There are two ways to represent the LDPC for DSC application: (1) by (N,M), where N (input) is the number of information bits and M (output) is the number of syndromes; that is, only the syndromes are transmitted (2) by (K, N - K), where K (input) is the number of information bits and N - K (output) is the number of parity bits; that is, only parity bits are transmitted.
Source nodes are combination of information bits S1 to S6 and parity bits [P.sub.1] to [P.sub.3].
The proposed scheme punctures the extended parity bits according to the compression ratio and estimates the punctured parity bits using the modified decoding scheme at the decoder, whereas the algorithm described in  does not do estimation.
For every additional syndrome requested, instead of sending the parity bits [P.sub.1] to PL, the encoder sends uncoded source bits to the decoder.
The turbo encoder has three outputs: u--the sequence of information bits and [p.sub.1], [p.sub.2]--the sequences of parity bits. For error detections, unlike  where a "genie" cyclic redundancy check (CRC) was used, we used a 16-bit CRC code.
In this case we have to determine the weakest parity bits for which CIM should be applied.
Parity bits are generated for the quantized random signal X and are transmitted to the decoder side.
The generated Y is corrected with the transmitted parity bits through channel decoding.
With fewer parity bits, we can reconstruct a more accurate X' that is similar to X from SI Y.