Product codes have been an effective coding method for communication channels where both random and burst error occur. We present a new approach to the structure and maximum likelihood (ML) decoding of product codes using Tanner (1981) graphs. For product codes having a sub-code which is a product of simple parity codes and repetition codes, we show how to obtain a sub-code with an acyclic Tanner graph and the largest possible distance. We show that in all cases of interest, a n-dimensional product code has such a structure. Wagner rule decoding is used on this sub-code and its cosets to obtain an effective and efficient maximum likelihood decoding of the given product code.
Product codes have been an effective coding method for communication channels where both random and burst error occur. We present a new approach to the structure and maximum likelihood (ML) decoding of product codes using Tanner (1981) graphs. For product codes having a sub-code which is a product of simple parity codes and repetition codes, we show how to obtain a sub-code with an acyclic Tanner graph and the largest possible distance. We show that in all cases of interest, a n-dimensional product code has such a structure. Wagner rule decoding is used on this sub-code and its cosets to obtain an effective and efficient maximum likelihood decoding of the given product code.
The maximum-likelihood decoding of linear block codes by Wagner rule decoding is discussed. In this approach, the Wagner rule decoding, which has been primarily applied to single parity check codes, is employed on acyclic Tanner graphs. Accordingly, a coset decoding equipped with Wagner rule decoding is applied to the decoding of a code C having a Tanner graph with cycles. A subcode C1 of C with acyclic Tanner graph is chosen as the base subcode. All cosets of C1 have the same Tanner graph and are distinguished by their values of parity nodes in the graph. The acyclic Tanner graph of C1, together with a trellis representation of the space of the parity sequences, represent the code C. This graphical representation provides a unified and systematic approach to search for an efficient method for the maximum-likelihood decoding of a given linear block code. It is shown that the proposed method covers the most efficient techniques known for the decoding of some important block codes, including the hexacode H6, extended Golay codes, Reed–Muller codes, Hamming code and (32, 16, 8) quadratic residue codes. The generalisation to the decoding of lattices is briefly explained.
Iterative decoding techniques have become a viable alternative for constructing high performance coding systems. In particular, the recent success of turbo codes indicates that performance close to the Shannon limit may be achieved. In this thesis, it is showed that many iterative decoding algorithms are special cases of two generic algorithms, the min-sum and sum-product algorithms, which also include non-iterative algorithms such as Viterbi decoding. The min-sum and sum-product algorithms are developed and presented as generalized trellis algorithms, where the time axis of the trellis is replaced by an arbitrary graph, the "Tanner graph". With cycle-free Tanner graphs, the resulting decoding algorithms (e.g., Viterbi decoding) are maximum-likelihood but suffer from an exponentially increasing complexity. Iterative decoding occurs when the Tanner graph has cycles (e.g., turbo codes); the resulting algorithms are in general suboptimal, but significant complexity reductions are possible compared to the cycle-free case. Several performance estimates for iterative decoding are developed, including a generalization of the union bound used with Viterbi decoding and a characterization of errors that are uncorrectable after infinitely many decoding iterations.
A new method is given for performing approximate maximum-likelihood (ML) decoding of an arbitrary binary linear code based on observations received from any discrete memoryless symmetric channel. The decoding algorithm is based on a linear programming (LP) relaxation that is defined by a factor graph or parity-check representation of the code. The resulting "LP decoder" generalizes our previous work on turbo-like codes. A precise combinatorial characterization of when the LP decoder succeeds is provided, based on pseudocodewords associated with the factor graph. Our definition of a pseudocodeword unifies other such notions known for iterative algorithms, including "stopping sets," "irreducible closed walks," "trellis cycles," "deviation sets," and "graph covers." The fractional distance d/sub frac/ of a code is introduced, which is a lower bound on the classical distance. It is shown that the efficient LP decoder will correct up to /spl lceil/d/sub frac//2/spl rceil/-1 errors and that there are codes with d/sub frac/=/spl Omega/(n/sup 1-/spl epsi//). An efficient algorithm to compute the fractional distance is presented. Experimental evidence shows a similar performance on low-density parity-check (LDPC) codes between LP decoding and the min-sum and sum-product algorithms. Methods for tightening the LP relaxation to improve performance are also provided.
We present a method for maximum likelihood decoding of the (48,24,12) quadratic residue code. This method is based on projecting the code onto a subcode with an acyclic Tanner graph, and representing the set of coset leaders by a trellis diagram. This results in a two level coset decoding which can be considered a systematic generalization of the Wagner rule. We show that unlike the (24,12,8) Golay code, the (48,24,12) code does not have a Pless-construction which has been an open question in the literature. It is determined that the highest minimum distance of a (48,24) binary code having a Pless (1986) construction is 10, and up to equivalence there are three such codes.
Let E/sub m/ denote the binary (m, m-1, 2) even weight code and /spl ominus/ the Kronecker product operation. A linear code of length m/spl times/t containing E/sub m//spl ominus/(t,1,t) can be decoded efficiently using a Tanner graph and a trellis diagram together with the Wagner decoding algorithm. Thus, it is important to determine whether a given code contains such a subcode. We show that the (18,9,6) binary quadratic residue code does not contain E/sub 6//spl ominus/(3,1,3). The maximum minimum distance of such a code is 5.