Vote rigging on sdr.hu

G8JNJG8JNJ
in KiwiSDR Discussion
Hi All,

What's happening on sdr.hu ?

The votes for a few sdr's suddenly seem to have gone crazy.

There are a few sdr's that haven't been on line for too long and they have very quickly risen to the top of the list with huge numbers of votes. 

I could understand this if they had been publicised in some way, but they never seem to be that busy and the performance of some of them leaves a lot to be desired.

I suspect some 'bots' have been busy

Regards,

Martin - G8JNJ


«13

Comments

  • WA2ZKDWA2ZKD
    the voting has little relevance anymore

  • BjarneBjarne
    I think they should do away with the voting bit. Says me, who's number 168 on the list :-D What people need is to find one which can cover the geographical target one wants to listen to. With around 170 receivers online, it's increasingly difficult to find one. One remedy would be to sort the receivers by geography, like continents or even in some cases countries.

    bjarne
  • BjarneBjarne
    Actually I sent an email to Andras and suggested they be sorted geographically. If anyone here agrees, maybe more people should send a similar request.

    Bjarne
  • WA2ZKDWA2ZKD
    I agree on the geo thing, also a sub-sort on freq coverage
    Bjarne
  • G8JNJG8JNJ
    Agreed,

    I've also sent an email to Andras.

    Regards,

    Martin - G8JNJ
  • jksjks
    Guys, I can't speak for Andras obviously. But as a full-time university student he's even busier than I am. So don't expect too much for something like this.

  • danieldaniel
    List view just doesn't work for remote radio receivers, and neither the voting system fixes it. Personally, when picking a receiver, the most important thing to me is its location.

    For me, sdr.hu's map view has noise in form of narrowband and VHF receivers I'm not interested in, and lacks a day/night overlay that is very useful when planning band scans. I personally stopped using sdr.hu completely when rx.linkfanel.net came out.
  • BjarneBjarne
    Yes, but isn't rx.linkfanel.net the same as the map on sdr.hu?
  • danieldaniel
    Not exactly the same. It only has HF receivers that cover at least 5 MHz, and that includes all KiwiSDR's and two WebSDR's. It also has color coding based on availability status (free slots, all slots busy, temporary downtime). sdr.hu lacks the latter.
  • jksjks
    edited September 2017
    There is also the Pavlova dispatcher: https://github.com/priyom/pavlova

    With a little work you could setup your own custom dispatcher using only the Kiwis you like to use. And then define custom "area codes" (see https://github.com/priyom/pavlova/blob/master/receivers.js) that group receivers by region or even functionality (e.g. Kiwis with good VLF reception in EU)

    Pavlova will automatically find a Kiwi in the requested area code that is running with an available connection slot.

  • linkfanellinkfanel
    edited September 2017
    Hello Bjarne,

    I agree with your analysis that it would be useful to sort receivers by geography. However if you're aware of the existing maps, aren't you satisfied with the way they "sort" and present the receivers by geography? That would sound just obvious to me so I'm a bit confused o_O Maybe I'm missing something obvious from the point of view of some use cases. Are there reasons why you would prefer a linear listing sorted by geographic groups, over a map presentation? Something missing in them?
  • BjarneBjarne
    A valid point, linkfanel. But the list is what you will look at first, and if the voting system gives no added value for the user, why not change how the list is presented. But somebody's got to do it, and if Andras' main occupation is elsewhere, as John suggests, it may not be high on the list for him. Oh well. It may be the least of our worries.
  • BjarneBjarne
    A valid point, linkfanel. But the list is what you will look at first, and if the voting system gives no added value for the user, why not change how the list is presented. But somebody's got to do it, and if your main occupation is elsewhere, as John suggests, it may not be high on the list for Andras.
  • KQ6XAKQ6XA
    edited September 2017
    HF receiver users are probably interested in:

    1. Location of receiver? 
    2. Does it have an antenna capable of receiving the frequency range?
    3. Does the receiver have good SNR? 

    It takes a lot of trial and error to find a remote receiver that might meet the user's needs at any given moment.  

    Perhaps each receiver operator would like to rate their own receiver for:

    1. Receiver's coverage map or description of best reception regions for various frequency ranges. 
    2. Overall Signal-To-Noise 
    3. Best frequency ranges for reception 
    4. Level or frequency of RFI or local interference to reception

    Receive noise and interference is probably the #1 problem most HF users face. 
    Increasing RFI emissions from power supplies, processors, displays, mains lines, and powerline modems have become the dominant sources of background noise in many populated areas of the world. 
    Local HF/MF/LF noise interference was a background issue in the 20th century, but now in the 21st century, it has become the foreground.
  • G8JNJG8JNJ
    Hi All,

    Bonnie makes some good points.

    I wonder if it would be possible to use something like Peter's IBP beacon monitor (or similar) to provide some form of automated 'score' or figure of merit for each SDR based on the criteria Bonnie has already outlined ? This would provide a unified rating system and would limit the possibility of 'Vote Rigging' or folks 'bigging' up their own SDR - not that I would need to do that, as everyone already knows that my KiWi is the best  ;-)

    Regards,

    Martin - G8JNJ
  • SM2BYCSM2BYC
    Why not use the WSPR functionality of the KIWI SDR to score different receivers? Much more information can be obtained from WSPR compared to IBP.

    Regards,

    Mauritz / SM2BYC
  • G8JNJG8JNJ
    Hi Mauritz,

    Good idea - especially if WSPR could run automatically as a background service when the SDR isn't otherwise in use (as proposed some time ago).

    Regards,

    Martin - G8JNJ
  • SM2BYCSM2BYC
    Yes, should be sufficient to run e.g.1 hour at daytime on the higher bands (up to 20 meters) and 1 hour at night time on the lower bands (down to 160 meters).

    Regards

    Mauritz / SM2BYC
  • linkfanellinkfanel
    I like the WSPR idea, but I'm not familiar with WSPR so I'm not too sure about the practicalities. My idea was to connect either internally or externally, open a waterfall channel and capture the signal power data, averaging it over a few seconds. Then run some simple statistical distribution and arithmetic to calculate a rough signal-to-noise score. A basic example could be, taking the ratio between the median power level (background noise), and say the 95th percentile (estimate of power of received signals). This could be refined for more accurate, or multi-band results. I wonder if it would be worth starting to work on right away, with a simple external script to extract scores by hand, and incorporate and visualize them with different color shades of marker on my rx.linkfanel.net map.
  • SM2BYCSM2BYC
    There is a lot of information about WSPR on the "net" including WSPRNET.ORG

    The KIWISDR WSPR extension is also a good starting point.

    Regards

    Mauritz / SM2BYC
  • G8JNJG8JNJ
    Hi All,

    Thinking about his a bit further, I don't believe that using WSPR in its present form (requiring some form of initiation on each WEB SDR, rather than it running as a background task) will work particularly well.

    However I've noticed that generally speaking, the better SDR's tend to be in use a lot of the time. 

    So maybe the easiest way would simply to rank SDR's based on the total amount of time the four receivers on each SDR are in use over a rolling period of perhaps 7 days or similar ?

    This data is already gathered by sdr.hu, so it shouldn't be too difficult to collate and sort by rank.

    Regards,

    Martin - G8JNJ



  • SM2BYCSM2BYC
    Using receiver usage as a basis for rating will give some information on their performance. The drawback I see is that the results will easily be offset by user and/or administrator behaviour. I could connect to all 4 channels on my Kiwi SDR and leave them running for 24/7. That would guarantee me top ranking without even having to connect an antenna to the RX.

    I still think that WSPR based ranking is the best alternative.

    Perhaps the easiest way would be to have the central server (e.g. rx.linkfanel.net) connect to one RX at a time and run WSPR for some time. The results could then be extracted either directly from the RX or from WSPRNET.ORG. The server could then move on to the next RX. This way no action would be needed from the Rx owner and no modification would be required to the Kiwi SDR software.

    Regards

    Mauritz / SM2BYC
  • VE3SUNVE3SUN
    I like the idea of using signals to rank the capabilities of the receivers. Perhaps this could be done from the waterfall info. Currently I glance at the wide band view to see what frequencies signals are arriving at before bothering to use a receiver for a particular band. This kind of check could be automated or crowd sourced to a new receiver view.

    I have already created a scraped sortable list of receivers for my own use but am awaiting permission from Andras before sharing the link. It is sortable by Grid square, type of receiver, from and to frequency. it would be pretty easy to add more parameters that we could crowd source.

    I got the feeling looking at the voting system on sdr.hu that it might be screwed up by search bots. Knowing from my own web hosting that bots create a lot of noise on such buttons. It didn't seem very useful to me.




  • G8JNJG8JNJ
    Hi All,

    I see that in V1.121, John has added an updated set of Icons to sdr.hu indicating GPS lock, receiver performance and VLF coverage (and perhaps more).

    I think this is a step in the right direction, but I'm not sure how these are derived or what the criteria is for an icon being displayed (other than the GPS one), but I'm assuming it's not fully automated.

    One point to note. If a KiWi has an antenna switch to allow better coverage of different regions or the VLF bands (like mine) I'm not sure how this is handled.

    Regards,

    Martin - G8JNJ


  • jksjks
    edited September 2017
    As you correctly guessed, the GPS icon is auto-derived. The other two are selected based on a database entry on kiwisdr.com (i.e. they're not something the Kiwi admin can specify). This is just an experiment. I'm not sure it's a good idea. My criteria for the "superior performance" icon is entirely subjective.

    I've only evaluated a few Kiwis for the database and I'm already struggling to decide who makes the list. And as you point out Martin there are special cases where the scheme breaks down. Although having gone to the trouble of installing a switch with multiple antenna choices should probably qualify you based on effort alone, lol.
  • linkfanellinkfanel
    edited September 2017
    > Perhaps the easiest way would be to have the central server (e.g.
    rx.linkfanel.net) connect to one RX at a time and run WSPR for some
    time. The results could then be extracted either directly from the RX or
    from WSPRNET.ORG. The server could then move on to the next RX. This
    way no action would be needed from the Rx owner and no modification
    would be required to the Kiwi SDR software.

    I don't think that would be easiest. It's already troublesome enough to properly manage automatic probing of each receiver's status, and that's just one simple HTTP query. Running a full WSPR session for some time would be much worse in difficulty and complexity. Running probes from a central server to each receiver one by one doesn't scale up when the size of the network increases. If WSPR could be run as a background process when slots are available like some people have expressed they wished, and the results reported by the receiver itself along with other health information like current user count, I do think that would be valuable; but implementing this might be a long way from here.

    > Using receiver usage as a basis for rating will give some information on
    their performance. The drawback I see is that the results will easily
    be offset by user and/or administrator behaviour. I could connect to all
    4 channels on my Kiwi SDR and leave them running for 24/7. That would
    guarantee me top ranking without even having to connect an antenna to
    the RX.

    It could give some insights that would be right in some cases, but even without malicious manipulation, I'm quite skeptical it would be reliable. Some people do some day-long monitoring on some receivers, sometimes you can see the user has been connected and sitting on some obscure frequency for several days... Some people also like to use several slots to run WSPR, and although that can be useful, that's not a relevant indicator of popularity. And again, popularity is not quality; there's a reason why we have the expression "degenerate into a popularity contest" :)

    According to the changelog, the v1.121 icons are set from a central kiwisdr.com database, that I can only assume too is populated by hand by John. That's one option, I know first-hand it still demands significant manual review. But regardless of where the data comes from, I like this kind of vector to display relevant quality information in the sdr.hu listing, without requiring work from Andr?s.

    This afternoon I've written a quick proof of concept for my signal-to-noise idea. It's on test at rx.linkfanel.net Darker, greyer markers are receivers with a lower score; and brighter, redder ones have a better score. Bear in mind that the color scheme and score formula still need to be tweaked, and it was run manually once at one given time of the day, so it would still be important to integrate and average successive scoring samples over a period at least longer than a day-night cycle. For the details, the ratio is calculated from a few seconds of fully-zoomed-out waterfall data, using the median signal power as background noise value, and the 5% of most powerful signals across the whole spectrum as a signal value. It's also not (yet) a multi-band approach, so at the moment scores can be skewed by things like a very strong MW band.

    Just like I said about WSPR, I hope that this could be done as a periodic background service by the KiwiSDR board itself, instead of the small script I ran by hand.
  • G8JNJG8JNJ
    OK all interesting stuff and very thought provoking.......

    MY SDR will also be skewed by the fact that I notch out the strong broadcast bands in order to be able to run the rest of the frequencies at a higher level.

    Hmmm..... thinking about it you can also have too much of a good thing, so maybe we should also have an automatically generated warning icon for SDR's that exhibit frequent ADC overload :-)

    Regards,

    Martin - G8JNJ
  • linkfanellinkfanel
    > MY SDR will also be skewed by the fact that I notch out the strong
    broadcast bands in order to be able to run the rest of the frequencies
    at a higher level.

    Don't worry, Martin, the current crude score samples rank your receiver 13th out of 134 rated receivers :)
  • WA2ZKDWA2ZKD
    what bands/freqs do you notch?  I have to do that in the 550-1700 range on a few stations but still have pretty good coverage even there.


  • G8JNJG8JNJ
    Hi,

    >Don't worry, Martin, the current crude score samples rank your receiver 13th out of 134 rated receivers

    That's what alerted me to the problem, I had been watching my score gradually climb over the past six months to about 400 votes, until I was sitting at position number three. Then, almost overnight I was overtaken by some SDR's I'd not previously spotted with ridiculously large numbers of votes. Some had more than the original KiWi deployments, which would take some beating. 

    I'm not overly competitive (or maybe I am), but I like my SDR to be somewhere up near the top as I use it to showcase my antenna design.

    >what bands/freqs do you notch? 

    In Europe in addition to many LW and MW broadcast stations, we have very strong Short Wave transmissions, which are even stronger at night.

    So I'm currently notching the following by about 10dB

    LW 148-284KHz
    MW 525-1600KHz
    49 metres 5900-6200KHz
    41 metres 7200-7450KHz
    31 metres 9400-9900KHz
    25 metres 11600-12100KHz
    22 metres 13570-13870KHz
    19 metres 15100-15800KHz

    With recent propagation conditions 17MHz is also becoming a bit of a problem, so I may need to notch that too.
    16 metres 17480-17900KHz

    I also have some amplitude / frequency slope correction in place, which further reduces the signal levels on the lower frequencies in order to allow me to run the HF frequencies without having to add broadband attenuation ahead of the KiWi to prevent occasional ADC overloads. 

    However it doesn't seem necessary today as all the frequencies above 2MHz are completely dead :-(

    Regards,

    Martin - G8JNJ




Sign In or Register to comment.