DVSI Boards for P25 regeneration?

[motoham88]

Going against the grain as usual by not saving for a Quantar, I have been wondering if anyone is working on or thinking about buying one of DVSI's development kits for IMBE/AMBE repeater building? With respect to all the wonderful work that has been done by MattSR and others for decoding P25, I am interested in re-encoding the 4PSK signal for NAC gating through a real analog repeater with discriminator/direct modulator inputs and not a garbage-in/garbage-out or double-vocode box. I have a friend that is passing raw P25 through an MSF which is working great, but needs a third box for the "COR" line.

The DVSI page has some development kits but maybe I'm out in left field with this idea...along with considering Asterisk for linking. Astro25 linking is well behind MotoTRBO which requires an internet connection at the repeater, while Quantars need full-duplex synchronous data. Using an OEM device if they are inexpensive enough would provide the zeros and ones to work with versus "screen scraping" the data off the air.

This would be for playing around with on ham, and I have read the Astro manuals about the Abacus chips and what they do.

Thanks.

 

[kayfox]

Last time I checked, and from all the quotes I have seen, its expensive. But I'm not surprised, for the pleasure of integrating their products into your design companies charge anywhere from $80-3,000 for development kits. The chips can be cheap, Ive heard $20 in 1k quantities, but how to get them is not at all clear and DVSI does not seem to want to return my emails.

For what I'm doing I will probably either borrow libmbe (sp?) from OP25 or just omit IMBE decoding altogether.

 

[Astro Spectra]

The DVSI codec converts audio to and from digital. It can be had as a handy USB key for 300 bucks, see:

http://www.dvsinc.com/products/usb_3k.htm

Note they now have a P25 version of the USB stick, originally it was AMBE.This was posted on your DIU replacement thread kayfox, here:

https://www.p25.ca/threads/2440-Quantar-DIU-replacement-board


But why do you want to decode the audio at the repeater? The Quantar doesn't. Since the audio is not decoded there is no DVSI licence issue, although you would require the free Motorola essential patents portfolio for a commercial product.

For an error correcting repeater the 4FSK signal must be demodulated, the data frames recovered then error sorted before remodulation. I'm not an OP25 expert but I understand a repeater module is included that does just this.

@Motoham88, have you considered building a simple 4FSK slicer out of 'old school' op-amps and then using something like the CD4046 PLL to recognise the presence of the 4800 baud P25 modulation (9600 bits per second encoded as two bits per 4FSK symbol hence 4800 baud modulation)? You're not going to be able to decode the NAC or recover anything from the CAI but it will allow COR indication from incoming P25.

 

[MattSR]

or what I'm doing I will probably either borrow libmbe (sp?) from OP25 or just omit IMBE decoding altogether.

Just to clarify, libmbe is a part of DSD, which was anonymously authored and published through radioreference.com - OP25 has 2 different codec implementations - a proof of concept codec which is slightly buggy and another codec that is much better which was provided by Pavel Yazev.

 

[kayfox]

Just to clarify, libmbe is a part of DSD, which was anonymously authored and published through radioreference.com

Yeah, I will readily admit to my memory being some sorta sieve. Although I'm not surprised at the "anonymously authored" part at all. Does anyone know when the patents expire on this stuff anyway?

 

[Astro Spectra]

Here ya go:

http://www.google.com/?tbm=pts#q=imbe+dvsi&hl=en&tbo=d&tbm=pts

And US patents last for 20 years since 1995. I'm not an IP attorney, but looks like 2016 for IMBE and 2021 for AMBE.

 

[kayfox]

Based on this list here: http://wiki.multimedia.cx/index.php?title=IMBE

US#6199037 Joint quantization of speech subframe voicing metrics and fundamental frequencies
Filing date: Dec 4, 1997
Issue date: Mar 6, 2001
Expiry: (20 years after filing) Dec 4, 2017

US#5870405 Digital transmission of acoustic signals over a noisy communication channel
Filing date: Mar 4, 1996
Issue date: Feb 9, 1999
Expiry: (20 years after filing) Mar 4, 2016

US#5754974 Spectral magnitude representation for multi-band excitation speech coders
Filing date: Feb 22, 1995
Issue date: May 19, 1998
Expiry: (the longer of 17 years from issue date or 20 years from filing date) May 19, 2015

US#5664051 Method and apparatus for phase synthesis for speech processing
Filing date: Jun 23, 1994
Issue date: Sep 2, 1997
Expiry: (the longer of 17 years from issue date or 20 years from filing date) Sept 2, 2014

US#5630011 Quantization of harmonic amplitudes representing speech
Filing date: Dec 16, 1994
Issue date: May 13, 1997
Expiry: (the longer of 17 years from issue date or 20 years from filing date) Dec 16, 2014

US#5517511 Digital transmission of acoustic signals over a noisy communication channel
Filing date: Nov 30, 1992
Issue date: May 14, 1996
Expiry: (the longer of 17 years from issue date or 20 years from filing date) May 14, 2013

US#5491772 Methods for speech transmission
Filing date: May 3, 1995
Issue date: Feb 13, 1996
Expiry: (the longer of 17 years from issue date or 20 years from filing date) May 3, 2015

US#5247579 Methods for speech transmission
Filing date: Dec 3, 1991
Issue date: Sep 21, 1993
Expiry: (the longer of 17 years from issue date or 20 years from filing date) Dec 3, 2011

US#5226108 Processing a speech signal with estimated pitch
Filing date: Sep 20, 1990
Issue date: Jul 6, 1993
Expiry: (the longer of 17 years from issue date or 20 years from filing date) Sept 20, 2010

US#5226084 Methods for speech quantization and error correction
Filing date: Dec 5, 1990
Issue date: Jul 6, 1993
Expiry: (the longer of 17 years from issue date or 20 years from filing date) Dec 5, 2010

US#5216747 Voiced/unvoiced estimation of an acoustic signal
Filing date: Nov 21, 1991
Issue date: Jun 1, 1993
Expiry: (the longer of 17 years from issue date or 20 years from filing date) Nov 21, 2011

US#5081681 Method and apparatus for phase synthesis for speech processing
Filing date: Nov 30, 1989
Issue date: Jan 14, 1992
Expiry: (the longer of 17 years from issue date or 20 years from filing date) Nov 30, 2009


I tend to think that since in 1993 DVSI was already marketing IMBE and it entered service with at least Motorola's iDEN system in 1994, that most of the later patents may relate to AMBE but not to IMBE.

 

[motoham88]

@kayfox, thanks for the follow-up information.

@Astro Spectra, I was thinking of putting in a device that would do all the conversion work and provide a stream of 0's and 1's to work with. I did read the manuals for the USB-3K and a few other products they have, but no mention of prices on the DVSI website.

At this point it's just a wild idea.

 

[Magnus]

I don't think IDEN ever used IMBE, it was VSELP.

 

[Viper1-6]

I don't think IDEN ever used IMBE, it was VSELP.

That is correct.

 

[kayfox]

I don't think IDEN ever used IMBE, it was VSELP.

Yeah, digging around it looks like at some point IMBE and AMBE were introduced to iDEN. Because the original system guide says VSELP, but the technical documentation for the Condor line of iDEN handsets talks about IMBE and AMBE.

 

[kayfox]

Back to the issue at hand, the list of patents is from that wiki article, it says its sources are:

References:
ANSI/TIA/EIA 102.BABA-1998, "Association of Public Safety Communications Officials (APCO) Project 25 Vocoder Description".
ANSI/TIA/EIA 102.BABA-1, "Association of Public Safety Communications Officials (APCO) Project 25 Half-Rate Vocoder Annex" (July 2009)

This means that list contains both IMBE and AMBE patents. It would be nice to find a list of patents for IMBE only, because I think that most of them would be expired. Unfortunately I don't have access to the TIA-102 docs and I'm not too keen on paying $100-200 each for them.

 

[Magnus]

Ok, it was AMBE. When they were TDMA 3:1 it was VSELP and when they switched to 6:1 they also switched to AMBE.

 

[Astro Spectra]

The 2005 patent 6912496, not listed in the Wiki from Christmas Island (!), specifically mentioned prior art dating to 1991, and includes mention of the earlier versions of the TIA IS102 spec dating back to 1993. So 17 years (for patents prior to 1995) after that is 2008 or 2010. But (and this is where you need an IP attorney if you want to progress to actually selling something) care is needed with the newer and pretty broad 6377916 patent. Still if you wanted to use the prior art argument or a 'Gillette defense' against the 916 patent, you would at least have a chance.

Having said all that, nothing in a patent stops you from building something with the published IP otherwise protected by patent for personal use without pecuniary gain. Indeed this is how improvements of previously patented ideas are researched. Giving implementations away or selling them where you infringe.

As for the motoham's thread, as I said in #3, you don't need to implement IMBE to build a P25 repeater since the forward error correction techniques that are the key to correcting received errors and the regeneration of a clean signal are not encumbered and the OP25 repeater code is worth further investigation.