Kiwi TDoA Direction Finding Increases SW Pirate Institutional Paranoia Levels

[Token]

How does this work when there are more than one station on a frequency?

Like, since hams have been mentioned here, could they use this to pinpoint the guys who play music and cuss a lot?

Because those frequencies have other transmissions going on simultaneously.

Does this feature only work when there is a sole signal on a frequency?

I have not taken the Kiwi TDOA code apart, so let me start by saying that what I am about to say is supposition, based on an understanding of the basic techniques and what I have observed from the system.  I could be out to lunch on some details, but I think I probably am correct on the major points.

I suspect anytime two separately transmitted sources of audio are in the same passband it will mess with the software, reducing the quality of the plots, but the below is a more sterile description, assuming worst case.

The system starts by taking simultaneous ~30 second samples from your selected SDRs.  You typically select SDRs that have the signal you are interested in as strong as possible.  You manually, one at a time, check the signal on each SDR to determine which SDR will be on your "list" of sampling machines.  Ideally you will have at least 3, and as many as 6, machines in your list.  Less than 3 will result in a curve, but not an intersection.

The software then runs all the samples you have taken to find identifiable features, something unique in the audio that the software can identify in all the samples.

After finding this (or these, I don't know how many it does per sample, but I assume it tries to find more than one) unique features it establishes the time of arrival at each SDR for the selected features.  It then plots the curves for each pair of receivers, calculates the intersection points and sigmas, and puts them on the map.

If the software attempts to find a single unique feature, and if there are two or more signals on the frequency, and you have selected receivers where the desired signal is strong and the second signal is weak, it is likely it will still find a unique feature for one of the stations.  In which case it will still plot only one transmitter.  Which one it plots is anyone's guess.

If the software attempts to find multiple features and time each of them, then it is less likely to plot only a single transmitter, but rather might indeed plot combinations of both transmitters, resulting in junk.

Regardless of how it does it though, if you had two widely separated stations on SSB simulcasting the same program (SSB so no destructive interference, AGC capture, or jamming, at worst a bit of an echo on the signal), with the same audio at the same time (not just two stations running the same tapes at the same time, but actual simulcast of the same audio stream) it would find corresponding features in each, and that would probably ruin the calculations.  The software might still plot something, but it would probably be junk, combinations of the curves for the two stations.  If each curve is nice and tight (good signal quality and good, tight, features to ID) it may calculate curves on the wrong time deltas, show curves and intersections in totally wrong areas.

Again, the above is based on some assumptions.

Could it be used to find the ham jammers?  Maybe, but remember what I said about the 30 second samples.  Transmissions from multiple stations, even if one at a time, that have handoff times inside that 30 second window will likely mess with the system.  So if the jammer does not stay on the air for the entire sample time that will probably result in errors, jumbled combinations of the data for the jammer and jammed stations.

T!

[KaySeeks]

If the software attempts to find multiple features and time each of them, then it is less likely to plot only a single transmitter, but rather might indeed plot combinations of both transmitters, resulting in junk.

In fact, I was wondering about exactly this so I attempted to "find" WWV on 5 MHz, 10 MHz and 15 MHz with receivers on the west coast of North America in the presence of WWVH (and passband narrowed to ~100-200 Hz so that differences in the male voice on WWV and the female voice on WWVH were reduced and the focus was on the carriers only). I don't have the results available right now but let's just say that things became "confused". It seems to me that as a general principle with TDoA, anytime there is a significant amount of skywave propagation present, the derived result varies quite a bit over time (see more on this at the bottom of this post). In this case, this was worse than I expected. I suspect (but was not able to prove beyond a doubt) that the presence of WWVH confused the algorithm.

Could it be used to find the ham jammers?  Maybe, but remember what I said about the 30 second samples.  Transmissions from multiple stations, even if one at a time, that have handoff times inside that 30 second window will likely mess with the system.  So if the jammer does not stay on the air for the entire sample time that will probably result in errors, jumbled combinations of the data for the jammer and jammed stations.

There has been some discussion about allowing for sampling intervals shorter than 30 seconds for reasons like this.

Getting back to what I mentioned before, skywave propagation is a big confounder here - it will cause a fair amount of variation from location attempt to location attempt. (This makes sense to me as the ionosphere will produce significant variable amplitude and phase shifts when the signal returns to earth.) Also, weak signals are more difficult to get a clean result on. Contrast this with strong, groundwave signals, which are relatively easy to locate. I was able to locate BBC - Droitwich (198 KHz) and RTL (234 KHz) to within ~10 km using only 3 receivers and I was able to repeat both using a second completely different group of receivers as a check. I did the same with WWV on 2500 KHz (only one set of three receivers though). However, 5, 10 and 15 MHz are not trivial and derived location was highly variable. I assume that this is due to the significant skywave propagation there.

All this means that HF pirates probably have little to worry about by the Kiwi TDoA network. You are quite unlikely to be successfully located down to your house. (I can't comment on what the authorities might be able to do though, so beware.) A MW or LW pirate might want to pay attention though.

[ChrisSmolinski]

All this means that HF pirates probably have little to worry about by the Kiwi TDoA network. You are quite unlikely to be successfully located down to your house. (I can't comment on what the authorities might be able to do though, so beware.) A MW or LW pirate might want to pay attention though.

Exactly. 50 or even 25 miles of accuracy (typical with skywave HF signals) is not going to uniquely identify anyone.  Some operators dislike the fact that now their general location (Chicago vs Boston for example) will be known, however. There's no changing that now. And to be honest, that's been possible previously without TDoA.

[Josh]

If the kiwis would employ vertical antennas, it would help the tdoa work with regards to skywave.

[KaySeeks]

50 or even 25 miles of accuracy (typical with skywave HF signals) is not going to uniquely identify anyone. 

You are doing well if you can repeatably locate an HF signal that isn't from some 10+ kW transmitter to within those distance ranges. I haven't been able to do that well generally. That Link11 signal on ~6942 KHz we have been talking about in that other thread is an exception and is a very strong signal.

Some operators dislike the fact that now their general location (Chicago vs Boston for example) will be known, however. There's no changing that now. And to be honest, that's been possible previously without TDoA.

Right. For example, ways of doing this include: 1) looking at propagation characteristics and also 2) sniffing around with the many receivers on the internet.

[KaySeeks]

If the kiwis would employ vertical antennas, it would help the tdoa work with regards to skywave.

If the pirates employed more NVIS, it would make it more difficult too.

[Brian]

All this means that HF pirates probably have little to worry about by the Kiwi TDoA network. You are quite unlikely to be successfully located down to your house. (I can't comment on what the authorities might be able to do though, so beware.) A MW or LW pirate might want to pay attention though.

Exactly. 50 or even 25 miles of accuracy (typical with skywave HF signals) is not going to uniquely identify anyone.  Some operators dislike the fact that now their general location (Chicago vs Boston for example) will be known, however. There's no changing that now. And to be honest, that's been possible previously without TDoA.

The monitoring stations here in Europe land are pretty good at DFing HF signals. They have me down to about 10 or 12 miles and another station possibly even closer than that. That's within groundwave, if they really wanted to nail me, they could.

They don't always get it right though. They think another station is VERY close to my location. I'ts actually about 80/90 miles away.

[Token]

Getting back to what I mentioned before, skywave propagation is a big confounder here - it will cause a fair amount of variation from location attempt to location attempt. (This makes sense to me as the ionosphere will produce significant variable amplitude and phase shifts when the signal returns to earth.) Also, weak signals are more difficult to get a clean result on. Contrast this with strong, groundwave signals, which are relatively easy to locate. I was able to locate BBC - Droitwich (198 KHz) and RTL (234 KHz) to within ~10 km using only 3 receivers and I was able to repeat both using a second completely different group of receivers as a check. I did the same with WWV on 2500 KHz (only one set of three receivers though). However, 5, 10 and 15 MHz are not trivial and derived location was highly variable. I assume that this is due to the significant skywave propagation there.

All this means that HF pirates probably have little to worry about by the Kiwi TDoA network. You are quite unlikely to be successfully located down to your house. (I can't comment on what the authorities might be able to do though, so beware.) A MW or LW pirate might want to pay attention though.

Those really accurate plots of signals in Europe tend to be with all the RX locations in groundwave, and working with very good SNR.  On the other hand, I have plotted WWV to within 9 km on 15 MHz, using 3 RXs located more than 1000 miles away.

Skywave itself is not a problem to TDOA.  It is variations in the number of hops and altitude of reflective layers to each of the receive stations.

Basically if all of the receive stations are roughly the same distance from the target then the increased distance (and so time) caused by skywave vs direct path will be approximately the same for all the receive stations.  In other words, despite the fact each receive location will get the signal a little later than ground wave, they will all have roughly the same skywave induced delta T, and it will be factored out in the calculations.

The inaccuracies start to get larger as you have receive stations of widely varying distances from the target.  A worst case would be say one receiver in ground wave and another in double or triple hop.  This would skew the resultant curve between those two stations by the skywave delta T and away from the ground wave station.  Even if the third receive station is in double or triple hop skywave only one curve will be accurate, the curve between the two skywave receive stations.  So two of the three resultant curves (A vs B, A vs C, and B vs C) will have significant errors, and the intersection of all three curves (the plotted position) will be shifted a good bit.

I have modeled plots with one RX in groundwave and the other two in high double hop skywave, and moved the resulting mapped TX location by greater than 200 km.  Just because it is a nice, tight, plot does not make it right

So, good signal strength (good SNR) and all receive locations evenly distributed around the target and roughly the same distance from the target can yield very good results, skywave or not.  This is particularly true if all the receive locations are in say daylight, or all are in night, vs some in daylight and some in night.

T!

[John Poet]

While some of us have been posting DF maps of broadcast and utility stations, I expect that all users understand that it would be best to refrain from posting any maps of pirate stations.

Why not make that a RULE for this forum, Chris?  NO POSTING DF MAPS OF PIRATE STATIONS! (or other location information, for that matter).
There are certainly enough moderators to make it stick and quickly remove any such thing...

It would probably make a lot of the ops feel better.  What's worse than individual listeners being able to do this DFing, is if that information would become published or widely disseminated.  This would surely result in a reduction in stations and broadcasts...

[ChrisSmolinski]

While some of us have been posting DF maps of broadcast and utility stations, I expect that all users understand that it would be best to refrain from posting any maps of pirate stations.

Why not make that a RULE for this forum, Chris?  NO POSTING DF MAPS OF PIRATE STATIONS! (or other location information, for that matter).
There are certainly enough moderators to make it stick and quickly remove any such thing...

It would probably make a lot of the ops feel better.  What's worse than individual listeners being able to do this DFing, is if that information would become published or widely disseminated.  This would surely result in a reduction in stations and broadcasts...

We can make it even slightly more broad than that, no DOXing of anyone, operator or listener. This would include TDoA maps, real names, locations, etc. Both posting the information here, as well as links to such information on other sites.   Reasonable?

[TheRelayStation]

While some of us have been posting DF maps of broadcast and utility stations, I expect that all users understand that it would be best to refrain from posting any maps of pirate stations.

Why not make that a RULE for this forum, Chris?  NO POSTING DF MAPS OF PIRATE STATIONS! (or other location information, for that matter).
There are certainly enough moderators to make it stick and quickly remove any such thing...

It would probably make a lot of the ops feel better.  What's worse than individual listeners being able to do this DFing, is if that information would become published or widely disseminated.  This would surely result in a reduction in stations and broadcasts...

We can make it even slightly more broad than that, no DOXing of anyone, operator or listener. This would include TDoA maps, real names, locations, etc. Both posting the information here, as well as links to such information on other sites.   Reasonable?

i agree and yes, that is very reasonable.
please apply this rule to the chat as well as the forum.
i would have these warnings posted as a sticky for all forum topics as well as for new members who register, they would have to read it and accept the terms before they can continue.
all moderators should be notified of the changes and immediately remove any content that reveals the location and other personal info of an OP or Hobbyist.
repeat offenders should be banned depending on how many times they infringe the rules (my opinion).
this is an excellent forum and hobby and although i do not mind that my location may be known, its better that everyone keeps that info to themselves since there is really nothing i can do to avert it from becoming known anyway (experienced individuals could easily obtain location info before TDoA on Kiwi, TDoA just made it easier for the average person).
hopefully, those who do know sensitive information about OP's do not spread it around using other means such as social media.
just about every HFU member is a respectable adult that knows better, so i dont expect any problems to surface as a result of TDoA on Kiwi though it is good practice to cover those bases.

[Oliver]

I fully agree with the statements of JP,   Chris & The Relay Station.
Those should be in my opinion the rules of the game for every board member.
It would be counterproductive to our hobby, to use new technologys and spreading the gained information with the possibility to harm any of the operators outhere who make this hobby so enjoyable.

One last thought, would it be possible to come to a similar agreement with the moderators of Iann's Chat?
I assume that some of you have a thigther relationship with some of the moderators on chat to discuss this topic.

Thanks,
Oliver

[John Poet]

While some of us have been posting DF maps of broadcast and utility stations, I expect that all users understand that it would be best to refrain from posting any maps of pirate stations.

Why not make that a RULE for this forum, Chris?  NO POSTING DF MAPS OF PIRATE STATIONS! (or other location information, for that matter).
There are certainly enough moderators to make it stick and quickly remove any such thing...

It would probably make a lot of the ops feel better.  What's worse than individual listeners being able to do this DFing, is if that information would become published or widely disseminated.  This would surely result in a reduction in stations and broadcasts...

We can make it even slightly more broad than that, no DOXing of anyone, operator or listener. This would include TDoA maps, real names, locations, etc. Both posting the information here, as well as links to such information on other sites.   Reasonable?

YES!   Absolutely.

[Josh]

If the kiwis would employ vertical antennas, it would help the tdoa work with regards to skywave.

If the pirates employed more NVIS, it would make it more difficult too.

Yeah, nvis done right will result in a df fix straight up.

[ChrisSmolinski]

Yeah, nvis done right will result in a df fix straight up.

Not necessarily with TdoA. The height of the ionosphere can be taken into account.