takes place in evenings or at night-time. Typical event durations
are from 30 minutes to several hours. When the solar activity is at its maximum in a
solar cycle, these effects can be strong and occur every evening [3, Chapter 4.2]. The
polar areas are generally less affected than the equatorial zones, but aurora phenomena
can cause scintillation at high latitudes [4].

2
 Due to all these inter-connected phenomena, deriving a full statistical model for the
ionosphere is nearly impossible, according to Allnut [1, page 119]:
"the concept of annual statistics is of dubious merit for ionospheric phenom-
ena."
The goal here will therefore be to derive a suitable average value for ionospheric scin-
tillations margin to be used in the link budget.
A frequently used parameter to describe ionospheric properties, is the S4 index, which
characterizes the severity of amplitude scintillation. It is given by Briggs and Parkin [5]:

( I 2 − hIi2 )


2
(S4) = (2)
hIi2
I is the received signal amplitude and h i means average [1]. To find an estimate for
peak-to-peak values, equation ( 3) is used, where the S4-index from equation ( 2) is used.
This is an empirical formula estimated from observations [3]:

Pf luc = 27.5S41.26 (3)

Figure 1 (from [3, Chapter 4.1]), shows one example of VHF and UHF measurements
from Sweden in 2003. From this data, it is found that the scintillation indices, for VHF
and UHF respectively, are between approximately 0.4 to 1.5 and 0.2 to over 1. This
corresponds to peak-to-peak values between 3.5 to 27 dB (by using equation ( 3)). The
duration of high scintillation values are several tenths of minutes. This means that
the system might experience a signal fade during the whole satellite pass, with loss of
communication as the consequence given a large loss.
However, these peak-to-peak values do not provide an usable value for the link budget.
The scintillations can be described by the Nakagami-distribution [3, 1], and this can be
used to derive a margin. This distribution models deviations from an RMS value. The
parameter for the Nakagami-distribution, m = 1/S4, and its cumulative distribution,
shown in Figure 2, show for which percentage of the time the fading is worse than a set
level.
Scintillation effects are typically classified into weak (S4 ≤ 0.3), moderate (0.3 < S4 < 0.6),
and strong (S4 ≥ 0.6) regimes. Here, a value for S4 = 0.5 is chosen. This is in the
moderate regime. If S4 is chosen too low, the resulting link budget might be too opti-
mistic and higher outages or less throughput than expected will be experienced. On the
other hand, if S4 is estimated too high, then the link budget will be too pessimistic and
the full potential of the channel will not be exploited. S4 = 0.5 ⇒ m = 4 is believed to
be a reasonable choice.
teste
The red line in Figure 2 shows the situation when m = 4 (the curve for m = 4 is
lacking in the figure from [6], so the value is interpolated between the lines for m = 3
and m = 5) and the margin, 7 dB, shall not be exceeded by more than 1%. The blue
line shows the case for m = 4 and a 10% level and corresponds to 3 dB. These results
will be further discussed and concluded in the link budget calculation in Section ??.

3
 Figure 1: Scintillation indices measured in northern Sweden. From [3, Figure 5].

Figure 2 shows that the distribution is very sensitive. This causes the estimate of the
required margin to vary a lot with regards to the choice of S4. If we chose S4 = 0.3
(weak), m = 11. From Figure 2, we then get that an estimate for the fading margin
to be between 3 and 4 dB for both 99% and 90% levels. On the other hand, if S4 = 1
(strong), m = 1, then the required margin will be between 10 to 20 dB. For even higher
S4, the required margin will have to be huge.
It should be noted that uncertainties like this are further arguments to make use of
Adaptive Coding and Modulation (ACM) and Variable Coding and Modulation (VCM)
techniques to adapt the use of the channel to its real true-time properties (see Section 5).
Implementation of these concepts are suggested for future work.
STK can implement the ITU-models for atmosphere, troposphere, rain and ionosphere.
For a simple setup with a satellite in a 500 km 98° polar orbit with an UHF transmitter
and a ground station receiver on Svalbard, it is found that the average level for the
ionospheric scintillations is 7.3 dB. The fading value varies from 26 dB to close to 0
dB. This is plotted in Figure ?? on page ??. For S-band STK reports less than 1 dB
estimated ionospheric loss.

4
Figure 2: Nakagami-m distribution for ionospheric scintillation. Adapted from [6].

5
1.4 Polar Cap and Auroral Absorption
Polar cap absorption and auroral absorption are rare events. Auroral absorption can last
for hours. For elevation angles greater than 20° the signal loss at VHF is expected to be
less than 1 dB for most of the time as stated in ITU Recommendation 531 [3, Table 2].
Polar cap absorption is a very rare event, usually occurring 10-12 times a year during
sunspot maxima only. However, the signal loss can be significant, and the duration can
be on the order of days [3, Chapter 5.2].

1.5 Atmospheric Effects

Other atmospheric effects such as rain fading are not present at the frequencies considered
here; namely VHF/UHF and S-Band. In addition, a low amount of precipitation is
expected in the Arctic areas.
When modeling with STK, it is found that losses due to atmospheric, rain, and cloud
conditions are all less than 0.1 dB. They are therefore chosen to be neglected in the link
budget. For very low elevation angles, tropospheric scintillation is modeled to be 40 dB
loss, while for elevation angles above > 3°, this loss component is 0 dB. This effect will
also be neglected in the link budget as we assume operating at elevation angles above 3°.
These effects are similar for UHF and S-Band cases.

1.6 Other Effects

In addition to the losses mentioned above, the system will be prone to signal degradation
due to several other factors. Examples are antenna pointing errors due to misbehaving
Attitude Determination and Control System (ADCS). Ocean wave movements can drown
or shadow the node antenna. Icing of antennas can occur as well as multipath fading
effects due to reflections from ocean or ice surface. The value of these effects is hard to
estimate. A large system margin could account for some of these effects. Link losses can
in some cases be expected, especially due to antenna icing and shadowing. In addition, a
large (conservative) system margin will in most cases give conservative results indicating
lower data throughput than what is actually possible. Again, VCM and ACM should be
employed in order to make the best possible use of the link.
Since both polar cap absorption and the periods of high scintillation values are tied
to periodic ionosphere activity due to solar activity, we can argue that the link budget
should not account for the peak values of these two parameters. At rare and extreme
events, ionospheric scintillation can cause losses greater than the system fade margin. For
a non-critical communications system, it can be accepted that we lose communication
during strong/extreme events. However, some losses should be taken into account in the
link budget.

2 Receiver Noise Calculations

Several physical phenomena lead to noise and disturbance of a signal. Traveling through
the ionosphere and atmosphere, rain and so on will also increase the noise level of a

6
 Figure 3: Noise calculations for a receiver.

signal. These phenomena contribute to the ambient environment and temperature the
antenna observes. In addition, the components in a receiver contribute to the noise level.
Also, noise from the surroundings, it being as interference or "general" increased RF
noise levels should be accounted for. However, for sensor systems placed in solitude in
the desolate Arctic, this contribution is small.
An estimate of the resulting system noise can be calculated by using a model shown
in Figure 3. In this case, we reference the system noise at the input of the Low-Noise
Amplifier (LNA), after the loss from the feed-line between the antenna and the LNA.
From this, we can write the corresponding system noise temperature Ts as:

Ta 1
Ts = + T0 (1 − ) + Tr , (4)
L L
where Ta is the antenna temperature, T0 = 290K is the ambient temperature denoted
Tf in Figure 3, L is the loss (linear) in the feed-line and Tr is the equivalent noise of
the receiver. The noise of the receiver can in most cases be approximated to the noise
temperature of the LNA [7, Chapter 5.5.2.5], as long as the gain of the LNA is high.

2.1 Satellite Antenna Temperature - Uplink

The satellite antenna will, depending on the antenna pattern, see a portion of the Earth
as well as the empty space. The true antenna temperature is an integral over the bright-
ness temperature the antenna diagram "sees". For a low-gain antenna, the Earth will
contribute only a fraction of this temperature. However, as a conservative estimation
Ta = 290 K can be used [7, Chapter 5.5.3.1].

2.2 Ground Antenna Temperature - Downlink

For a receiving station on Earth, the observed antenna temperature will be lower, as the
antenna sees the cold space, in addition to the noise contribution by the atmosphere. If
the antenna has a broad lobe and a low elevation angle, the ground will also contribute.
From [7, Chapter 5.5.3.2.1] the antenna temperature can be estimated to be as low as 10
- 50 K for clear sky conditions. Referring to [8], galactic noise must also be taken into

7
account. Hence an estimate of 250 K for UHF and 100 K for S-Band will be used in the
link budget calculations in Section ??.
Figure 4 shows some of the contributions to the antenna noise temperature. It is for
example noise from the ground, the sky and from rain. Depending on the amount of
rain clouds, all contribution from the sky might be attenuated through the rain clouds.
Further, the noise temperature due to the rain clouds will then be greater. As mentioned,
the contribution from for example rain will be small in our case; both due to very little
precipitation and due to that UHF frequencies are not affected by rain at any great
degree.

Figure 4: General representation of antenna noise and temperature. Several physical

processes can contribute to the antenna noise. Some are shown in this figure.

3 Channel Coding
By introducing redundant bits in the data transmitted over the RF channel, we can
increase the probability of successful decoding, for a given signal to noise ratio. The cost
of this is that the introduced redundant bits reduce the usable bit rate. Various existing
Forward Error-correction Code (FEC) schemes can be used. The codes have different
properties; both when it comes to code strength and complexity in decoding [7, Chapter
4.3]. Figure 5 shows typical code gains for a selection of code rates, based on Viterbi
decoding of a convolution code. The figure is based on Table 4.5 in [7, Chapter 4.3.2].

8
 Figure 5: Typical coding gain. Adapted from Table 4.5 in [7, Chapter 4.3.2].

4 Modulation
For low-power operations, a simple, but spectral effective, modulation should be selected.
Figure 6 shows Bit Error Rate (BER)-curves for a selection of digital modulations. As
observed, the modulations BPSK and QPSK have the lowest Eb /N0 -requirement for a
given BER. As QPSK transmits two bits per symbol, giving twice the data rate compared
to BPSK, this modulation is chosen as baseline for link budget calculations.

Figure 6: Curves for Bit-Error-Rates for a selection of digital modulations.

9
5 Adaptive Links
The previous sections have shown that since the environment and the radio channel
are changing over time, the use of adaptive links should be considered. This means
that if the link budget for a given link condition, with respect to range, weather, and
ionospheric conditions, has room for extra capacity compared to other link conditions,
this improvement could be cashed out in different ways. For example:

• Power saving – less power is needed to maintain a link, still supporting the same
bit rate
• Higher throughput – same amount of power used, but a higher rate and/or higher
order modulation can be used when the link margin supports it

The challenging part can be how to define the signaling channel and the fall-back
modes, as the contact time during each pass is quite short. The link yield is depending
upon the transmit power, the distance between the transmitter and receiver and the
channel properties.

5.1 Causes of Variability

The following phenomena may cause high variability in link quality, and could therefore
be justifications on why adaptive links are sought for.

Distance The distance between the transmitter and receiver can easily be calculated
by knowing the satellite orbit. The maximum distance for a given pass can be calculated
in advance. As an example, the distance can vary between around 3000 km (horizon)
and 600 km (zenith). This corresponds to around 14 dB (20log( 3000
600 )) change in received
power level. An example of this is shown in Figure 7. Here, the usable dynamic range
is 10 dB. The steep cut-off is due to implementation of ITU-models for ionospheric and
tropospheric scintillation which STK estimates to be severe for low elevation angles.

Atmospheric and ionospheric conditions VHF and UHF frequencies are not very prone
to effects of atmospheric variations, such as attenuation due to rain or water vapor. Also,
the Arctic is mainly considered a desert. However, for south-north-moving satellites, the
link might go through parts of the atmosphere that contain more vapor. In addition,
ionospheric scintillations can occur as shown in Section 1.3. Even more, the link will also
be affected by solar energy bursts. These are rare events that can be monitored and a
forecast can be issued. If such effects occur, they can cause losses that are larger than
the system fade margin even if very low bit rates are used. Outages due to this will be
rare occurrences and should be tolerated for a system as discussed here.

Local conditions The link can be affected by fading due to several local conditions,
such as icing and wave movements if the node is floating. Reflections due to waves and
local surroundings can also lead to a changing multipath environment. The value of this

10
Figure 7: Example of estimation of received C/N during a near overhead pass. The black
line is the C/N-value, the green is distance from ground station to satellite. The
left Y-axis is in dB, the right Y-axis is km and the X-time unit is hours.

fading can be several dB. According to [9, Chapter 6.3], multipath can be severe for
any elevation angle in a low-frequency system using low gain antennas. The impact of
different parameters such as elevation angle and sea state is discussed in [1].

5.2 Adaptive Coding and Modulation

Since the received power level varies during the pass due to for example varying distance,
the link properties are quite different in the start and end of a pass, compared to during
the middle section of a pass (especially valid for overhead passes).
In order to make the most use of channel capacity, Adaptive Coding and Modulation
(ACM) or Variable Coding and Modulation (VCM) can be employed. This can be a
change of the code rate; so that packets transmitted in the start of a pass can have a
simpler modulation or have a stronger code than packets transmitted when the received
power is higher. This dynamics can be as high as 20 dB, according to [10]. If we slice
the curve in Figure 7 for every 2 dB, we can have 5 different Coding and Modulation
Scheme (CMS) if desired.
The extra implementation cost due to increased complexity on adaptive links must be
traded against the chance of achieving more data throughput in some passes.

5.2.1 Variable Coding and Modulation Schemes

In order to make ACM work, the receiving node (satellite, sensor node or gateway) must
constantly inform the transmitter of how the signal is received. This can be implemented
in several ways, one way is to note the output from the FEC process and derive an
estimate for the BER. If few packets are received, the transmitter should be requested to
reduce the data rate or change to a stronger code. A time-out should be in place both in
transmitter and receiver in order to fall back to a basic CMS in case of failure of a more
"advanced" CMS.

11
 Figure 8 shows how the loop controlling the ACM-functionality can behave. In the
satellite there will be a function estimating the received link quality which in turn in-
structs the sensor node to use a specific CMS suitable for the present radio channel. An
implementation for an SDR is proposed in [10].

Figure 8: Model of ACM control loop.

5.2.2 Rate-less Codes - Hybrid ARQ

A method to transmit as much usable data as possible, which also enables the ability to
handle varying reception properties is to employ rate-less codes. A data transfer between
the sensor node and the satellite can start with a coding scheme where data with just a
little code is transmitted. However, the transmitter still makes a very long code word.
If the packet is received correctly (determined by a CRC-code), then an ACK can be
issued. However, if the decoding of the packet fails, then an ARQ is issued. The source
will then issue an additional segment of the code word, that the receiver will add to the
first received packet. This process may continue until the packet is successfully decoded,
or until a timeout is reached. This method is also called Hybrid ARQ [11].

References
[1] J. E. Allnut. Satellite-to-Ground Radiowave Propagation. The Institution of Engineering and
Technology, 2nd edition, 2011. ISBN 978-1-84919-118-0.

[2] Recommendation ITU-R P.618-12 - Propagation data and prediction methods required
for the design of Earth-space telecommunication systems. ITU, 07 2015. URL
https://www.itu.int/rec/R-REC-P.618/en.

[3] Recommendation ITU-R P.531-12 - Ionospheric propagation data and prediction meth-
ods required for the design of satellite services and systems. ITU, 09 2013. URL
https://www.itu.int/rec/R-REC-P.531/en.

[4] Paul M. Kintner jr, Todd Humphreys, and Joanna Hinks. GNSS and Ionospheric Scintillation -
How to survive the next solar maximum. Inside GNSS, 2009.

12
 [5] B. H. Briggs and I. A. Parkin. On the variation of radio star and satellite scintillations with zenith
angle. Journal of Atmospheric and Terrestrial Physics, 25:pp 339–365, 1963. doi: 10.1016/0021-
9169(63)90150-8.

[6] Simon R. Saunders and Alejandro Aragón-Zavala. Antennas and Propagation for Wireless Com-
munication Systems. Wiley, second edition, 2007. ISBN 978-0-470-84879-1.

[7] G. Maral and M. Bousquet. Satellite Communication Systems. Wiley, 5th edition, 2009.

[8] Recommendation ITU-R P.372-13 - Radio noise. ITU, 09 2016. URL

https://www.itu.int/rec/R-REC-P.372/en.

[9] Louis J. Ippolito Jr. Radiowave Propagation in Satellite Communication. Van Nostrand Reinhold
Company, 1986. doi: 10.1007/978-94-011-7027-7. ISBN 978-94-011-7027-7.

[10] E. Grayver, A. Chin, J. Hsu, S. Stanev, D. Kun, and A. Parower. Software defined ra-
dio for small satellites. In 2015 IEEE Aerospace Conference, pages 1–9, March 2015. doi:
10.1109/AERO.2015.7118901.

[11] P. Frenger, S. Parkvall, and E. Dahlman. Performance comparison of harq with chase com-
bining and incremental redundancy for hsdpa. In IEEE 54th Vehicular Technology Conference.
VTC Fall 2001. Proceedings (Cat. No.01CH37211), volume 3, pages 1829–1833 vol.3, 2001. doi:
10.1109/VTC.2001.956516.

[12] Alliance for Telecommunications Industry Solutions. ATIS telecom glossary 2000 : T1.523-2001,
2001. URL http://www.atis.org/glossary/. Accessed February 2018.

Acronyms
ACK ACKnowlegde. 12
ACM Adaptive Coding and Modulation. 4, 6, 11, 12
ADCS Attitude Determination and Control System. 6
ARQ Automatic Repeat reQuest. 12
AWGN Additive White Gaussian Noise. 2

BER Bit Error Rate. 9, 11

CMS Coding and Modulation Scheme. 11, 12

EIRP Effective Isotropic Radiated Power. 1

FEC Forward Error-correction Code. 2, 8, 11

ITU International Telecommunications Union. 2

LNA Low-Noise Amplifier. 7

SDR Software Defined Radio. 12

STK Systems Toolkit. 4, 6, 10

VCM Variable Coding and Modulation. 4, 6, 11

13
List of Definitions
RMS Root-mean-square, a way of describe an average signal level. 3

S-Band A radio communication band that spans the 2 - 4 GHz frequency range (IEEE definition). 6, 8
sensitivity The sensitivity (level) of a receiver can be described as the signal needed at the receivers
input in order to get the required signal-to-noise ration on the output for a given bit-error-rate
and modulation [12]. 1

UHF Ultra High Frequency. A radio communication band that spans 300 - 3000 MHz (ITU definition).
Usually used to denote frequencies from 300 - 1000 MHz (IEEE definition). 2–4, 6, 8, 10

VHF Very High Frequency. A radio communication band that spans 30 - 300 MHz (ITU and IEEE
definition). 2, 3, 6, 10

14