Long Range AIS Channels

VHF Marine Band radios, protocol, radio communication theory, practical advice; AIS; DSC; MMSI; EPIRB.
jimh
Posts: 5712
Joined: Fri Oct 09, 2015 12:25 pm
Location: Michigan, Lower Peninsula
Contact:

Long Range AIS Channels

Postby jimh » Tue Oct 18, 2016 6:05 pm

An interesting outcome of the 2012 World Radio Conference (WRC 12) that affected VHF Marine Band frequencies was a provision to permit use of Channels 75 (156.775-MHz) and 76 (156.825-MHz). These two channels have generally never been put into use except to serve as guard channels for Channel 16. They functioned as unused frequencies that were adjacent to Channel 16 (156.800-MHz) in order to keep that channel free from any adjacent channel interference. In WRC 12, a provision was made to permit Channels 75 and 76 to be used for a long-range AIS broadcast message, Message 27.

According to the relevant ITU recommendation, ITU-R M.1371-5, AIS Message 27 is:

...primarily intended for long-range detection of AIS Class A and Class B “SO” equipped vessels (typically by satellite). This message has a similar content to Messages 1, 2 and 3, but the total number of bits has been compressed to allow for increased propagation delays associated with long-range detection.


This provision appear to endorse and facilitate use of AIS transmission for global tracking of vessels by satellites in space. I find this quite interesting. AIS was introduced as a collision-avoidance system, but it has since become more useful as a vessel tracking system. The introduction of private and government satellites in space orbit for the purpose of receiving AIS signals in order to track vessels far at sea and out of range of any land-based receiver has become a new aspect of AIS. Apparently now or in the future there is a specifically allocated additional frequency which can be employed specifically to facilitate receivers listening from space.

For more see the ITU document at

Recommendation ITU-R M.1371-5
Technical characteristics for an automatic
identification system using time division
multiple access in the VHF maritime
mobile frequency band

https://www.itu.int/dms_pubrec/itu-r/rec/m/R-REC-M.1371-5-201402-I!!PDF-E.pdf

jimh
Posts: 5712
Joined: Fri Oct 09, 2015 12:25 pm
Location: Michigan, Lower Peninsula
Contact:

Re: Long Range AIS Channels

Postby jimh » Sun Nov 12, 2017 1:19 pm

On the general topic of reception of AIS signals by satellites in various earth orbits, it is unknown to me if the original design of the AIS system took any consideration of satellite reception, but it seems a reasonable assumption that, at some point, the incorporation of AIS MESSAGE 27 into the protocol must have been done with satellite reception in mind.

There are more problems with reception of AIS signals by a satellite than one might consider at first glance. The AIS signals were initially designed for a range of communication of perhaps 20 to 50-miles between ships. Satellites in Low Earth Orbit (LEO) typically have an altitude of more than 100-miles and up to 1,200-miles. A typical AIS LEO satellite might be in a orbit with a elevation of about 400-miles. As a result of this very high point of view, the satellite can see many more AIS transmitting ships than any terrestrial receiver. Since the coordination of transmission of AIS signals is in small groups of self-organized networks, the self-organized networks only cover ships that are within mutual receive-transmit communication. A satellite AIS receiver may receive signals from many ships, all participating in different self-organized networks, and these transmission are likely to overlap, using the same time slots for a transmission. The satellite receiver then is faced with trying to receive two or more signals at once, all in a single time slot.

SRT Marine, a leading manufacturer in AIS transceiver development and manufacturing, reported that a satellite AIS receiver may only successfully decode about 2-percent of the AIS transmissions it receives. The other 98-percent are unreadable due to interference from competing signals.

To overcome the poor receiver performance, several very complex methods have been proposed and perhaps one or more are already in use. One method to discriminate overlapping signals is to exploit the effects of Doppler Shift. Because the satellite is in relative motion to all the transmitters, the apparent frequency of the signal from a particular ship arriving at the satellite receiver will be shifted from the original frequency, based on its relative motion to the satellite's. The satellite receiver can re-tune its receiver slightly up or down in frequency to better receive a particular ship's transmissions.

Another method that can help a satellite separate AIS signals is use of a directional antenna. Electrically-steerable beam-forming antennas can be used to improve the signal strength from certain directions, allowing perhaps better reception of the AIS signals from ships in the antenna's main lobe while discriminating against interfering signals from other ships that arrive from a different heading,

Even more sophisticated methods have been proposed for improving the yield of successfully decoded AIS signals from overlapping transmissions. In one method, as soon as the first (usually strongest) signal is decoded successfully, the receiver then uses the decoded data to re-encode a new transmission, mimicking the original, and this regenerated signal is then injected into the original mix of radio-frequency signals being analyzed, but with a phase reversal. The stronger and now isolated signal is thus subtracted from the jumble of received signals, removing it from the mix. Now the next strongest signal can often be detected and decoded. If this is possible, then that second signal is then regenerated and subtracted from the incoming signals, perhaps yielding another layer of detectable and decodable signals.

Signals that are detected but when decoded produce an error may also be eventually successfully decoded by application of very complex mathematical analysis. With intensive computational processing, it may be possible to extract an error-free message.

And one manufacturer of AIS transceivers, SRT Marine, has devised their own trade-secret method of transmission that they say will significantly improved detection and decoding by satellite receivers, particularly by their own satellite receivers. Their method is not disclosed, but they claim a ten-fold improvement of successful message decoding, to 20-percent of all received messages up from 2-percent.

jimh
Posts: 5712
Joined: Fri Oct 09, 2015 12:25 pm
Location: Michigan, Lower Peninsula
Contact:

Re: Long Range AIS Channels

Postby jimh » Sun Nov 12, 2017 1:26 pm

One of the proposed methods of enhancing reception by satellite of AIS signals is described in a U.S. Patent. The patent description offers a reasonably clear explanation of the method. See

https://patents.google.com/patent/US20140113546A1/en

Here are three illustrations and accompanying text that explain them, excerpted from the patent application.

Figure1.png
Figure1.png (8.56 KiB) Viewed 730 times


    DETAILED DESCRIPTION OF EMBODIMENTS

    FIG. 1 schematically represents a system 10 for detecting broadcast signals transmitted by terrestrial sources 40.

    The following description will be based on the nonlimiting case in which the terrestrial sources 40 are ships, and in which the broadcast signals transmitted by said ships are AIS signals.

    Nothing precludes, according to other examples, from considering other types of terrestrial sources and broadcast signals, for example ADS-B signals transmitted by aircraft. Advantageously, the detection system 10 can be implemented to detect several types of broadcast signals, for example both AIS signals and ADS-B signals.

    The detection system 10 comprises a satellite 20 in earth orbit. It should be noted that any suitable orbit can be considered, for example a low earth orbit LEO, a medium earth orbit MEO, etc.

    The detection system 10 illustrated by FIG. 1 comprises a single satellite 20. Nothing precludes, according to other examples, from having a detection system 10 comprising a plurality of such satellites 20 in earth orbit.

    The satellite 20 comprises an antenna array 22 comprising a plurality of individual antennas 24 adapted to receive the AIS signals transmitted by the ships 40. Hereinafter in the description, M will be used to designate the number of individual antennas 24.

    In practice, several AIS signals are likely to be received simultaneously by the different individual antennas 24 of the antenna array 22. Thus, each individual antenna 24 of the antenna array 22 supplies an individual signal which can prove to be a composite signal combining, inter alia, several AIS signals transmitted by different ships 40 and received during one and the same acquisition interval.

    The satellite 20 also comprises conventional means adapted to transmit all or part of the individual signals, obtained respectively from the different individual antennas 24, to a terrestrial processing device 30, which comprises conventional means adapted to receive said individual signals transmitted by the satellite 20.

    The processing device 30 comprises means configured to detect the AIS signals in the individual signals received from the satellite 20, obtained respectively from the different individual antennas 24, in accordance with a detection method 50 described in more detail hereinbelow.

    The processing device 30 to this end comprises a unit for processing the individual signals which for example takes the form of a processor and an electronic memory in which is stored a computer program product, in the form of a set of program code instructions which, when executed by the processor, implement all or part of the steps of the method 50 for detecting AIS signals. In a variant, the processing unit comprises programmable logic circuits, of FPGA, PLD, etc. type, and/or custom integrated circuits (ASIC), adapted to implement all or part of the steps of said method 50 for detecting AIS signals.

There are several important aspects of the system described here that should be noted. First, the satellite has several antennas, and the signals of each antenna are preserved and stored for later signal analysis. Second, the system is a bent-pipe system, that is, the heavy computational analysis is not done in the satellite, but by a terrestrial station that receives stored signals from the individual antennas and then performs the method's proposed signal analysis.

Figure2.png
Figure2.png (29.35 KiB) Viewed 725 times


FIG. 2 represents the main steps of a detection method 50 according to a particular embodiment.

As illustrated by FIG. 2, a detection method 50 according to the invention comprises at least two detection iterations:

--a first detection iteration 51 a, and
--a second detection iteration 51 b.

The main steps of each of said first 51 a and second 51 b detection iterations are as follows:

--52: formation, from the individual signals obtained from the individual antennas 24 of the satellite 20, of virtual beams of different respective main directions of radiation,
--53: search for the presence of AIS signals in the virtual beams.

Thus, in its general principle, a detection method 50 is based on the formation of virtual beams.

A beam corresponds to a particular radiation pattern of the antenna array 22, and is formed conventionally by combining the individual signals by using a particular set of weighting coefficients. Thus, it is possible to form, from the individual signals, beams of different respective main directions of radiation by using different sets of weighting coefficients.

The expression “virtual beams” should be understood to mean that said beams are formed in a deferred manner (unlike a forming in real time). The individual signals are consequently first of all stored, and can then be used to form as many virtual beams as necessary.

In the particular embodiment illustrated by FIG. 2, the nonlimiting case is assumed in which one and the same number N of virtual beams are formed both during the first detection iteration 51 a and during the second detection iteration 51 b. Nothing precludes, according to other examples, from forming different numbers of virtual beams during the first 51 a and second 51 b detection iterations.

During the virtual beamforming step 52 of the first detection iteration 51 a, the virtual beams are predefined virtual beams, of main directions of radiation that are evenly distributed in a radiofrequency field of view of the antenna array 22.

The expression “predefined virtual beams” should be understood to mean that the weighting coefficients considered for each virtual beam are static and known in advance. Such a forming of virtual beams is known in the literature as “conventional beamforming”.

During the virtual beamforming step 52 of the second detection iteration 51 b, the virtual beams can, according to the embodiment considered, be predefined virtual beams (conventional beamforming) and/or adaptive virtual beams.

The expression “adaptive virtual beams” should be understood to mean that the weighting coefficients considered for each virtual beam are dynamic and determined as a function of parameters that are not known in advance, for example as a function of the AIS signals detected during the first detection iteration 51 a. Such virtual beamforming is known in the literature as “adaptive beamforming”.

The step 53 of searching for AIS signals is executed for each of the N virtual beams formed during each of said first 51 a and second 51 b detection iterations. Said search for AIS signals can be performed conventionally.

Furthermore, for at least one AIS signal detected during the first detection iteration 51 a, the respective contributions of said at least one detected AIS signal to the different individual signals are attenuated, relative to said first detection iteration, for all or part of the virtual beams formed during the second detection iteration 51 b.

Preferentially, the respective contributions of a detected AIS signal to the different individual signals are attenuated in all the virtual beams formed during the second detection iteration 51 b in one and/or the other of the following ways:
by estimating and by suppressing said respective contributions in the different individual signals, the virtual beams being formed during the second detection iteration 51 b from individual signals obtained after suppression, and/or
by forming, during the second detection iteration 51 b, adaptive virtual beams determined as a function of said detected AIS signal such that the respective contributions of said detected AIS signal are combined destructively in said adaptive virtual beams of the second detection iteration 51 b.

In the particular embodiment illustrated by FIG. 2, the nonlimiting case is considered in which the respective contributions of at least one detected AIS signal to the M individual signals are estimated during a step 54, and suppressed in said individual signals during a step 55, before the formation of the virtual beams during the second detection iteration 51 b.

Preferentially, such an estimation/suppression of the respective contributions of a detected AIS signal to the individual signals is performed for each detected AIS signal during the first iteration.

For example, the step 54 of estimating the respective contributions of a detected AIS signal comprises steps of:
--540 estimating a demodulated signal from the AIS signal detected in a virtual beam,
---541 estimating a direction of arrival and an amplitude of arrival, on the antenna array 22, of said detected AIS signal,
---542 computing the M respective contributions of said detected AIS signal to the M individual signals, as a function of the demodulated signal, of the direction of arrival and of the amplitude of arrival of the detected AIS signal.

The estimated amplitude of arrival is preferably a complex amplitude, that is to say one that encompasses information on a phase of arrival, on the antenna array 22, of said detected AIS signal.

The step 540 of estimating the demodulated signal from the detected AIS signal can be performed conventionally, by demodulating said detected AIS signal according to the predefined modulation format of the binary data included in the AIS signals. The demodulated signal is then an estimation of the binary data included in said detected AIS signal. As is known, said binary data included in an AIS signal comprise, in principle, a Cyclic Redundancy Check (CRC) field making it possible to detect errors in the binary data received. Advantageously, the CRC is used to confirm that the detected signal is indeed an AIS signal, the detected signal being considered to be an AIS signal only if the decoding of the CRC indicates that there are no errors in the binary data received.

It should be noted that it is also possible, when an AIS signal has been detected in a virtual beam, to form, still during the first detection iteration 51 a, a new virtual beam with a main direction of radiation identical to the direction of arrival of said detected AIS signal. Thus, the estimation of the demodulated signal will be improved because the antenna array 22 gain will be at maximum in said direction of arrival of said detected AIS signal.

The step 541 of estimating the direction of arrival and the amplitude of arrival of the detected AIS signal can implement methods considered to be within the scope of the man skilled in the art.

For example, the direction of arrival can be estimated by means of the MUSIC (Multiple Signal Classifier) algorithm or by means of the ESPRIT (Estimation of Signal Parameters via Rotational Invariant Techniques) algorithm, both based on the computation of a covariance matrix from the M individual signals.

In a preferred variant, the direction of arrival of the detected AIS signal on the antenna array 22 is estimated as a function of information, included in said detected AIS signal, on the position of the ship 40 having transmitted said detected AIS signal. In practice, it is known that an AIS signal includes the GPS coordinates of the ship that transmitted it, such that the direction of arrival of this AIS signal on the antenna array 22 of the satellite 20 can be estimated provided that the position and the attitude of the satellite 20 are also known.

For example, the amplitude of arrival of the detected AIS signal is estimated in the virtual beam formed during the first detection iteration 51 a for which the main direction of radiation is closest to the estimated direction of arrival of said detected AIS signal. The amplitude of arrival of the detected AIS signal in the M individual signals can then be estimated from:
--the amplitude of arrival of said detected AIS signal in the virtual beam considered,
--the complex gain of the antenna array 22, for the virtual beam considered, in the direction of arrival of said detected AIS signal.

Then, the respective contributions of the detected AIS signal can be regenerated, during the step 542, from the demodulated signal, the direction of arrival and the amplitude of arrival of said detected AIS signal.

The M respective contributions of said detected AIS signal to the M individual signals are then known, and are subtracted from said individual signals during the suppression step 55.

By thus suppressing, in the individual signals, the contributions of all or part of the detected AIS signals, the detection of new AIS signals, previously masked by the detected AIS signals, will be facilitated during the second detection iteration 51 b.

Preferably, and as illustrated by FIG. 2, the detection method 50 comprises, during the step 54 of searching for AIS signals in the virtual beams formed, a step 543 of determining whether one and the same AIS signal has been detected in several virtual beams. If one and the same AIS signal is detected in several virtual beams, the respective contributions of said detected AIS signal in several virtual beams are then regenerated and suppressed only once in the individual signals.

The determination whether one and the same AIS signal has been detected in several virtual beams can implement methods considered to be within the scope of a person skilled in the art. For example, it is possible to compare the demodulated signals obtained in different virtual beams. In practice, demodulated signals obtained in different virtual beams, if they correspond to one and the same AIS signal, should theoretically contain identical binary data.

In particular embodiments, the detection method 50 also comprises, for each detected AIS signal for which the respective contributions to the individual signals are to be estimated, the estimation of a frequency of arrival and/or of an instant of arrival of said detected AIS signal on the antenna array 22. The respective contributions of said detected AIS signal to the different individual signals are regenerated as a function also of said frequency of arrival and/or of said instant of arrival.

Such provisions make it possible to improve the accuracy of the regeneration of the contributions of the detected AIS signals, because said contributions will be realigned in frequency (by taking into account, inter alia, any frequency shifts induced by Doppler effect) and in time with all of the individual signals.

The estimation of the frequency of arrival and/or of the instant of arrival can implement frequency and/or time synchronization methods that are considered to be within the scope of the man skilled in the art, and is preferably performed in the virtual beam that led to the detection of the AIS signal considered. Such an estimation can be performed during the step 53 of searching for AIS signals and/or during the step 540 of estimating the demodulated signal.

In the particular embodiment illustrated by FIG. 2, the virtual beams, during the second detection iteration 51 b, are formed from the M individual signals obtained after suppression of the respective contributions of AIS signals detected during the first detection iteration 51 a.

In such a case, the virtual beams formed during the second detection iteration 51b can be predefined virtual beams (conventional beamforming) and/or adaptive virtual beams (adaptive beamforming).

In a preferred embodiment, the virtual beams formed during the second detection iteration 51 b are predefined virtual beams of different main directions of radiation from the main directions of radiation of the virtual beams formed during the first detection iteration 51 a.

For example, by considering a radiofrequency field of view of the antenna array 22 extending between −30° and 30°, and by considering a number N of virtual beams formed equal to six, it is possible to form:
--during the first detection iteration 51 a: virtual beams of respective main directions of radiation −30°, −20°, −10°, 0°, 10° and 20°,
--during the second detection iteration 51 b: virtual beams of respective main directions of radiation −25°, −15°, −5°, 5°, 15° and 25°.

As a variant or in addition, it is possible to form, during the second detection iteration 51 b, adaptive virtual beams determined as a function of the AIS signals detected during the first detection iteration 51 a.

According to a nonlimiting example, the adaptive virtual beams formed during the second detection iteration 51 b are orthogonal to the respective directions of arrival of at least a part of the AIS signals detected during the first detection iteration.

The expression “orthogonal to the direction of arrival of an AIS signal” should be understood to mean that the radiation pattern of the antenna array 22 for the virtual beam considered exhibits, in said direction of arrival, a local minimum. Said local minimum is such that, in said direction of arrival, the virtual beam exhibit an antenna array 22 gain at least 20 decibels (dB) less than the antenna array 22 gain in the main direction of radiation of said virtual beam, even at least 30 dB less.

It should be noted that the number M of individual antennas 24 limits the number of directions of arrival to which a virtual beam can be orthogonal. In theory, it is possible to form (M−1) blind directions in a virtual beam, a blind direction being a direction in which no signal can be received (known as “null steering” in the literature). It will therefore be understood that it will not necessarily be possible to form a virtual beam orthogonal to the directions of arrival of all the detected AIS signals, if there are too many thereof.

In such a case, preference will be given to forming virtual beams orthogonal to the directions of arrival of the detected AIS signals for which the measured powers were the highest and/or of the AIS signals detected in several virtual beams of the first detection iteration 51 a. In a variant, it is possible to form virtual beams orthogonal to the respective main directions of radiation of at least a part of the virtual beams in which several AIS signals have been detected during the first detection iteration 51 a.

It should be noted that the detection method 50 can comprise more than two detection iterations. In the nonlimiting example illustrated by FIG. 2, the second detection iteration 51 b then advantageously comprises, like the first detection iteration 51 a, steps 54 of estimating and 55 of suppressing the contributions of the AIS signals detected during the second detection iteration 51 b. The individual signals thus obtained are used for the forming of virtual beams during a third detection iteration, etc.

Hereinafter in the description, a particular embodiment of the antenna array 22 is considered, in which each individual antenna is designed to measure the signals received according to two distinct linear polarizations, preferably orthogonal. Thus, it is possible to form, from the individual signals, virtual beams of different respective main directions of radiation and/or of linear polarizations by using different sets of weighting coefficients.

In such a case, virtual beams of different respective linear polarizations are preferably formed during the first detection iteration 51 a and/or during the second detection iteration 51 b, for example evenly distributed in a radiofrequency field of view of the antenna array 22.

The AIS signals are transmitted with a vertical linear polarization. When propagating, an AIS signal will retain a substantially linear polarization, but the latter will turn. Consequently, the AIS signals are all received with a substantially linear polarization, but with an orientation that will be able to vary from one received AIS signal to another. By forming virtual beams of different linear polarizations, it will consequently be possible to distinguish AIS signals received with different linear polarizations.

In a preferred embodiment, different linear polarizations are considered from one detection iteration to a next detection iteration.

For example, the virtual beams formed during the second detection iteration 51 b are predefined virtual beams of different linear polarizations from the linear polarizations of the virtual beams formed during the first detection iteration 51 a.

As a variant or in addition, when adaptive virtual beams are formed during the second detection iteration 51 b, the latter are, for example, orthogonal to the respective linear polarizations of arrival of at least a part of the AIS signals detected during the first detection iteration.

The expression “orthogonal to the linear polarization of arrival of an AIS signal” should be understood to mean that the linear polarization of the virtual beam considered is orthogonal to the linear polarization of arrival of said AIS signal.

If the number of AIS signals detected is too great, it will be possible to form virtual beams orthogonal to the linear polarizations of arrival of the detected AIS signals for which the measured powers were highest and/or orthogonal to the linear polarizations of arrival of the AIS signals detected in several virtual beams of the first detection iteration 51 a. In a variant, it is possible to form virtual beams orthogonal to the respective linear polarizations of at least a part of the virtual beams in which several AIS signals have been detected during the first detection iteration 51 a.

In the case illustrated by FIG. 2 where the respective contributions of at least one detected AIS signal to the M individual signals are estimated during a step 54, and suppressed in said individual signals during a step 55, the M respective contributions of said detected AIS signal to the M individual signals are preferably also computed as a function of the linear polarization of arrival of said detected AIS signal, for example estimated during the estimation step 541.

Figure3.png
Figure3.png (17.22 KiB) Viewed 725 times


FIG. 3 represents curves, obtained by simulation, illustrating AIS signal detection efficiency of several detection methods. More particularly, FIG. 3 represents the average number of AIS signals detected as a function of the total number of AIS signals present in individual signals, in the case of an antenna array 22 comprising five individual antennas 24 spaced apart by one wavelength, each of said individual antennas 24 having double linear polarization (horizontal and vertical).

Three curves are represented, illustrating the results obtained for three different detection methods: one detection method according to the prior art and two detection methods according to the invention.

For comparison purposes, the same total number of virtual beams formed was considered for the three detection methods, in order to obtain a substantially equivalent computation complexity. In the case in point, a total of 72 virtual beams were formed to obtain each of the curves represented in FIG. 3.

In the detection method according to the prior art, twelve main directions of radiation were considered (from −30° to 25° in 5° steps) and, for each main direction of radiation, six different linear polarizations (from 0° to 150° in 30° steps) were also considered. The results obtained correspond to the curve designated “AA” in FIG. 3.

The detection methods 50 according to the invention, considered to obtain the results represented in FIG. 3, both comprise four detection iterations and are both based on the estimation/suppression of the AIS signals detected in the individual signals, the individual signals obtained after suppression being considered for the forming of virtual beams during the next detection iteration.

In the first detection method 50 according to the invention considered (for which the results obtained correspond to the curve designated “INV1”), the same virtual beams are formed during the four detection iterations. More particularly, six main directions of radiation (from −30° to 20° in 10° steps) and, for each main direction of radiation, three different linear polarizations (0°, 60°, 120°) were considered.

In the second detection method 50 according to the invention considered (for which the results obtained correspond to the curve designated “INV2”), different virtual beams are formed during the four detection iterations. For each detection iteration, six main directions of radiation were considered and, for each main direction of radiation, three different linear polarizations were considered. More particularly, the following were considered:
--during the first detection iteration: the main directions of radiation from −30° to 20° in 10° steps and, for each main direction of radiation, the linear polarizations 0°, 60° and 120°,
--during the second detection iteration: the main directions of radiation from −25° to 25° in 10° steps and, for each main direction of radiation, the linear polarizations 0°, 60° and 120°,
--during the third detection iteration: the main directions of radiation from −30° to 20° in 10° steps and, for each main direction of radiation, the linear polarizations 30°, 90° and 150°,
--during the fourth detection iteration: the main directions of radiation from −25° to 25° in 10° steps and, for each main direction of radiation, the linear polarizations 30°, 90° and 150°.