We seek to understand and document all radio transmissions, legal and otherwise, as part of the radio listening hobby. We do not encourage any radio operations contrary to regulations. Always consult with the appropriate authorities if you have questions concerning what is permissable in your locale.

Author Topic: How Amalgamated DGPS works  (Read 110 times)

Offline ChrisSmolinski

  • Administrator
  • Marconi Class DXer
  • *****
  • Posts: 21336
    • View Profile
    • Black Cat Systems
How Amalgamated DGPS works
« on: January 03, 2019, 1315 UTC »
Something I just posted to a mailing list, perhaps interesting to folks here:

ADGPS is very aggressive in trying to find a valid DGPS message, that is one that has the correct preamble, passes the checksum, and the station ID, frequency, and baud rate match the list of stations. After demodulating the audio and passing it through some very tight filters, it shift the audio samples in one by one, looking for a valid message (which is why it can spin the CPU fans on your computer at a high speed). Considering the number of audio samples per bit, this means a given message, if it exists, is sampled a few dozen times. Of course random noise is also sampled many more times, and something that appears to be a valid DGPS message can appear out of the noise. If you're processing an I/Q file of the entire DGPS band, it is looking about one million times per second for a valid message. Over a 12 hour overnight recording session, over 40 billion times, if my math is correct.

Unfortunately the checksum algorithm DGPS uses is not very robust, and multiple bit errors can produce a message that still passes the checksum, there is nothing that can be done about this. My guess as to why other DGPS decoders don't necessarily produce decodes when this happens is because they are looking for (relatively) strong signals only. Of course if you do that, you can miss the weak DX targets we're looking for :)

This is why a few random invalid DGPS decodes are generated each ADGPS session.

Some human intervention is required to determine if decode(s) are valid. Everyone needs to come up with their own set of "rules" to decide what is real and what is not. However this is a fun hobby, while you do not want to fool yourself with amazing DX catches that are not real, it's also not a life threatening situation, it doesn't matter if you occasionally guess wrong. These are the rules I happen to use:

Are the decodes due to another station with the same ID/baud/frequency?  There's a few cases of this, where two stations send the same ID on the same frequency and baud rate. There's no way ADGPS can tell them apart. Fortunately one is usually on the other side of the planet, and can be ignored.

Often, false decodes are actually corrupted decodes of another DGPS station. If you look at the frequency and bad rate of a suspect falsely detected station, you can usually match it up with another station much closer to you, that you regularly receive. But static/noise has corrupted the ID as well as some of the other bits, so it still passes the checksum test. Sometimes you can compare the ID of the two stations and see how just a bit or two being changed causes this (remember the ID is sent as a binary number). There's a dozen or so stations I routinely get false decodes from. The usual suspects as I call them. It's quite easy for me to ignore them, I know they are not real.

Is the catch even possible based on what we know about propagation?  If I see decodes from South Africa or China during the local daytime here in the USA, I know that is not possible, and they can be discarded.

How many decodes are there, and are there at least a few decodes closely spaced in time? If I get 5 decodes from a DX target widely spaced over the entire night, say an hour or more apart, I am not necessarily convinced I received it. If OTOH those 5 decodes are received within a minute or two, then I am much more convinced it was one of those brief DX openings, and the decodes were legitimate.

Is it a station I regularly receive? For example, I routinely get Horta Azores during the winter. I'm not surprised when I see it, and maybe I will check the timestamps to see how closely spaced they are, but maybe I won't. If I see 25 decodes, like I did this past night, I won't even bother to check the timestamps, I am sure it was legitimate.

Are you getting a cluster of decodes from several stations in the same region around the same time, indicating a DX opening?  For example, reception of European stations for me in the USA is somewhat rare, but occurs often in the winter. The usual pattern is that I should be seeing Horta Azores, it is on a clear frequency of 308 kHz. Often I will get it and nothing else from Europe and that region. If I see decodes from some European station, but only a few, and nothing from Azores, then I am suspicious, and suspect they could be fake, I'd look at the timestamps to decide. Chances are, they are indeed false decodes.   In my case, the next most likely station is Cabo Cavoeiro Lt from Portugal, as it is on a split channel, and also is fairly common for me, I got 3 decodes last night.  Then maybe if those two are coming in, I also lucked out, and got some more European stations. Last night I didn't, but I few nights ago I did. And last winter I had a few nights when I got a lot from Europe, even the UK, Ireland and Germany. Very rare here but it happens.

Another example for me are Pacific NW stations. There's a few that are common. I tend to use Richmond BC Canada as my "beacon". The more decodes I got from it, the more likely I am going to see decodes from other stations in BC and even Alaska sometimes.

Also going back to propagation - I find that it is most likely (but not always the case) to get decodes from a DX target near the "end" of the propagation window, that is when it is an hour or two before local sunrise for European stations, and an hour or even less from my local sunrise for west coast USA/Canada stations, especially Alaska and even more so Hawaii. Hawaii is very rare for me, I think I have gotten decodes once or twice. It's a long haul.  I believe this is the case because the ionosphere has been dark the longest time, and is most suitable for low frequency propagation.

Of course you also want to consider overall propagation conditions. If the ionosphere is unsettled with a high K index, DGPS results will likely be poor. I also like to use the DST index as a guide, it should be relatively stable and high for a few days. If it is for a week or two, then you can get some of the amazing DX openings.  Some useful tools are here https://www.hfunderground.com/propagation/#longwave

ADGPS is another tool in your DX ToolBox. (Hey, happen to have an app called DX ToolBox :) https://www.blackcatsystems.com/software/ham-shortwave-radio-propagation-software.html )  ADGPS can be very useful, but is not infallible. But with some analysis of the results, you can filter out the invalid decodes fairly easily, and be left with what are likely real DX catches.
« Last Edit: January 11, 2019, 1247 UTC by ChrisSmolinski »
Chris Smolinski
Westminster, MD
eQSLs appreciated! csmolinski@blackcatsystems.com
NRD 545 / netSDR / AFE822x / AirSpy HF+ / KiwiSDR / 670 ft horizontal loop / 500 ft northeast beverage / 270 ft west-south-west beverage / 300 ft south beverage / 43m / 20m / 10m  dipoles / Crossed Parallel Loop