decode

Block decoder

Syntax

msg = decode(code,n,k,'hamming/fmt',prim_poly)
msg = decode(code,n,k,'linear/fmt',genmat,trt)
msg = decode(code,n,k,'cyclic/fmt',genpoly,trt)
msg = decode(code,n,k)
[msg,err] = decode(...)
[msg,err,ccode] = decode(...)
[msg,err,ccode,cerr] = decode(...)

Optional Inputs

InputDefault Value
fmtbinary
prim_polygfprimdf(m) where n = 2^m-1
genpolycyclpoly(n,k)
trt Uses syndtable to create the syndrome decoding table associated with the method's parity-check matrix

Description

For All Syntaxes

The decode function aims to recover messages that were encoded using an error-correction coding technique. The technique and the defining parameters must match those that were used to encode the original signal.

The For All Syntaxes section on the encode reference page explains the meanings of n and k, the possible values of fmt, and the possible formats for code and msg. You should be familiar with the conventions described there before reading the rest of this section. Using the decode function with an input argument code that was not created by the encode function might cause errors.

For Specific Syntaxes

msg = decode(code,n,k,'hamming/fmt',prim_poly) decodes code using the Hamming method. For this syntax, n must have the form 2m-1 for some integer m greater than or equal to 3, and k must equal n-m. prim_poly is a row vector that gives the binary coefficients, in order of ascending powers, of the primitive polynomial for GF(2m) that is used in the encoding process. The default value of prim_poly is gfprimdf(m). The decoding table that the function uses to correct a single error in each codeword is syndtable(hammgen(m)).

msg = decode(code,n,k,'linear/fmt',genmat,trt) decodes code, which is a linear block code determined by the k-by-n generator matrix genmat. genmat is required as input. decode tries to correct errors using the decoding table trt, where trt is a 2^(n-k)-by-n matrix.

msg = decode(code,n,k,'cyclic/fmt',genpoly,trt) decodes the cyclic code code and tries to correct errors using the decoding table trt, where trt is a 2^(n-k)-by-n matrix. genpoly is a row vector that gives the coefficients, in order of ascending powers, of the binary generator polynomial of the code. The default value of genpoly is cyclpoly(n,k). By definition, the generator polynomial for an [n, k] cyclic code must have degree n-k and must divide xn-1.

msg = decode(code,n,k) is the same as msg = decode(code,n,k,'hamming/binary').

[msg,err] = decode(...) returns a column vector err that gives information about error correction. If the code is a convolutional code, err contains the metric calculations used in the decoding decision process. For other types of codes, a nonnegative integer in the rth row of err indicates the number of errors corrected in the rth message word; a negative integer indicates that there are more errors in the rth word than can be corrected.

[msg,err,ccode] = decode(...) returns the corrected code in ccode.

[msg,err,ccode,cerr] = decode(...) returns a column vector cerr whose meaning depends on the format of code:

  • If code is a binary vector, a nonnegative integer in the rth row of vec2matcerr indicates the number of errors corrected in the rth codeword; a negative integer indicates that there are more errors in the rth codeword than can be corrected.

  • If code is not a binary vector, cerr = err.

Examples

On the reference page for encode, some of the example code illustrates the use of the decode function.

The example below illustrates the use of err and cerr when the coding method is not convolutional code and the code is a binary vector. The script encodes two five-bit messages using a cyclic code. Each codeword has 15 bits. Errors are added to the first two bits of the first codeword and the first bit of the second codeword. Then decode is used to recover the original message. As a result, the errors are corrected. err reflects the fact that the first message was recovered after correcting two errors, while the second message was recovered after correcting one error. cerr reflects the fact that the first codeword was decoded after correcting two errors, while the second codeword was decoded after correcting one error.

m = 4; n = 2^m-1; % Codeword length is 15.
k = 5; % Message length
msg = ones(10,1); % Two messages, five bits each
code = encode(msg,n,k,'cyclic');  % Encode the message.
% Now place two errors in first word and one error
% in the second word.  Create errors by reversing bits.
noisycode = code;
noisycode(1:2) = bitxor(noisycode(1:2),[1 1]');
noisycode(16) = bitxor(noisycode(16),1);
% Decode and try to correct the errors.
[newmsg,err,ccode,cerr] = decode(noisycode,n,k,'cyclic');
disp('Transpose of err is'); disp(err')
disp('Transpose of cerr is'); disp(cerr')

The output is below.

Single-error patterns loaded in decoding table.
 1008 rows remaining.
2-error patterns loaded.  918 rows remaining.
3-error patterns loaded.  648 rows remaining.
4-error patterns loaded.  243 rows remaining.
5-error patterns loaded.  0 rows remaining.
Transpose of err is
     2     1

Transpose of cerr is
     2     1

More About

expand all

Algorithms

Depending on the decoding method, decode relies on such lower-level functions as hammgen, syndtable, and cyclgen.

Was this topic helpful?