Progress Report
C.C. Li 2007/06/14
What I have done this week
1. Survey paper：
[1] High-Throughput Turbo-Sum-Product Decoding of QC LDPC Codes
Turbo Decoding of QC LDPC Codes
The turbo decoding algorithm can be directly used to decode QC LDPC codes. Since
for each ˜Hs there is exactly one 1 in each column, each ˜Hs defines a supercode
(0 ≤ s ≤ j − 1).
TSP Algorithm of QC LDPC Codes
We formulate the TSP decoding algorithm as follows:
1. Iteration 0 (Initialization): For i = 1, 2, · · · ,N, calculate the intrinsic LLR
information obtained from channel λi =
variable message
2. Iteration n (n ≥ 1):
i) For l = 1, 2, · · · ,m and ∀ i ∈ Rl calculate
, and initialize the check to
SSP ALGORITHM OF QC LDPC CODES
The group shuffled decoding algorithm can be directly applied to decode a QC LDPC
code. The parity check matrix of the QC LDPC code can be arbitrarily partitioned into
G groups column-wise.
SSP Algorithm of QC LDPC Codes
Given a QC LDPC code, we can view its H as the concatenation of two submatrices:
H = [¯H1 ¯H2], where ¯H1 is simply the first block column in H. Note that this
concatenation ensures that for ¯H1 there is exactly one 1 in each row. This
concatenation of H partitions the set of variable nodes in the Tanner graph of the
given LDPC code into two sets V1 and V2, which correspond to ¯H1 and ¯H2
respectively.
BER performance comparisons for the (211, 3, 5) SFT code (dash line: 5 iterations;
solid line: 15 iterations; dash-dot line: 50 iterations)
Comparisons of average number of iterations to converge for the (211, 3, 5) SFT code
6/14~6/20

Survey paper about H-QC & QC.