From: fsm12 on


I want to know that for a particular viterbi decoder e.g for 1/2,7 type
,with traceback depth of 48, if input data frame contains 2000 bits ,IS it
possible to correct only upto 4 errors in this data stream and is there any
relation between error correcting capability and location of these errors
in the data pattern?

From: Vladimir Vassilevsky on


fsm12 wrote:
> I want to know that for a particular viterbi decoder e.g for 1/2,7 type
> ,with traceback depth of 48, if input data frame contains 2000 bits ,IS it
> possible to correct only upto 4 errors in this data stream and is there any
> relation between error correcting capability and location of these errors
> in the data pattern?

D >= 2t + 1

STUPIDENT

From: Tim Wescott on
On Fri, 01 Jan 2010 01:10:15 -0600, fsm12 wrote:

> I want to know that for a particular viterbi decoder e.g for 1/2,7 type
> ,with traceback depth of 48, if input data frame contains 2000 bits ,IS
> it possible to correct only upto 4 errors in this data stream and is
> there any relation between error correcting capability and location of
> these errors in the data pattern?

OK, think. "Can I only correct four errors ever, no matter how long my
data stream is".

If yes, why would Viterbi decoders be so popular? Yes, it's a thing of
mathematical beauty, but engineers are paid because we help our clients
make money.

So, the answer must be no, because if it were yes then the cell phone
companies (which use pretty good protocols, all in all) wouldn't touch it
with a ten foot pole (or, for that matter, a ten foot russian).

In particular, a Viterbi decoder can massively reduce the error _rate_ in
a stream that has a given error _rate_ -- there will always be clusters
of errors that the Viterbi decoder cannot fix, but as the incoming error
rate goes down the outgoing error rate goes down way, way faster.

Why don't you experiment with a Viterbi decoder for a bit, and see for
yourself how this is true?

--
www.wescottdesign.com
From: fsm12 on
>
>
> I want to know that for a particular viterbi decoder e.g for 1/2,7
type
>,with traceback depth of 48, if input data frame contains 2000 bits ,IS
it
>possible to correct only upto 4 errors in this data stream and is there
any
>relation between error correcting capability and location of these
errors
>in the data pattern?
>
>

This does not answer my query at all> Is it not possible to Quantify the
values of errors corrected by a Viterbi decoder for a given input data
stream
(e.g. for a 2000 inputs bits fed to a decoder having free distance of
10.)

From: Eric Jacobsen on
On 1/2/2010 8:19 AM, fsm12 wrote:
>>
>> I want to know that for a particular viterbi decoder e.g for 1/2,7
> type
>> ,with traceback depth of 48, if input data frame contains 2000 bits ,IS
> it
>> possible to correct only upto 4 errors in this data stream and is there
> any
>> relation between error correcting capability and location of these
> errors
>> in the data pattern?
>>
>>
>
> This does not answer my query at all> Is it not possible to Quantify the
> values of errors corrected by a Viterbi decoder for a given input data
> stream
> (e.g. for a 2000 inputs bits fed to a decoder having free distance of
> 10.)

Unlike algebraic codes, convolutional codes are more sensitive to input
error distribution. Clumps of errors, especially as the clump lengths
compare to the constraint length, are far harder to correct with a
convolutional code than an algebraic code.

So, no, it is not possible to strictly say that a CC can correct N out
of M bits, because the distribution of the errors matters. Randomly
distributed errors are easier to correct than clumps. This is why
channel interleavers are sometimes used.


--
Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com