초록 ▼

A binary error-correction coding method that produces certain codewords such that a collection of all valid codewords is a resulting error-correction code that can have large minimum distance. The method's assignment of codewords enables them to be represented as particular paths through a single tr

A binary error-correction coding method that produces certain codewords such that a collection of all valid codewords is a resulting error-correction code that can have large minimum distance. The method's assignment of codewords enables them to be represented as particular paths through a single trellis with branches labeled with permuted information symbols and permuted parity symbols in addition to conventional sequential information and parity symbols. Not all paths through trellis represent valid codewords. The resulting code is linear so real-time encoding can be implemented with a large generator matrix. Also an iterative method of error-correction decoding that is based on a single trellis with branches labeled with both permuted and sequential information and parity symbols, despite not all paths in the trellis corresponding to valid codewords.

대표청구항 ▼

I claim: 1. A method for error-correction coding, for a given number of N bits, of a given value for a binary vector of N information bits wherein a codeword value is assigned for a vector of N parity bits, and wherein the method comprises the steps of: (a) encoding the said vector of N information

I claim: 1. A method for error-correction coding, for a given number of N bits, of a given value for a binary vector of N information bits wherein a codeword value is assigned for a vector of N parity bits, and wherein the method comprises the steps of: (a) encoding the said vector of N information bits, by using a first binary convolutional encoder, to create an output vector, which the output is a first truncated output to a vector of N bits; (b) permuting the said vector of N information bits, with respect to a first permutation, to create a vector of N permuted information bits; (c) encoding the said vector of N permuted information bits, by using a second binary convolutional encoder, to create an output vector, which the output is a second truncated output to vector of N bits; (d) setting a first internal binary vector of N bits equal to a result value from binary addition of: the said first truncated output vector of N bits from the first binary convolutional encoder; and the said second truncated output vector of N bits from the second binary convolutional encoder; (e) permuting the said vector of N parity bits, with respect to a second permutation, to create a vector of N permuted parity bits; (f) encoding the said vector of N permuted parity bits, by using a third binary convolutional encoder, to create an output vector, which the output is a third truncated output to a vector of N bits; and (g) setting a second internal binary vector of N bits equal to a result value from binary addition of: the said vector of N parity bits; and the said third truncated output vector of N bits from the third binary convolutional encoder; and wherein the said codeword value is assigned to equalize the value of the said vector of N parity bits for which a result value for the said second internal binary vector of N bits is equal to the value for the said first internal binary vector of N bits, and wherein an encoding to produce the said codeword value is a matrix multiplication of a binary generator matrix value, which is stored in a memory, multiplying the said given value of said binary vector of N information bits. 2. The method of error-correction coding of claim 1, wherein the said binary addition of said step (d) further including one or more additional truncated vectors of N bits, each said additional truncated vector of N bits is created by permuting the said vector of N information bits, with respect to an additional permutation, to create an additional vector of N permuted information bits which is encoded, by using an additional binary convolutional encoder, to create another output vector, which the output is an additional truncated output to a vector of N bits. 3. The method of error-correction coding of claim 1, wherein the said binary addition of said step (g) further including one or more additional truncated vectors of N bits, each said additional truncated vector of N bits is created by permuting the said vector of N parity bits, with respect to an additional permutation, to create an additional vector of N permuted parity bits which is encoded, by using an additional binary convolutional encoder, to create another output vector, which the output is an additional truncated output to a vector of N bits.