Ok so if there are multiple solutions then there is data to ECC, not
100% ECC of a constant, so it is indeed correct that there is a need for
a registry, and I was wrong.
Leigh/WA5ZNU
On Fri, 1 Dec 2006 4:07 pm, Patrick Lindecker wrote:
> Hello Leigh and Jose,
>
>> The parameters to look for from Patrick are the RS code groups and
>
>> maximum correction coefficient. It is likely that the same seed data p
>> is used in all cases, so there is no need for a registry of data codes.
>> (I.e., if you can decode it, you don't need to know what it means since
>> you have already found the right decoder g and frequency f).
> I use a limited sub-set of the RS code groups because through
> translation in time and in frequency RS codes are not orthogonal
> between themselves which is logic as RS has a set of solutions which
> are cyclic by parts (I don't see this in the litterature but it is
> obvious when you see the solutions...).
>
> The problem is to stay with a fix relationship between each RS solution
> and its mode label ("0" for "BPSK31"...), exactly as in SSTV each code
> sent previously to the SSTV picture is related to a given SSTV code.
>
> 73
>
> Patrick
>
>> ----- Original Message -----
>>
>> From: Leigh L Klotz, Jr.
>>
>> To: [email protected]
>>
>> Sent: Friday, December 01, 2006 5:38 AM
>>
>> Subject: Re: [digitalradio] USA: No Advanced Digital HF Data Comms
>>
>> From what I can gather, the code is just an ECC'd data block and the
>>
>> contents of the data aren't that important; it is the decodability
>> itself that is. If you can use arbitrary modem decoder g() and decode
>> data p with center frequency f with g(p,f) yielding null|(q,r) where r
>> is the correction coefficient, then you iterate over all decoders g and
>> pick the one that produces the highest r, without reference to q. In
>> all likelihood, only one decoder g() will yield a non-null result,
>> however, if you also iterate over frequencies f then you may get
>> multiple succesful decodes.
>>
>> The parameters to look for from Patrick are the RS code groups and
>> maximum correction coefficient. It is likely that the same seed data p
>> is used in all cases, so there is no need for a registry of data codes.
>> (I.e., if you can decode it, you don't need to know what it means since
>> you have already found the right decoder g and frequency f).
>>
>> 73,
>> Leigh/WA5ZNU
>> On Thu, 30 Nov 2006 1:08 pm, Jose A. Amador wrote:
>>> I see it would be interesting to "register and publish" the codes that
>>> Patrick used by some "central authority"
>>> (maybe, Patrick himself). For me, it is similar to the I2C codes
>>> that the semiconductor industry
>>> uses, and Philips, as inventor of the concept, is the registering
>>> authority.
>
>
