> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1
Digital Modes for Ionospheric Radio Network
Inter American University of Puerto Rico – Bayamón Campus
In this paper, the HF band and channel characteristics will
Abstract—This paper presents a recommendation of a modern be discussed first. Then, the commonly used Watterson model
digital mode to transmit textual information over the HF band. will be summarized as along with the CCIR 520-1 and ITUR
This mode will be used in an interstation communication system F.1487 recommendations used to simulate the HF channel.
for a network of geophysical measurement stations in the The methodology used and setup of the test will also be
Antarctic. An HF channel simulator was used to develop effective explained in detail. Finally, the simulation results will be
comparative tests for all selected digital modes. The results from
shown in conjunction with the conclusions drawn from them.
the channel simulator helped in choosing the correct digital mode
according to the most reliable text transmission.
Index Terms—HF Channel, Watterson Model, Digital Radio II. HF CHANNEL
Modes, HF Propagation The high frequency (HF) band ranges from 3 to 30 MHz
and is used primarily for maritime, aeronautical, and military
systems as well as for long-distance AM broadcasting and
I. INTRODUCTION amateur radio operations. The nominal bandwidth allocation is
A S a propagation channel, the ionosphere has proven very
useful for transmitting over long distances using the High
Frequency (HF) radio band. The ionosphere, however, is
3 kHz. HF radio has been proven to be an inexpensive, low-
power, yet reliable, means of long-distance communication
without need for additional relay stations such as satellite or
typically characterized by interrelated temporal effects on telecommunications line . It is excellent for isolated and
communication, making it a highly variable medium over a remote areas where other communication infrastructure is
wide range of time scales. These effects include, but are not nonexistent.
limited to, common white noise, Doppler spread, Doppler The HF band, as well as other bands, possesses direct and
shift, and time delay . These factors cause distortion in both ground wave propagation, but with the addition of sky wave
propagation, which is not present on other frequency bands
signal amplitude and phase, causing loss of synchronization
higher than HF. This propagation mechanism is responsible
and data misinterpretation in a communication system. Digital
for long-distance communication because the radio signals
mode selection is important as it affects error correction as
that enter the ionosphere are refracted back toward the earth
well as character synchronization. by the E and F layers of the ionosphere, depending on
This paper formulates a recommendation for the use of a ionospheric conditions, which can change with location, time,
modern digital mode to transmit textual information over the and space weather conditions.
HF band. The recommended mode will be used in an The ionosphere as a channel of transmission is generally
interstation communication system for a network of characterized as a multi-path time-varying fading channel that
autonomous and dynamically adaptive, low-power contains time delay, Doppler shift, noise, and interference,
geophysical measurement stations in the Antarctic . resulting in a degradation of transmission quality.
Even though on-air tests can be performed, ionospheric The channel exhibits fluctuating signal-to-noise ratio (SNR)
conditions are impossible to control and do not yield caused by several sources. Radio noise can arise from natural
satisfactory results for comparative tests between modes. forms such as atmospheric noise, caused mainly by cloud-to-
What is needed is a way of simulating an ionospheric test ground flashes, as well as cosmic noise that mostly affects the
medium where parameters can be controlled. A model that higher portion of the HF band. In addition, manmade noise is
simulates an HF channel is the key to developing an effective present and can range from electric power sources and electric
comparative test of all digital modes . A computer cables to computer clocks. Additionally, crosstalk caused by
simulation is the best way to obtain significant results. signals in the same frequency at various power levels as well
An HF channel simulator is used to choose the correct as adjacent channels close in frequency with significant power
digital mode according to the most reliable text transmission. can cause de-sensing in the receiver.
A station will send a predetermined message to all other Ionospheric turbulence causes Doppler spread in the form of
stations on the chain when needed. A nearly error-free fading of the signal, originating from multiple reflections
transmission is needed to ensure the message gets through in a (multi-paths) of radio signals between the earth and different
timely manner. layers of the ionosphere. Furthermore, since the heights of the
layers of the ionosphere change during the day, different
frequency shifts on each of the multi-path components will
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2
III. HF CHANNEL MODEL
The commonly used Watterson model forms the basis for
most modern HF channel simulations. This model proposes a
narrow-band HF channel that assumes it is both stationary and TABLE I
stable . Although HF channels are nonstationary in both
Modes Tested Mode Production Modulation
frequency and time, this model was validated with on-air Capability of Modem
measurements and found to be valid for only short periods of AMTOR FEC AMTOR FEC & ARQ FSK
time (about 10 min) and small bandwidths (<12 kHz). This ASCII 110 ASCII FSK
model helps to develop a controlled and repeatable
CW 20 CW OOK
environment for performing tests.
Based on the Watterson model, several channel conditions PACTOR FEC GTOR FEC & ARQ FSK
can be used for an HF ionospheric channel simulator. The PSK31 PACKET FSK
International Radio Consultative Committee (CCIR), now RTTY 45 PACTOR FEC & ARQ FSK
called the International Telecommunication Union (ITU), has
listed various channel conditions for the HF band. This
document, CCIR Recommendations 520-1, defines values for RTTY FSK
frequency spread and delay times . These recommendations
include the commonly used CCIR Poor, Moderate, and Good
channel conditions. The ITU-R F.1487  adds some
additional models based on latitude. It has conditions for low,
mid, and high latitudes, including three sub-conditions for
The modes used for the tests were chosen according to the
modes the selected HF digital modem was capable of
transmitting and receiving. The tested modes are shown in
Table I. Note that the only modes considered are of the
broadcast type; in other words, those not requiring
acknowledgement from the receiving end since this cannot be
simulated. The goal of the tests was to determine suitability of
the modes to be used in an interstation communication system
in the Antarctic. Relative comparisons were made so as to Fig. 1 Block diagram of components used to create one of three possible
make a recommendation, and compare their performance simulation paths.
under certain conditions.
A message containing 330 characters, including spaces, was SELECTED CHANNEL CONDITION PARAMETERS
used for transmission. It contained a pangram repeated six
Condition Differential Doppler Spread
times. This message was generated using software called
Delay (ms) (Hz)
MultiPSK , which is used for digital mode operation and
CCIR Good 0.5 0.1
creates corresponding audio tones for each mode. Each
message was then recorded in a wave audio file format. This CCIR Moderate 1.0 0.5
audio file was then imported on the path simulator software
CCIR Poor 2.0 0.5
 that did the work of simulating the ionosphere using the
Watterson model. Fig. 1  shows the overall simulator data High-Latitude Moderate 3 10
path. The parameters of the simulator were adjusted according
High-Latitude Disturbed 7 30
to CCIR 520-1 and ITU-R F.1487 channel conditions. Table II
shows the selected conditions’ parameters. The end result of
the path simulator was played back onto another computer V. SIMULATION AND RESULTS
where MultiPSK decoded the message. The message received Fig. 2 illustrates how the modes performed according to
was compared with the original, and the character error rate different SNR conditions. All SNR values refer to a 3 kHz
(CER) calculated. BW. The graph depicts the modes’ behavior as they go from 0
In some modes it was possible to change the values of the to a 100 percent of CER as the SNR was decreased. PACTOR
baud rate and the frequency shift between tones. In these FEC and ASCII 110 were found to be the most sensitive
cases, baud rates were adjusted according to the standard modes. Meanwhile, PSK31 outperformed all the other modes,
conventions used for amateur radio operation; a 170-Hz shift being able to get a 1.69% of CER at a SNR of -12dB. This is
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 3
mainly because narrower bandwidth modes tend to be the best
performers at low SNRs.
For CCIR Good through Poor conditions, the condition
name of the channel can create confusion because it was found
that, as the conditions decayed, some modes performed better
(Fig. 3 to 5). PSK31 didn’t seem to be particularly affected by
the different channel conditions as it had similar CER among
tests. RTTY 45 was observed to perform almost similarly,
with low CER, as the SNR conditions got better, being the
only two modes that consistently got good results with
approximately two percent of CER on moderate SNR
When high latitude channel, quiet, and moderate conditions
were tested only PACTOR FEC and RTTY 45 had a CER Fig. 4 CCIR good conditions on character error rate (CER).
below four percent. Figs. 6 and 7 exhibit how the other modes
performed poorly even with a high SNR value. In disturbed
conditions only RTTY 45 demonstrated moderate
performance if the SNR was high; the remaining modes had a
100 % of CER.
PSK31 performed very well under noisy channel
conditions. On the other hand, it started showing complete loss
of signal when Doppler spread was increased, which is due to
the nature of its modulation scheme. RTTY 45 is the other
protocol that showed overall good results, but it was found to
be sensitive to the noise present in the channel.
Redundancy proved to be crucial when transmitting a
message. This ensures the message arrival at least most of the
time, especially when synchronization is lost for a brief Fig. 5 CCIR poor conditions on character error rate (CER).
Fig. 2 Character error rate (CER) of various digital modes as a function of
signal-to-noise ratio in AWGN conditions Fig. 6 High-latitude moderate conditions on character error rate (CER).
Fig. 3 CCIR good conditions on character error rate (CER).
Fig. 7 High-latitude disturbed conditions on character error rate (CER).
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 4
A recommendation for the use of a modern digital mode to
transmit textual information over the HF band was done. Tests
performed on an HF channel simulator were used to choose
the correct digital mode according to that which provided the
most reliable text transmission under different channel
Based on the simulation results and analyses, the RTTY 45
digital mode was selected as the recommendation for the
interstation communications system for a network of
geophysical measurement stations in the Antarctic. This
digital mode was chosen because of its good performance on
all tests, provided that there is a strong enough signal.
This paper was based on using MultiPSK for generating and
receiving the messages. Other tests could be done using actual
modems to compare the result with the simulations. Also,
different baud rates should be considered and tested to prove
the assumption that a slower baud rate will increase its
The author thanks R. Nealy, Dr. R. Clauer, Dr. J. Baker, and
Dr. H. Vo. The author also extends his gratitude to Dr. T.
Bose, director of the Wireless@VT REU program.
 Yang Guo; Ke Wang, “A Real-Time Software Simulator of Wideband
HF Propagation Channel,” ICCSN ’09 International Conference on
Communication Software and Networks, 2009, vol., no., pp.304–308,
27-28 Feb. 2009.
 S. Musko, C. R. Clauer, A. Ridley, and K. Arnett , “Autonomous low-
power instrument platform to enable remote high latitude array
deployment,” Rev. Sci Inst., Vol. 80, Issue 4, p. 044501, April 2009.
 Johnson, E.E., “Fast propagation predictions for HF network
simulations,” MILCOM 97 Proceedings, vol.1, no., pp.341–345, 2–5
 Riley, T.J., “A comparison of HF radio digital protocols,” Seventh
International Conference on HF Radio Systems and Techniques (Conf.
Publ. No. 441), vol., no., pp.206–210, 7–10 Jul 1997.
 Watterson, C.; Juroshek, J.; Bensema, W, “Experimental Confirmation
of an HF Channel Model,” IEEE Transactions on Communication
Technology, vol.18, no.6, pp.792–803, December 1970.
 CCIR Recommendation 520-1 19821, “Use of High Frequency
Ionospheric Channel Simulators,” ITU, Geneva C. J. Kaufman, Rocky
Mountain Research Lab, Boulder, CO, private communication, May
 ITU-R Rec. F1487. “Testing of HF Modems with Bandwidth of up to
about 12kHz Using Ionospheric Channel Simulator.” International
Telecommunication Union, Radiocommunication Sector. Geneva 2000.
 Lindecker, P. MultiPSK Software [Online]. Available:
 Wheatley,M.. PathSim Software [Online]. Available:
 Wheatley,M. (2000, December, 01). PathSim User and Technical Guide