Probability theory - spot yield for hopping versus continuous reception?


Bruce KX4AZ
 

I wonder if anyone has ever done any rough calculations using the WSPR database to do some predictive calculations around the spot numbers that could be achieved in a continuous decoding (i.e. with a KiwiSDR) versus hopping (i.e. a WSJT-x software with a single frequency receiver).

My current  'KX4AZ/T' KiwiSDR site is working reasonably well with WSPR decoding, but I don't like to tie up all 8 of the Kiwi channels with the wsprdaemon.  Recently I have had it decoding on just 4 channels (80/40/30/20).  Now and then I'll turn on a separate Airspy HF+ Discovery  "hopper" receiver for the less active WSPR frequencies...but I've always wondered about how many spots I am missing with the hopper receiver.  Seems like the WSPR database would be an excellent tool to use for constructing some predictive calculations, but I can't quite wrap my mind around the mathematics & probability calculations it would require.  And for all I know this has already been done long ago.


WA2TP - Tom
 

Have you considered upgrading the beaglebone on the kiwi to a BBAI? 
You would then be able to decode 14rx channels simultaneously. Albeit you do loose the full waterfall. 

I currently have 54rx channels by doing this with 5 kiwi/BBAI and using wsprdaemon to decode.

On Dec 5, 2021, at 1:28 PM, Bruce KX4AZ <bruce@...> wrote:

I wonder if anyone has ever done any rough calculations using the WSPR database to do some predictive calculations around the spot numbers that could be achieved in a continuous decoding (i.e. with a KiwiSDR) versus hopping (i.e. a WSJT-x software with a single frequency receiver).

My current  'KX4AZ/T' KiwiSDR site is working reasonably well with WSPR decoding, but I don't like to tie up all 8 of the Kiwi channels with the wsprdaemon.  Recently I have had it decoding on just 4 channels (80/40/30/20).  Now and then I'll turn on a separate Airspy HF+ Discovery  "hopper" receiver for the less active WSPR frequencies...but I've always wondered about how many spots I am missing with the hopper receiver.  Seems like the WSPR database would be an excellent tool to use for constructing some predictive calculations, but I can't quite wrap my mind around the mathematics & probability calculations it would require.  And for all I know this has already been done long ago.


Jim Lill
 

A couple thoughts....  

- look at raw data from wsprnet.org or equiv. site and see what your reception is over the course of a day

- once understanding that distribution, you might be able to reduce the number of kiwi channels you use by taking advantage of the scheduling function


On 12/5/21 1:28 PM, Bruce KX4AZ wrote:
I wonder if anyone has ever done any rough calculations using the WSPR database to do some predictive calculations around the spot numbers that could be achieved in a continuous decoding (i.e. with a KiwiSDR) versus hopping (i.e. a WSJT-x software with a single frequency receiver).

My current  'KX4AZ/T' KiwiSDR site is working reasonably well with WSPR decoding, but I don't like to tie up all 8 of the Kiwi channels with the wsprdaemon.  Recently I have had it decoding on just 4 channels (80/40/30/20).  Now and then I'll turn on a separate Airspy HF+ Discovery  "hopper" receiver for the less active WSPR frequencies...but I've always wondered about how many spots I am missing with the hopper receiver.  Seems like the WSPR database would be an excellent tool to use for constructing some predictive calculations, but I can't quite wrap my mind around the mathematics & probability calculations it would require.  And for all I know this has already been done long ago.


Glenn Elmore
 


I can't answer for 'anyone' but the short answer for myself is "no, I haven't". But I think it depends what you are after. To maximize the raw spot count with limited receiver count probably following the herd as they travel will get you more total spots. If the goal is unique spots over a day, I think following the MUF may pay off better.

By watching unique spots vs band and time of day, I try to center the 6 receivers I use on WSPR at N6GN/K2, 3/4ths of a KiwiSDR,  on where the MUF /absorption has center of mass. at a particular time of day and year.  I change the schedule in wsprdaemon several times/day.

I'm more interested in the propagation than the counts but doing this does give more data so that watching it all in conjunction with the MUF map can give insight.


On 12/5/21 11:28 AM, Bruce KX4AZ wrote:
I wonder if anyone has ever done any rough calculations using the WSPR database to do some predictive calculations around the spot numbers that could be achieved in a continuous decoding (i.e. with a KiwiSDR) versus hopping (i.e. a WSJT-x software with a single frequency receiver).

My current  'KX4AZ/T' KiwiSDR site is working reasonably well with WSPR decoding, but I don't like to tie up all 8 of the Kiwi channels with the wsprdaemon.  Recently I have had it decoding on just 4 channels (80/40/30/20).  Now and then I'll turn on a separate Airspy HF+ Discovery  "hopper" receiver for the less active WSPR frequencies...but I've always wondered about how many spots I am missing with the hopper receiver.  Seems like the WSPR database would be an excellent tool to use for constructing some predictive calculations, but I can't quite wrap my mind around the mathematics & probability calculations it would require.  And for all I know this has already been done long ago.


Rob Robinett
 

I set all my Kiwis to 8 channel mode and configure WD to use 6 or 7 of the channels, leaving one waterfall rx channel free for my listening.

For publically shared sites, with two Kiwis fed by one antenna I can spot on all 14 WSPR bands and I have 4 waterfall channels for listeners.

On Sun, Dec 5, 2021 at 10:40 AM Glenn Elmore <n6gn@...> wrote:


I can't answer for 'anyone' but the short answer for myself is "no, I haven't". But I think it depends what you are after. To maximize the raw spot count with limited receiver count probably following the herd as they travel will get you more total spots. If the goal is unique spots over a day, I think following the MUF may pay off better.

By watching unique spots vs band and time of day, I try to center the 6 receivers I use on WSPR at N6GN/K2, 3/4ths of a KiwiSDR,  on where the MUF /absorption has center of mass. at a particular time of day and year.  I change the schedule in wsprdaemon several times/day.

I'm more interested in the propagation than the counts but doing this does give more data so that watching it all in conjunction with the MUF map can give insight.


On 12/5/21 11:28 AM, Bruce KX4AZ wrote:
I wonder if anyone has ever done any rough calculations using the WSPR database to do some predictive calculations around the spot numbers that could be achieved in a continuous decoding (i.e. with a KiwiSDR) versus hopping (i.e. a WSJT-x software with a single frequency receiver).

My current  'KX4AZ/T' KiwiSDR site is working reasonably well with WSPR decoding, but I don't like to tie up all 8 of the Kiwi channels with the wsprdaemon.  Recently I have had it decoding on just 4 channels (80/40/30/20).  Now and then I'll turn on a separate Airspy HF+ Discovery  "hopper" receiver for the less active WSPR frequencies...but I've always wondered about how many spots I am missing with the hopper receiver.  Seems like the WSPR database would be an excellent tool to use for constructing some predictive calculations, but I can't quite wrap my mind around the mathematics & probability calculations it would require.  And for all I know this has already been done long ago.



--
Rob Robinett
AI6VN
mobile: +1 650 218 8896


WA2TP - Tom
 

Interestingly I have been using the 5 kiwi/BBAI setup for just over a year and a half and I generally stand in the top 5 for uniques and top 10 for overall.

Cumulative last month November total spot count was recorded as #3 in the world per KB9AMG’s site.

And daily top 5 in the world on the wspr challenge.
 73’
WA2TP
Tom

On Dec 5, 2021, at 1:40 PM, Glenn Elmore <n6gn@...> wrote:




I can't answer for 'anyone' but the short answer for myself is "no, I haven't". But I think it depends what you are after. To maximize the raw spot count with limited receiver count probably following the herd as they travel will get you more total spots. If the goal is unique spots over a day, I think following the MUF may pay off better.

By watching unique spots vs band and time of day, I try to center the 6 receivers I use on WSPR at N6GN/K2, 3/4ths of a KiwiSDR,  on where the MUF /absorption has center of mass. at a particular time of day and year.  I change the schedule in wsprdaemon several times/day.

I'm more interested in the propagation than the counts but doing this does give more data so that watching it all in conjunction with the MUF map can give insight.


On 12/5/21 11:28 AM, Bruce KX4AZ wrote:
I wonder if anyone has ever done any rough calculations using the WSPR database to do some predictive calculations around the spot numbers that could be achieved in a continuous decoding (i.e. with a KiwiSDR) versus hopping (i.e. a WSJT-x software with a single frequency receiver).

My current  'KX4AZ/T' KiwiSDR site is working reasonably well with WSPR decoding, but I don't like to tie up all 8 of the Kiwi channels with the wsprdaemon.  Recently I have had it decoding on just 4 channels (80/40/30/20).  Now and then I'll turn on a separate Airspy HF+ Discovery  "hopper" receiver for the less active WSPR frequencies...but I've always wondered about how many spots I am missing with the hopper receiver.  Seems like the WSPR database would be an excellent tool to use for constructing some predictive calculations, but I can't quite wrap my mind around the mathematics & probability calculations it would require.  And for all I know this has already been done long ago.


Bruce KX4AZ
 

Setting up a BBAI would definitely be the best solution for me if I don't eventually add a second Kiwi, but I have the impression that would require a fair amount of effort software-wise that exceeds my current skill set -  i.e. I had just enough to get wsprdaemon running on a Pi4.  Perhaps I should do is make it my summer 2022 project, when I return to the KX4AZ/T site where the KiwiSDR is running.

Generally speaking, my goal has been to maximize the unique station counts in a 24 hour period, without tying up the KiwiSDR too much.  When I occasionally run a supplementary hopper receiver, I adjust the bands based on the time of day, to focus on what makes the most sense, i.e. at night I don't bother with 15/12/10 meters, but 630/160m are worth including.  

This thread actually belongs in the wsprbeacon forum, it occurs to me.  My interest/curiosity is in devising a mathematical algorithm to predict how many unique stations might be missed on a specific band, if a receiver is only listening for x% of the time over the course of a 24 hour period.  Essentially it becomes a cumulative probability of spotting, taking into account the the proportion of the day that the beacons are transmitting, combined with the probability that the receiver will "see" a signal of adequate S/N ratio.  My sense is that a 20% duty cycle on a single band would yield much more than 20% of the total in comparison to a continuous (100%) receiving mode.  The question is what that % value might be.  And actually, this value could be extracted from the WSPR data base if one could interrogate it in the correct way, and see how many unique spots remain if 80% of them were randomly removed from entire set of raw spots in a day.  So I should be able to answer my original question once I improve my WSPR database skills. 


WA2TP - Tom
 

It’s fairly simple. I am by no means any kind of debian expert. 

In fact, my very first experience with debian was a year and a half ago. 

Since then I have built well over a dozen kiwi/BBAI systems for fellow hams. 

The most difficult part was a reliable thermal solution which I was able design.

Feel free to reach out if I can be of any  assistance. 

On Dec 5, 2021, at 4:14 PM, Bruce KX4AZ <bruce@...> wrote:

Setting up a BBAI would definitely be the best solution for me if I don't eventually add a second Kiwi, but I have the impression that would require a fair amount of effort software-wise that exceeds my current skill set -  i.e. I had just enough to get wsprdaemon running on a Pi4.  Perhaps I should do is make it my summer 2022 project, when I return to the KX4AZ/T site where the KiwiSDR is running.

Generally speaking, my goal has been to maximize the unique station counts in a 24 hour period, without tying up the KiwiSDR too much.  When I occasionally run a supplementary hopper receiver, I adjust the bands based on the time of day, to focus on what makes the most sense, i.e. at night I don't bother with 15/12/10 meters, but 630/160m are worth including.  

This thread actually belongs in the wsprbeacon forum, it occurs to me.  My interest/curiosity is in devising a mathematical algorithm to predict how many unique stations might be missed on a specific band, if a receiver is only listening for x% of the time over the course of a 24 hour period.  Essentially it becomes a cumulative probability of spotting, taking into account the the proportion of the day that the beacons are transmitting, combined with the probability that the receiver will "see" a signal of adequate S/N ratio.  My sense is that a 20% duty cycle on a single band would yield much more than 20% of the total in comparison to a continuous (100%) receiving mode.  The question is what that % value might be.  And actually, this value could be extracted from the WSPR data base if one could interrogate it in the correct way, and see how many unique spots remain if 80% of them were randomly removed from entire set of raw spots in a day.  So I should be able to answer my original question once I improve my WSPR database skills.