Low complexity LDCP decoding
View Patent ↗A technique for low-density parity-check (LDPC) coding involves utilizing a fixed point implementation in order to reduce or eliminate reliance on floating point operations. The fixed point implementation can be used to calculate check node extrinsic L-value as part of an LDPC decoder in an LDPC system. The technique can include one or more of linear approximations, offset approximations, and node-limiting approximation. A system constructed according to the technique implements one or more of linear approximations, offset approximations, and node-limiting approximation.
1. A method comprising
receiving a codework;
repeating until no new values for an extrinsic L-value computation (ELC) parameter are called for in association with decoding the codeword:
selecting an ELC parameter value from a plurality of values based upon a variable;
computing extrinsic L-values for low-density parity-check (LDPC) check nodes and variable nodes using the ELC parameter for one or more iterations;
calculating a posteriori probability (APP) L-values for the codeword;
extracting information bits associated with the decoded codeword;
forwarding the information bits; and
computing the extrinsic L-values using an approach selected from the group of approaches consisting of: sum-product (SP), box-plus (BP), a posteriori probability (APP), variations of these, and combinations of these.
2. A method comprising
receiving a codeword;
repeating until no new values for an extrinsic L-value computation (ELC) parameter are called for in association with decoding the codeword:
selecting an ELC parameter value from a plurality of values based upon a variable;
computing extrinsic L-values for low-density parity-check (LDPC) check nodes and variable nodes using the ELC parameter for one or more iterations;
calculating a posteriori probability (APP) L-values for the codeword;
extracting information bus associated with the decoded codeword;
forwarding the information bits; and
computing the extrinsic L-values using an approximation selected from the group of approximations consisting of: min-sum (MS), linear (L), offset (O), non-linear (NL), variations of these, and combinations of these.
3. A system comprising:
codeword buffer;
an offset engine;
a low density parity-check (LDPC) decoding engine coupled to the codeword buffer and the offset engine;
a data buffer coupled to the LDPC decoding engine;
wherein, in operation,
the codeword buffer contains a codeword;
the offset engine provides a bias parameter, B, where 0≦B≦ln(d c −1) to the LDPC decoding engine, wherein ‘d c ' is a number of edges going into a check node, and further wherein ‘ln' is a natural log;
the LDPC decoding engine computes extrinsic L-values for low-density parity-check (LDPC) check nodes and variable nodes using the bias parameter;
the LDPC decoding engine calculates a posteriori probability (APP) L-values for the codeword and extracts information bits associated with the decoded codeword;
the data buffer has the information bits associated with the decoded codeword.
4. The system of claim 3 , wherein the bias parameter is selected from the group of parameters consisting of a bias parameter (B), an design parameter (q), or a combination of these.
5. The system of claim 3 , further comprising,:
an attenuation engine coupled to e LDPC decoding engine, wherein, in operation,
the attenuation engine provides an attenuation parameter to the LDPC decoding engine;
the LDPC decoding engine computes extrinsic L-values for LDPC check nodes and variable nodes using the attenuation parameter.
6. The system of claim 5 , wherein the attenuation parameter is selected from the group of parameters consisting of an attenuation parameter (A), a design parameter (q), or a combination of these.
7. The system of claim 3 , further comprising:
an approximation engine coupled to the LDPC decoding engine, wherein, in operation,
the approximation engine provides a node-limiting parameter (Q) to the LDPC decoding engine; the LDPC decoding engine computes extrinsic L-values for LDPC check nodes and variable nodes using the node-limiting parameter.
8. The system of claim 3 , further comprising:
a φ lookup table coupled to the LDPC decoding engine, wherein, in operation,
the approximation engine provides a φ value to the LDPC decoding engine, wherein φ(x)=ln[(exp(x)+1)/(exp(x)−1)] and the φ value is associated with a value derived from the function φ(x);
the LDPC decoding engine computes extrinsic L-values for LDPC check nodes and variable nodes using the φ value.
9. The system of claim 3 , wherein the LDPC decoding engine implements an approach selected from the group of approaches consisting of: sum-product (SP), box-plus (BP), a posteriori probability (APP), and a combination of these, and wherein the LDPC decoding engine computes the extrinsic L-value for the check node in association with the codeword in accordance with the implemented approach.
10. The system of claim 3 , wherein the LDPC decoding engine uses an approximation selected from the group of approximations consisting of min-sum (MS), linear (L), offset (O), or non-linear (NL), and wherein the LDPC decoding engine computes the extrinsic L-value fur the check node in association with the codeword in accordance with the implemented approximation.