LDPC decoder, corresponding method, system and computer program

1. A decoder of Low Density Parity Check codewords on GF(r q ), the decoder using the iterative belief propagation algorithm and comprising at least

storing means to store a posteriori information on variables;

updating means to update the a posteriori information on variables;

first computation means to compute variable to constraint messages from a posteriori information on variables and variable to constraint messages from previous iteration;

second computation means to compute constraint to variable messages from variable to constraint messages computed by the first computation means; and

third and fourth computation means to update the a posteriori information on variables;

wherein the decoder comprises:

first shuffle means to transfer the a posteriori information on variables to the first computation means;

second shuffle means to transfer information from the third computation means to the fourth computation means; and

compression-storage-decompression means of constraint to variable messages.

2. A decoder according to claim 1 , wherein at least one second computation means takes into account a predetermined number of the variable to constraint messages which have the lowest magnitude level.

3. A decoder according to claim 2 , wherein at least one second computation means outputs at least three different magnitudes of constraint to variable messages.

4. A decoder according to claim 2 , wherein the first computation means, the second computation means and the compression-storage-decompression means are gathered in parity check processing means, at least one parity check processing means comprising:

first decompression means to decompress constraint to variable messages of the previous iteration;

first computation means to compute variable to constraint messages from a posteriori information on variables and variable to constraint messages from previous iteration;

storage means to store compressed information representing constraints to variable messages;

pre-processing means to select lambda variable to constraint messages of lowest magnitude and respective references;

constraint to variable computation means to compute the lambda plus one different magnitudes of the constraint to variable messages generated; and

second decompression means to generate the constraint to variable messages of the current iteration.

5. A decoder according to claim 1 , wherein the third computation means generates the difference between constraint to variable messages of the current iteration and the one of the previous decoding iteration.

6. A decoder according to claim 1 , wherein the third computation means generates an updated version of a posteriori information on variables.

7. A decoder according to claim 1 , wherein the a posteriori information on variables is partially updated with constraint to variable information more than once during one decoding iteration.

8. A decoder according to claim 1 , wherein the a posteriori information on variables is partially updated with constraint to variable information as many times as there are parity constraints connected to the variable.

9. A decoder according to claim 1 , wherein the information related to a constraint is transferred sequentially through the first and second shuffle means.

10. A decoder according to claim 9 , wherein the sequential output order of the second computation means differs from its input order.

11. A system comprising a decoder of Low Density Parity Check codewords on GF(r q ), the decoder using the iterative belief propagation algorithm and comprising at least

storing means to store a posteriori information on variables on variables;

updating means to update the a posteriori information on variables;

first computation means to compute variable to constraint messages from a posteriori information on variables and variable to constraint messages from previous iteration;

second computation means to compute constraint to variable messages from variable to constraint messages computed by the first computation means; and

third and fourth computation means to update the a posteriori information on variables;

wherein the decoder comprises furthermore

first shuffle means to transfer the a posteriori information on variables to the first computation means; and

second shuffle means to transfer information from the second computation means to the storing means; and

compression-storage-decompression means of constraint to variable messages and wherein the system comprises furthermore means of reception of Low Density Parity Check codewords.

12. A system according to claim 11 , comprising further means to handle digital data belonging to the group consisting of:

data transmitted over a wireless channel;

data broadcast over a wireless channel; and

data stored on a medium.

13. A decoding method of Low Density Parity Check code words, the method using the iterative belief propagation algorithm and comprising at least following steps:

storing of a posteriori information on variables;

updating of the a posteriori information on variables;

first computation step computing variable to constraint messages from a posteriori information on variables and variable to constraint messages from previous iteration;

second computation step computing constraint to variable messages from variable to constraint messages from previous iteration; and

third computation step computing the a posteriori information on variables;

wherein the decoding method comprises furthermore

first shuffle step transferring the symbol a posteriori information to the first computation step;

second shuffle step transferring information from the second computation step to the storing step; and

compression-storage-decompression step of constraint to variable messages.

14. A computer program product to decode Low Density Parity Check code words, using the iterative belief propagation algorithm, the computer program product comprising program code instructions recorded on a carrier that can be used in a computer comprising computer-readable programming means to perform the following steps:

storing of a posteriori information on variables;

updating of the a posteriori information on variables;

first computation step computing variable to constraint messages from a posteriori information on variables and variable to constraint messages from previous iteration;

second computation step computing constraint to variable messages from variable to constraint messages from previous iteration;

third computation step computing the a posteriori information on variables;

first shuffle step transferring the symbol a posteriori information to the first computation step;

second shuffle step transferring information from the second computation step to the storing step; and

compression-storage-decompression step of constraint to variable messages.