LATVIAN JOURNAL OF PHYSICS AND TECHNICAL SCIENCES

2012, N 3
DOI: 10.2478/v10047-012-0015-3

METHODS FOR PROCESSING AND INTERPRETATION OF AIS SIGNALS

CORRUPTED BY NOISE AND PACKET COLLISIONS.

J. Poļevskis, M. Krastiņš, G. Korāts,

A. Skorodumovs, J. Trokšs
Ventspils International Radio Astronomy Center,
101 Inzenieru Str., Ventspils, LV-3601, LATVIA

The authors deal with the operation of Automatic Identification System (AIS)
used in the marine traffic monitoring to broadcast messages containing information
about the vessel: id, payload, size, speed, destination etc., meant primarily for
avoidance of ship collisions. To extend the radius of AIS operation, it is envisaged to
dispose its receivers on satellites. However, in space, due to a large coverage area,
interfering factors are especially pronounced − such as packet collision, Doppler’s
shift and noise impact on AIS message receiving, pre-processing and decoding. To
assess the quality of an AIS receiver’s operation, a test was carried out in which,
varying automatically frequency, amplitude, noise, and other parameters, the data on
the ability of the receiver’s ability to decode AIS signals are collected. In the work,
both hardware- and software-based AIS decoders were tested. As a result, quite
satisfactory statistics has been gathered − both on the common and the differing
features of such decoders when operating in space. To obtain reliable data on the
software-defined radio AIS receivers, further research is envisaged.
Key words: AIS, signal collisions, SDR, space-based.

1. INTRODUCTION
In the field of transport and logistics, cargo positioning and tracking plays an
important role. A dedicated system called the Automatic Identification System
(AIS) is used in the marine traffic monitoring to broadcast messages containing
information about the vessel: identification, payload, size, speed, destination, etc.,
used primarily for avoidance of ship collisions [1].
Each of the vessels broadcasts these messages with different frequency rate
depending on its current speed of movement. In the areas with increased traffic
density, packet collisions would appear. Such collisions are avoided using the Self-
Organized Time Division Multiple Access (SOTDMA) technology implemented
into the AIS [2]. This principle ensures self-organization of broadcast messages in
a local area, but when observed from a satellite, the collision of messages is
inevitable. This problem calls for further investigation into the packet collisions,
Doppler’s shift and noise impact on AIS message receiving, pre-processing and
decoding.
As the ground stations have a limited range and, therefore, not only are
unable to gather information about ships far from the coast but also require a large
number of such stations to cover the entire coastline, during the last decade there
has been a push to implement space-based AIS receivers.

25

Unauthenticated
Download Date | 2/4/20 3:59 PM
 AIS messages are transmitted in two channels – 161.975 MHz and
162.025 MHz, using 9.6 kbps GMSK modulation over 25 or 12.5 kHz channels.
Before modulation, the data are non-return-to-zero inverted (NRZI) encoded. The
SOTDMA scheme (employed for avoidance of collisions) uses the concept of one
minute long time frames consisting of 2250 message slots. Each potential
transmitter monitors the AIS channels for 1 min and chooses an available time slot
as nominal. From now on, whenever a ship has to transmit its AIS information (the
period depends on the ship's speed and whether it is changing course), it transmits
in a random time slot near the nominal one [2].
A satellite-based AIS receiver faces additional challenges when dealing not
only with a reduced signal-to-noise ratio, Doppler’s shift and environmental con-
ditions, but also with the possibility of packet collisions. As the TDMA scheme of
AIS signals works only on a local level while satellites have a large overlap area,
there is the possibility of collision when two ships are not in a direct line of sight of
each other and therefore transmit during the same time slot but both transmissions
can be received by a satellite. Collisions could also be caused by different path
lengths making the adjacent time slot messages overlap. Although such situations
are rare in open seas, in the areas with heavier maritime traffic (which are of the
greatest interest) the loss of data could be significant [3].
Currently, most AIS receivers tested in space are either commercial ter-
restrial hardware-based decoders or custom-made FPGA-based SDR solutions [4].
Collision problems are either ignored altogether or hopefully avoided by using
directional antennas with lower reception area, although the alternative for im-
proving the signal processing algorithms to recover higher percentage of collided
packets should be considered.
The goal of this study was to test how well AIS receivers decode signals that
are as close as possible to those received in space, and to compare the viability of
hardware- and software-based approaches.
The scope of this paper includes methods used in testing AIS receivers and
analysis of gathered data, as well as simple statistical comparison of AIS receivers
(analysis of different AIS receivers and methods for improvement of the decoding
performance are not included)..
Ideally, the method for testing AIS decoders would be to use pre-recorded
AIS signals received by satellites, and gather statistics on the number of messages
successfully decoded; however, as such signals are not readily available, they could
be simulated. The probability of decoding a simulated message is recorded along
with the parameters of simulation and the data gathered for analysis.
2. EXPERIMENTAL
A block diagram of the experimental setup can be seen in Fig. 1.
A signal based on series of one hundred faked AIS messages is generated on
USRP N210 SDR transceiver and through an attenuator sent to the testable AIS
decoder. Decoded messages are transmitted to PC, and their count as well as signal
generation parameters are logged. This process is repeated multiple times
automatically for various combinations of signal parameters.
Figure 2 shows a block diagram of AIS signal generation program.
A hundred numbered messages are encoded and modulated according to the
AIS standard. The carrier frequency is then shifted simulating Doppler’s shift and

26

Unauthenticated
Download Date | 2/4/20 3:59 PM
 Fig. 1. Block diagram of the experimental setup.

Fig. 2. Block diagram of AIS generation program.

Unauthenticated
Download Date | 2/4/20 3:59 PM
clock error. The signal is attenuated and summed with a secondary AIS signal
generated in the same way, which simulates a packet collision. Then the noise is
added and the signal is transmitted at 162.025 Mhz. The generator was developed
using GNU Radio (open source USRP-based SDR software development toolkit).
From this it follows that there are five basic parameters of simulation:
frequency shift and attenuation of primary signal, frequency shift and attenuation
of secondary signal, and amplitude of noise. All these parameters are automatically
iterated through in a regular grid and combination tested.
This is undoubtedly incomplete simulation of signal as seen in space, but it
is adequate for our goal − the comparison of decoders. Additionally, due to USRP
N210 (being designed for continuous transmissions as opposed to packet-based
ones) the generated AIS messages are transmitted without any pause between them,
which corresponds to the real world worst-case scenario.
The hardware-based AIS receivers under test were: DE70321 – a de-
monstration board of FPGA-based AIS baseband processor IC CMX7032 [5], and
Transas IPx7 − a commercial AIS class B transceiver.
For the software-based AIS decoder used for testing the USRP E100 was
employed to digitalize AIS data, while for a simple software-based AIS decoder −
the gr-ais (open source software released under the GPL licence, see N. Foster [6]).
The USRP N210 using WBX transceiver has an output power of 20dBm [7],
which was attenuated by 69dB in the case of hardware-based AIS decoders, and by
39dB in the case of an USRP e100 based decoder to account for the lack of a
narrow-band amplifier (to be added if it was to be dedicated to the AIS signal
decoding).
3. RESULTS AND DISCUSSION
The testing of hardware-based AIS receivers is considered successful. A
large body of data has been gathered, from which meaningful conclusions could be
drawn.

Fig. 3. The averaged decoding probability

as a function of primary signal frequency shift.

28

Unauthenticated
Download Date | 2/4/20 3:59 PM
Fig. 4. The averaged decoding probability as function of primary and secondary signal frequency shifts.

Unauthenticated
Download Date | 2/4/20 3:59 PM
 As the amount of data gained by iterating over all combinations of five
simulation parameters is huge, the impact of noise on the system (as of the time of
writing this publication) has not yet been fully processed, while preliminary results
show that it has a negligible qualitative impact on the overall relative performance
of AIS decoders.
The data gathered show that the probability of decoding a message sharply
drops when the secondary signal reaches 50% amplitude of the primary signal (see
Fig. 3).
Figure 4 shows the probability of receiving messages for any given pair of
primary and secondary signal frequency shifts averaged over all the amplitude
combinations.
The most interesting results seen in Fig. 4 are: first, there is a significant
increase in the probability of decoding if the frequency shift of a secondary signal
is 1 kHz. Our team has not been able to come to a conclusive answer as to the
cause of such a feature; second, the message collisions have the largest negative
effect on the probability of decoding the message if the primary and secondary AIS
signal frequencies are offset by about 2 kHz. For the LEO satellite this corresponds
to Doppler’s shift caused by a 30o angle or a 400 km ground distance between
transmitters.
The testing of software-based AIS decoder has so far been unsuccessful.
Although it adequately decoded real AIS signals, for generated signals this
occurred only sporadically, and even in these cases the decoded message was
corrupted by addition of several bytes. The causes of such behaviour are as of yet
unknown but being investigated.

4. CONCLUSIONS
The hardware-based AIS receivers have been shown to perform adequately
in a simulated space, although this highly depends on the AIS receiver itself. The
DE70321 decoding quality noticeably decreases with frequency shift, while the
Transas IPx7 is stable in this regard.
The testing of software-defined AIS receivers has as of yet been
unsuccessful. Although they have many promising qualities that could in theory be
better for space-based applications, in reality there are no readily available
solutions for them; therefore, to make them viable, they need to be developed
specifically for this kind of mission.
To analyze the data about the influence of noise in these systems as well as
to improve the overall testing model an additional research work is required.

REFERENCES
1. IALA, Operational issues. (Dec. 2004). In: IALA Guideline No. 1028 on The
Automatic Identification System (AIS), vol. 1 Part I, ed-n 1.3.
2. IALA, Technical issues. (Dec. 2002). In: IALA Guidelines on The Automatic
Identification System (AIS), vol. 1 Part II, ed-n 1.1.
3. Eriksen, T., Høye, G., Narheim, B. & Meland, B.J. (2006). Maritime traffic monitoring
using a space-based AIS receiver. Acta Astronautica, 58, 537−549.
DOI:10.1016/j.actaastro.2005.12.016

30

Unauthenticated
Download Date | 2/4/20 3:59 PM
4. Kgabo Frans Mathapo (2007). A Software-Defined Radio Implementation of Maritime
AIS. Unpublished master's thesis, University of Stellenbosch.
5. CML Microcircuits (2010). DE70321/DE70321T Demonstration Kit User Manual.
Retrieved from
http://www.cmlmicro.com/techregister/log.asp?fname=/products/datasheets/docs/de70
321ds.pdf
6. Foster, N. (2011). gr-ais [open source software]. Available from
https://www.cgran.org/wiki/AIS
7. Ettus Research (2011). Transceiver Daugherboards for the USRP Software Radio
System. Retrieved from
http://www.olifantasia.com/gnuradio/usrp/files/datasheets/ds_transceiver.pdf

METODES TROKŠŅU UN KOLĪZIJU REZULTĀTĀ SAKROPĻOTU

AIS ZIŅOJUMU APSTRĀDEI UN INTERPRETĀCIJAI
J. Poļevskis, M. Krastiņš, G. Korāts,
A. Skorodumovs, J. Trokšs
Kopsavilkums
AIS ir kuģu identifikācijas sistēma, kas starp kuģiem un zemes stacijām
pārraida datus par to identifikāciju, pozīciju, kursu, ātrumu utml. Lai palielinātu šīs
sistēmas darbības rādiusu, no zemes stacijām, tiek strādāts pie AIS uztvērēju
izvietošanas uz satelītiem. Kosmosa apstākļos ir daudz dažādi apstākļi, kas
uztveršanu apgrūtina, galvenais no tiem ir ziņojumu sadursmes plašā redzes lauka
dēļ.
Lai varētu novērtēt AIS uztvērēju darbības kvalitāti šajos apstākļos, tika
izveidots tests, kas automātiski, mainot frekvenču, amplitūdu un trokšņa para-
metrus, ievāc datus par AIS uztvērēja spējām tos dekodēt. Tika pārbaudīti gan uz
aparatūras bāzes AIS dekoderi, gan programmvadāmā radio bāzes AIS dekoderi.
Rezultātā tika ievākta ļoti laba statistika par aparatūras AIS dekoderiem, kas
parādīja gan kopīgās, gan atšķirīgās iezīmes par šādu dekoderu darbību kosmosā.
Derīgus datus par programmvadāma radio AIS uztvērēju neizdevās iegūt tā
programmas īpatnību dēļ.
Lai iegūtu labākus rezultātus, ir nepieciešams ieguldīt vēl daudz darba. Ir
nepieciešams ne tikai atrisināt problēmas ar programmvadāmo radio, bet arī līdz
galam apstrādāt neapstrādātos ievāktos datus par trokšņu ietekmi, kā arī izpētīt
iespējas šo testu vēl uzlabot.
04.03.2012

31

Unauthenticated
Download Date | 2/4/20 3:59 PM