SDRTrunk - p25 decoder test release

[Boatanchor]

This project looks promising but for us mere mortals, the whole setup process does seem insanely complicated when you compare it to other existing software based trunking systems.

Hopefully, there will be a way to streamline (simplify) the installation at some point in the future.

Great to see cutting edge work taking place in this area though..

Keep up the good work

 

[mtindor]

Denny,

I did get a chance to install the latest SDRTrunk + build the latest JMBE. However, the new simulcast system here that just came online yesterday had no testing on it today. So I have not been able to use SDRTrunk to gauge P25 LSM voice performance.

Hopefully they will do some testing tomorrow. As soon as I'm able to test I'll let you know. The other system that I usually test is 30 miles away, and I had thunderstorms that required me to lower my 800 mhz yagi to a level where it just wasn't picking up that 450 mhz LSM site.

Mike

 

[KA1RBI]

I'm working on an equalizer to see if that improves decode rates.

Quite correct, the equalization issue is currently an unsolved problem in OP25, Will be looking forward to your findings on this. One thought is that from a channel modelling standpoint, LSM-style simulcast is difficult to distinuish from plain ordinary multi-path. That is in plain ordinary multipath you have one or more copies of the signal that arrive at slightly different times. As Mike also pointed out, that sounds a lot like LSM... Separately, not sure if CNYICC does this, but some LSM systems have an additional low-speed channel that (apparently) modulates the "AM component" of the LSM signal. Long ago I put together a then-unpublished page on this topic. No reason now not to release it - comments welcome. Anyway, the reason for mentioning is that I speculated that it might be used for dynamic equalizer training...

LSM Gallery - Page 2

Max

 

[DSheirer]

This project looks promising but for us mere mortals, the whole setup process does seem insanely complicated when you compare it to other existing software based trunking systems.

Hopefully, there will be a way to streamline (simplify) the installation at some point in the future.

Sorry about the complexity. I hope to add a script or maybe a menu option that will automate the download/compile process for the JMBE library for P25 audio support in the program. I'll be adding radio reference data support into the program as well, later this year.

 

[DSheirer]

Denny,

I did get a chance to install the latest SDRTrunk + build the latest JMBE. However, the new simulcast system here that just came online yesterday had no testing on it today. So I have not been able to use SDRTrunk to gauge P25 LSM voice performance.

Hopefully they will do some testing tomorrow. As soon as I'm able to test I'll let you know. The other system that I usually test is 30 miles away, and I had thunderstorms that required me to lower my 800 mhz yagi to a level where it just wasn't picking up that 450 mhz LSM site.

Mike

Thanks for the feedback Mike. Are you using a directional or non-directional antenna for monitoring the LSM sites?

 

[DSheirer]

Quite correct, the equalization issue is currently an unsolved problem in OP25, Will be looking forward to your findings on this. One thought is that from a channel modelling standpoint, LSM-style simulcast is difficult to distinuish from plain ordinary multi-path. That is in plain ordinary multipath you have one or more copies of the signal that arrive at slightly different times. As Mike also pointed out, that sounds a lot like LSM... Separately, not sure if CNYICC does this, but some LSM systems have an additional low-speed channel that (apparently) modulates the "AM component" of the LSM signal. Long ago I put together a then-unpublished page on this topic. No reason now not to release it - comments welcome. Anyway, the reason for mentioning is that I speculated that it might be used for dynamic equalizer training...

LSM Gallery - Page 2

Max

Max, thanks for sharing. I spent a good bit of time studying your other page on LSM. I'll post a reply with the results from the equalizer testing.

Denny

 

[mtindor]

Thanks for the feedback Mike. Are you using a directional or non-directional antenna for monitoring the LSM sites?

A directional (770 mhz) yagi to monitor a 450 mhz system 30 miles away, and both a yagi and a simple ducky antenna [doesn't take much] for the local system [which there was no testing on today either].

Mike

 

[Boatanchor]

Sorry about the complexity. I hope to add a script or maybe a menu option that will automate the download/compile process for the JMBE library for P25 audio support in the program. I'll be adding radio reference data support into the program as well, later this year.

That would be fantastic - Looking forward to future developments.

 

[KD0TAZ]

As I drool in anticipation of an executable release that I can actually play with, I have a question about the resource demands.. What kind of hardware would this run on? As I would be using it for streaming, I have been looking for ways to keep everything battery backed without having to use an enormous UPS. Since my RS scanner precludes me from using a Raspberry Pi (can only provide alpha tags with a Uniden), I have come up with the idea of using a Dell Optiplex FX160 - which is Atom based (330/Dual core @ 1.6GHz) and draws less than 30 watts at 12vdc - to replace my current streaming tower. Since the demands of the streaming server are virtually non-existent, I can install a stripped down XP image and run it off a battery. Would I be able to use such lightweight hardware for SDRTrunk? I know once Unitrunker integrated support for the RTL dongle and eliminated the need for SDR#, the resource demand dropped drastically..

 

[DSheirer]

What kind of hardware would this run on?

Processing requirements depend on how many tuners you are using, sample rate for each tuner, number of decoding channels and, for P25, the signal strength/decode quality. Weaker signals can require more error correction processing.

If you just want to decode and stream one channel and you set the tuner to a low sample rate, you should be ok with average hardware. You can also set the spectral display to a low frame rate and a low FFT bin size to further reduce processing.

I have come up with the idea of using a Dell Optiplex FX160 - which is Atom based (330/Dual core @ 1.6GHz) and draws less than 30 watts at 12vdc - to replace my current streaming tower. Since the demands of the streaming server are virtually non-existent, I can install a stripped down XP image and run it off a battery. Would I be able to use such lightweight hardware for SDRTrunk?

An atom processor might not be big enough, but I never tested one, so I can't say for certain. Best bet is to give it a try and see if it meets your needs.

Denny

 

[jcardani]

Hi Denny,

It's been a few weeks, how are things working out with the equalizer method?

 

[DSheirer]

Hi Denny, It's been a few weeks, how are things working out with the equalizer method?

Spring has really been tapping my spare time, as I imagine the same for everyone.

I've made some really good progress with the P25 decoding.

After reading Max's comments about how LSM pulse shapes the symbols, I removed one of the filters and that cleaned up the decoding significantly.

I coded and started testing an equalizer and then put it on hold while I resolve some issues with the IMBE audio converter.

The audio converter was taking between 10 and 34 milliseconds to decode each 20 millisecond chunk of audio, causing breaks and sputters during playback. On finding the bottleneck, I replaced the DFT/IDFT portion of the unvoiced audio generation with a standard FFT library and that dropped conversion time down to 0 - 1 milliseconds per frame. Result: no choppy audio.

There's still an issue in the voiced audio generation code. The voiced portion sounds tinny (like a blown speaker?). I think I've still got an error somewhere that I'm trying to track down..

I'd also like to integrate downloading and compiling the JMBE library into sdrtrunk before the next release. I imagine most folks don't like compiling code.

Hope to have a new build for folks to try later this weekend.

Denny

 

[vince48]

Thanks Denny
We all appreciate your hard work.

Vince48

 

[jcardani]

Thanks Denny! Great work!

 

[DSheirer]

I posted a new test release of sdrtrunk version 0.1.4 to the download folder. This version greatly improves P25 LSM decoding and P25 audio playback (when you include the JMBE library).

I also created an automated download/compile utility for the JMBE library. Download the jmbe_builder.zip (windows) or jmbe_builder.tar.gz (linux), unzip the file, and run the included script. Instructions are on the wiki page at: https://code.google.com/p/sdrtrunk/wiki/JMBE

Denny

 

[MattSR]

Great work Denny!

 

[Boatanchor]

Life is just too damned short and there are not enough hours in the day..

I've spent hours, maybe even days reading through the 'Quick Start' info and reconfiguring various tuners on a Windows machine, but the best I can do is get the control channel decoder to display the decoded data. I can't for the life of me, get the damned thing to trunk track a basic (non LSM) P25 system letalone actually decode any P25 audio.

I've now given up.

Maybe I'm just plain stupid, but life is too short to spend days trying to get software working when I can do excatly the same thing in minutes using other, relatively user friendly software.

I like the idea of being able to decode LSM data & voice for others who have such systems nearby, but fortunately I don't have any LSM systems around here to worry about.

I will keep watching and hoping that a user frendly version hits the streets, but at the moment, I'm done.

Anyway, keep up the good work..

 

[DSheirer]

I've spent hours, maybe even days reading through the 'Quick Start' info and reconfiguring various tuners on a Windows machine, but the best I can do is get the control channel decoder to display the decoded data. I can't for the life of me, get the damned thing to trunk track a basic (non LSM) P25 system letalone actually decode any P25 audio.

The P25 decoder does not support voice channel following yet.

It will decode the audio if you setup a decoder on a voice channel frequency and it will decode the control channel signalling and display the control messages.

I'll be adding the voice channel following shortly.

Denny

 

[mtindor]

I can't believe there haven't been comments here recently. I don't want to see you get discouraged, Denny. To be honest, the directions for the JMBE stuff didn't work on my Windows machine, even after installing the Java JRE and JDK, or just one or the other.

I know how to build JMBE on Linux, so it's no big deal for me. But others might experience the same difficulties with me with the JMBE autobuilder on Windows, and I wanted to let you know in case they just don't want to be the bearer of bad tidings. I suspect your system builds it with ease because you probably already have the requisite software installed for your programming environment.

As much as I'd love to be using SDRTrunk all the time, I'm tired of dealing with the cheap driving RTL tuners and am holding off on the notion that at some point you might have Airspy support And if Airspy support means that you need to get your hands on an Airspy, let me know and I'll have one shipped to ya.

mike

 

[jmjumps]

I can't believe there haven't been comments here recently. I don't want to see you get discouraged, Denny. To be honest, the directions for the JMBE stuff didn't work on my Windows machine, even after installing the Java JRE and JDK, or just one or the other.

I know how to build JMBE on Linux, so it's no big deal for me. But others might experience the same difficulties with me with the JMBE autobuilder on Windows, and I wanted to let you know in case they just don't want to be the bearer of bad tidings. I suspect your system builds it with ease because you probably already have the requisite software installed for your programming environment.

As much as I'd love to be using SDRTrunk all the time, I'm tired of dealing with the cheap driving RTL tuners and am holding off on the notion that at some point you might have Airspy support And if Airspy support means that you need to get your hands on an Airspy, let me know and I'll have one shipped to ya.

mike

I did not have any problems with the Windows autobuilder, but I had successfully built the software a while back (which as far as building software goes even that is straightforward). Can you provide the output from the .bat file? Maybe that can provide some insight into why it isn't working for you. I assume there is no jmbe*.jar output, rather than there being a problem when you try to use the .jar?

I have been quiet because the software isn't generally usable yet for P25 voice (channel following being the last piece). Denny has been making steady progress on core functionality. I look forward to when the pieces are together and we can provide more substantive feedback.