Summary: | 碩士 === 國立中正大學 === 通訊工程研究所 === 96 === Abstract
Iterative decoding based on belief propagation (BP) has received significant attention recently, mostly due to its near Shannon-limit error performance for the decoding of low-density parity-check codes. Like maximum a posterior probability (MAP) decoding, it is a soft-in/soft-out decoding algorithm. It processes the received symbols recursively to improve the reliability of each symbol. It is shown that the BP algorithm [1][2]provide a powerful performance for iterative decoding of LDPC codes.
BP algorithm operating on the Tanner graph. The graph consists of two sets of nodes, namely, symbol nodes and check nodes. Each symbol node represents a symbol in a codeword and each check node represents a parity-check constraint. Every check node is connected by an edge to the symbol nodes it check. The TGs of LDPC codes are usually sparsely connected since the density of the non-zero elements in the parity-check matrix is low. The decoding complexity of LDPC code associated with BP algorithms evaluated based on the check-node and bit-node updates in [2] and [3], and it is reported the complexity of BP algorithm is much lower than Turbo codes. However, it is well known that BP algorithm converges after several tens to hundreds iterations because of its slow convergence speed [4].
BP algorithm is update with parallel scheme, in each iteration, all the symbol nodes and subsequently all the check nodes in the Tanner graph pass new messages to their neighbors. Another scheme is update neighbors nodes in accordance with sequence, for example [5][6], this decoding scheme have about the same complexity as BP algorithm, but it converge faster than BP algorithm, and its performance better than BP algorithm(low bit error rate). This paper decoding scheme is base on [5][6], and modify the algorithm in order to reduce average iterations, but require more cost of hardware, we compare performance propose scheme with BP algorithm and [5][6].
|