An apparatus including at least one processor configured to execute instructions and cause the apparatus to perform, obtaining for a first possible state (s) of a received sample at the current time step (k), log-likelihood ratio, Ilr, values Ilr.sub.old,min, Ilr.sub.old,max of a first transmitted bit (b.sub.j), wherein, the Ilr values Ilr.sub.old,min, Ilr.sub.old,max are respectively associated with a most likely state and a less likely state related to a received sample at the previous time step (k?1); determining based on path metrics and branch metrics corresponding to the received sample at the current time step (k); a first parameter (Q) related to a difference between likelihoods of the most likely state and the less likely state; updating magnitude of the Ilr value Ilr.sub.old,min at least based on the Ilr value Ilr.sub.old,min, the Ilr value Ilr.sub.old,max, and the first parameter, to obtain an updated Ilr value Ilr.sub.old,updated.

A cyclo-stationary characteristic of a communications channel and/or storage media is determined. The cyclo-stationary characteristic has K-cycles, K>1. Markov transition probabilities are determined that depend on a discrete phase ?=t mod K, wherein t is a discrete time value. An encoder to optimize the Markov transition probabilities for encoding data sent through the communications channel and/or stored on the storage media. The optimized Markov transition probabilities are used to decode the data from the communication channel and/or read from the storage media.

A method of soft-decision decoding including training a machine learning agent with communication signal training data; providing to the trained machine learning agent a signal that has been received via a communications channel; operating the machine learning agent to determine respective probabilities that the received signal corresponds to each of a plurality of symbols; and, based on the determined probabilities, performing soft decision decoding on the received signal.

Method and system for soft decision decoding are provided. A soft decision decoding method implemented by a receiver in a communication network may include: receiving a signal frame carrying a message through a communication network; obtaining data structure of the message; obtaining at least one bit of the message based on the data structure and known information; and decoding the received signal frame based on the at least one bit using soft decision decoding to obtain a decoding result. Decoding efficiency and accuracy may be improved.

A data storage device is disclosed comprising a non-volatile storage medium (NVSM), wherein a plurality of codewords and corresponding parity sector are written to the NVSM and then read from the NVSM. Each codeword read from the NVSM is processed using a Viterbi-type detector, thereby generating codeword reliability metrics. The codeword reliability metrics for at least some of the codewords are processed using a low density parity check (LDPC) type decoder, thereby generating a LDPC reliability metric for each symbol of at least one codeword. The LDPC reliability metrics for at least one of an un-converged codeword are processed using the parity sector, thereby updating the un-converged codeword reliability metrics. Processing the codeword reliability metrics with the LDPC decoder and updating the reliability metrics with the parity sector is repeated at least once before updating the LDPC reliability metrics of at least the un-converged codeword using the Viterbi-type detector.

A data storage device is disclosed comprising a head actuated over a disk comprising a data track having at least a first data segment and a second data segment. A first plurality of codewords are generated, and a first parity sector is generated over the first plurality of codewords. The first plurality of codewords and the first parity sector are written to the first data segment. A second plurality of codewords are generated, and a second parity sector is generated over the second plurality of codewords. The second plurality of codewords and the second parity sector are written to the second data segment. During a read operation the data segments of the data track are processed sequentially to decode the codewords using a low density parity check (LDPC) decoder, wherein the reliability metrics of un-converged codewords are stored in a codeword buffer and updated using the respective parity sector.

A method of processing a digitally encoded radio signal (102) comprising a bit to be determined is disclosed. The method comprises correlating a first bit sequence (103) comprising the bit with a plurality of predetermined filters (104a-h) to create a first set of filter coefficients (110a-h); calculating (120) a first likelihood data set (124) comprising a likelihood of said bit having a given value for each bit position from the first set of filter coefficients. A second bit sequence (103) comprising the bit at a different position is then correlated with the filters to create a second set of filter coefficients (10a-h), from which a second likelihood data set (124) is calculated. A soft output bit (26) comprising a probability weighted bit value from data corresponding to the bit at a first and second bit positions from the first and second likelihood data sets respectively is then calculated.

In one embodiment, a system provides for optimizing an error rate of data through a communication channel. The system includes a data generator operable to generate a training sequence as a Markov code, and to propagate the training sequence through the communication channel. The system also includes a Soft Output Viterbi Algorithm (SOVA) detector operable to estimate data values of the training sequence after propagation through the communication channel. The system also includes an optimizer operable to compare the estimated data values to the generated training sequence, to determine an error rate based on the comparison, and to change the training sequence based on the Markov code to lower the error rate of the data through the communication channel.

According to one embodiment, a memory controller includes: a memory I/F that reads a codeword written in a NAND memory as any one of hard-bit information, first soft-bit information, and second soft-bit information; a codeword processor that generates a codeword of an first soft-decision value from the first soft-bit information, and generates a codeword of a second soft-decision value from the second soft-bit information; a first decoder that executes hard-decision decoding on a codeword of a hard-decision value configured from the hard-bit information; a second decoder that executes first soft-decision decoding on the codeword of the first soft-decision value; and a third decoder that executes second soft-decision decoding on the codeword of the second soft-decision value, wherein the first soft-bit information includes information having a first number of bits greater than the number of bits of the hard-bit information, and the second soft-bit information includes information having a second number of bits greater than the number of bits of the first soft-bit information.

Methods and apparatus are provided for calculating branch metrics, associated with possible transitions between states of a trellis, in a sequence detector for detecting symbol values corresponding to samples of an analog signal transmitted over a channel. For each sample and each transition, the method calculates a plurality of distance values indicative of distance between that sample and respective hypothesized sample values for that transition. In parallel with calculation of the distance values, the sample is compared with a set of thresholds, each defined between a pair of successive hypothesized symbol values arranged in value order, to produce a comparison result. An optimum distance value is selected as a branch metric for the transition in dependence on the comparison result.