VHF-transmission identification challenge

[StuM]

Just to add to the confusion, vessels talking to "the pilot station or coast guard" are likely to be using a ship to shore duplex channel through a repeater, so the received signal will be coming from the repeater station, not the vessel.

[CatNewBee]

I guess, vessels talking to officials use the protocol carefully and say they vessel name, potentially call sign, to distinguisch between vessels with the same name etc. and comply to the rules of communication. They will also tell their GPS position, either directly or upon request.

I dont think there are too many reckless pranks out there, that use the VHF to trick the officials, but this happens...

[barnakiel]

All dsc calls from a dsc enabled vhf unit will contain a dsc header:

continuousWave: Whaler: Reference: Data Interface in Digital Selective Calling Class-D Radios

Plain voice transmissions from a dsc vhf radio contain no dsc data, hence no way to identify the caller.

I use dsc exclusively, never plain voice, when mmsi of the boat is know to us (this implies the boat either has an active ais unit onboard or else the boat has initiated our comm with a dsc call.

Cheers,
b.

[hellosailor]

Cat, i didn't say there ere no regulations. I said there was no mandated LINK or other means to connect a vessel to a transmission. Whether the radio operator happens to follow regulations and tradition, or not, has no bearing on that. There is no LINK that makes this happen automatically and inevitably.

You certainly could do speech processing on the VHF audio and then pattern recognition to extract a vessel name and then look it up in a database...if you had the resources of the NSA or a large corporation. And as Nicholas has apparently started to do, by transcribing [sic] the VHF audio into a text file. But I can go out and buy a ten year old copy of DragonDictate for $25 and do that. Speech recognition/transcription is an old thing by now, the trick is getting the error rate up. (I find that even Google, Microsoft, Apple, or Amazon's products are sometimes entirely unable to understand very clear speech, and other times, incredibly
good at it.)
But even then, there's no assurance that information will be presented intrinsically in any transmission.

Nicholas-
In answer to your question of "Or does DSC not get engaged in any way when using the regular VHF-communication?" AFAIK it does not. I'm not intimate with the DSC specifications but from what I have heard about the USCG's "new" communications systems for distress calls, they DO NOT get any information "with" the voice call. The voice call stands alone, DSC is not used by default.
You would need to query your national authority (here it would be the FCC or the USCG) to find out if there were any national regulations, or query the ITU to confirm international standards that should apply. In my limited knowledge, there is no DSC broadcast of the MMSI or anything else made invisibly when the voice transmission on normal channels is made.
Very few marine users actually conform to suggested standards, except perhaps the military. With commercial mariners, simply getting them to maintain a proper watch, or to use proper navigation lights, does not always happen.
If you set up a DSC radio receiver alongside a voice channel VHF radio, you can find out for yourself if there is DSC traffic being sent along with every voice call.

[CatNewBee]

DSC is only transmitted on channel 70 and only if you use the DSC call function, all other channel transmit voice only. No DSC at all. DSC datagramms are re-transmitted automatically after a time period for distress calls until turned off or acknowledged by DSC by the MRCC afaik.

[RaymondR]

Two land based receiving stations signal processing for phase shift in the AIS signal.

Same stations processing for similar or same phase shift in the voice transmission signal.

If proportion of phase shift is the same in both cases then either, most likely, the signals are sourced from transmitters at the same geographical location or, less likely, different geographical locations with a transmitter/base station geometry which would provide similar phase shifts.

[CatNewBee]

Originally Posted by RaymondR

Two land based receiving stations signal processing for phase shift in the AIS signal.

Same stations processing for similar or same phase shift in the voice transmission signal.

If proportion of phase shift is the same in both cases then either, most likely, the signals are sourced from transmitters at the same geographical location or, less likely, different geographical locations with a transmitter/base station geometry which would provide similar phase shifts.

That implies, that a rogue, pirate vhf handheld prank, A is on the water and B has turned on transmitting AIS on board. Very unlikely.

The good guys do not need to be identified, thei say their name and position. And the others are very unlikely keen to show off with DSC traffic or AIS. They even may not use ITA certified gear for transmitting / receiving...

[Wotname]

Originally Posted by CatNewBee

That implies, that a rogue, pirate vhf handheld prank, A is on the water and B has turned on transmitting AIS on board. Very unlikely.

The good guys do not need to be identified, thei say their name and position. And the others are very unlikely keen to show off with DSC traffic or AIS. They even may not use ITA certified gear for transmitting / receiving...

I think the OP's project needs to identify the good guys, not the rogue pirates.

[RaymondR]

Originally Posted by Wotname

I think the OP's project needs to identify the good guys, not the rogue pirates.

Yep, my reading was that he only wanted to connect the voice transmissions to a specific AIS ship identifier.

If looking for a common timing signal for the base stations in order to detect the phase shift I think the GPS system broadcasts a highly accurate clock signal as part of their data packet broadcasts.

[Paul Elliott]

Originally Posted by RaymondR

Yep, my reading was that he only wanted to connect the voice transmissions to a specific AIS ship identifier.

If looking for a common timing signal for the base stations in order to detect the phase shift I think the GPS system broadcasts a highly accurate clock signal as part of their data packet broadcasts.

Difference of phase at different receiving sites is pretty difficult to use at VHF frequencies, since the wavelength is less than two meters, and for any typical station separation there will be thousands of conflicting potential solutions to the hyperbolic equation.

Time of Arrival (TOA) can be used, and yes, GPS can provide timing accuracy to nanoseconds, which is probably better than the variations in propagation delay of the various paths. This needs a signal with good characteristics for automating the detection though. AIS is good, VHF voice not as good, but I suppose possible. You need communications among multiple receiving stations to be able to calculate the differences in TOA -- the math is essentially the reverse of LORAN, using hyperbolic lines of constant difference.

You can use directional antennas at multiple locations to locate transmitters, and these days a doppler turnstile antenna array is usually used, which can simultaneously calculate bearings on multiple sources. The USCG uses this technique in their "Rescue 21" system.

But the OP mentioned that their coastline isn't conducive to using multiple radio-location stations, so perhaps this isn't too useful.

[RaymondR]

I was not suggesting trying to define a position using the techniques exploited by Decca as the problem is to connect a voice transmission to an AIS transmission which provides a position anyway.

Assume station A and station B each of which "listens" for both the AIS and voice signals and uses the GPS time signal to time stamp the zero crossing of the voice and AIS transmission signals.

Add or deduct the zero crossing time for the voice signal to arrive at a single lead or lag value (L or L) for the subject voice signal and each of the AIS signals.

Compare the subject voice L or L value with each of the values from the various AIS signals.

The most similar L or L values are probably being transmitted from the same geographical locations which can be defined from the AIS signal.

I have not done any maths to check the feasibility of the scheme and have no intention of doing so, the poster can do that. However I think it's a possibility worth checking out.

[NicholasM]

Originally Posted by RaymondR

Assume station A and station B each of which "listens" for both the AIS and voice signals and uses the GPS time signal to time stamp the zero crossing of the voice and AIS transmission signals.

Add or deduct the zero crossing time for the voice signal to arrive at a single lead or lag value (L or L) for the subject voice signal and each of the AIS signals.

Compare the subject voice L or L value with each of the values from the various AIS signals.

The most similar L or L values are probably being transmitted from the same geographical locations which can be defined from the AIS signal.

Thank you for the answer Raymond!

That is a very interesting point of view, that I haven't considered before. I am currently looking into the technical and financial feasibility of this scheme.

A couple of things I can immediately come up with:

Infrastructure-wise, we only have one receiving station at the moment. Getting access to another one for testing purposes should not be too difficult. I do not have any experience with zero crossing detection, I assume the stations would have to be equipped with some kind of detector? Are there any precedents of zero cross detection with VHF or AIS signals?

An issue that I can spot out straight away is the fact that idle vessels (whose status is 'anchored') only send out their AIS signal once every 3 minutes. In busy waters this could make the matching process very tricky. Moving vessels (faster than 3 kts) send out the signal every 3-10 seconds. This means the solution could not be standalone, but will have to be complemented with another one.

Hence, if this theory proves feasible, I am considering doing the following:

- Use a daily 'expected vessel' list and filter vessel names out of the conversation if available and clear enough;
- Complement with zero cross detection if the ship doesn't announce its name or if it isn't pronounced clear enough;

For larger vessels requiring pilots to board the ship we are also looking into voice detection to identify the pilot (after a pilot boards the vessel, they take over all communication with the shoreline) and link directly to the 'pilot planning' to identify the vessel.

Is anyone else familiar with the principles of zero cross detection and its reliability? Is the scheme reliable when there are multiple vessels close to each other?

[Paul Elliott]

Zero-crossing detection on a VHF signal is going to be extremely difficult. Which zero crossing are you looking at? The first one will be indistinguishable from broadband noise and your detectors will be putting out full-amplitude noise when there is no signal.

If you mean looking at the zero-crossings of the demodulated signal, that's a bit easier, but again, which zero-crossing? There will be thousands of them per second, and telling one from another will be tough.

A technique I would consider is cross-correlation of the two decoded audio signals. But to do this you need access to time-stamped copies of the decoded audio so you can correlate one with the other. But to use this method for location, you need three receiving sites -- two sites just give you a hyperbolic line of position. Also, the audio filtering done in the receivers will add some delay uncertainty, perhaps on the order of microseconds, and each microsecond gives you 300 meters of receiver-position uncertainty (and much more uncertainty with the hyperbolic line). This filtering delay can probably be calibrated out by regular monitoring of transmitters of a known location.

With two sites you can use radio direction finding to give you two bearings that intersect at one position (using the doppler turnstile antenna method I previously mentioned). The accuracy is apparently +/- a few degrees, which gives significant position uncertainty at longer ranges or tight acute angles.

Look at TDOA (Time Difference of Arrival) and Radiolocation to see how others are doing this: https://cdn.rohde-schwarz.com/pws/dl...OA_english.pdf

[Unregistered-1]

The KiwiSDR remote receivers will group together and do TDOA in a semi-automated fashion. Script up something that would "lock" a very approximate VHF real-time TDOA (on actual microphone PTTs) to a much more accurate "MMSI over AIS" inside your database, and then "track" it with both detects. Once you had made that link in your database you could make quite a few assumptions from it.

Rephrasing ;

Use an AIS receiver, and enter every station in range into your MySQL database.

Use the KiwiSDR stations to RDF track any VHF transmissions, and find close matches in your AIS database and make the database link/assumption. There will not be too many vessels very close together, and if there are then they will not stay that way for long. If they do stay tight together then they are not underway and likely you aren't interested in them at that time.

Use KiwiSDR stations (these are GPS-locked) to measure VHF transmissions precise and exact transmit frequency, down to the single Hz.

Now you can match RDF targets and single station transmissions with MMSI numbers.

You will need a downconverter for the KiwiSDR stations, and its' local oscillator will need to be GPS locked.

The KiwiSDR stations are not cheap, but they are 30MHz wide so you can do a sizable chunk of the VHF spectrum in real time.

If you can make this work it will be an amazing toy and a useful and powerful tool that you can resell to a LOT of people.

[Unregistered-1]

Originally Posted by Wotname

RF surveillance experts tell me that every transmitter has it's own unique "fingerprint" for a the first few milliseconds after it is keyed. This is just a product of the transmitter coming up to its carrier frequency.

You would need some seriously expensive spectrum analysers and recorders to make use of this apparent fact.

It's easy to see with a $12 RTL SDR USB receiver