Provided herein are methods and systems for applying an ensemble comprising a plurality of neural network based decoders trained using actively selected training samples for decoding error correction encoded codewords which are also encoded for error detection before transmitted over transmission channels subject to interference. In particular, each of the neural network based decoders is associated with only a limited size region of the distribution space of the error correction code where the distribution space is partitioned based on error detection values computed for the encoded codewords. As such each of the decoders is specialized for decoding encoded codewords mapped to its limited size associated region. During run-time a received encoded codeword may be mapped to one of the regions and may be fed accordingly to one of the neural network based decoders of the ensemble which is associated with the mapped region.

Provided herein are methods and systems for applying an ensemble comprising a plurality of neural network based decoders trained using actively selected training samples for decoding error correction encoded codewords which are also encoded for error detection before transmitted over transmission channels subject to interference. In particular, each of the neural network based decoders is associated with only a limited size region of the distribution space of the error correction code where the distribution space is partitioned based on error detection values computed for the encoded codewords. As such each of the decoders is specialized for decoding encoded codewords mapped to its limited size associated region. During run-time a received encoded codeword may be mapped to one of the regions and may be fed accordingly to one of the neural network based decoders of the ensemble which is associated with the mapped region.

A memory controller system includes error correction circuitry and erasure decoder circuitry. A retry flow is triggered when the memory controller's error checking and correction (ECC) detects an uncorrectable codeword. Error correction circuitry generates erasure codewords from the codeword with uncorrectable errors. The memory controller computes the syndrome weight of the erasure codewords. For example, the erasure decoder circuitry receives the erasure codewords and computes the syndrome weights. Error correction circuitry orders the erasure codewords based on their corresponding syndrome weights. Then error correction circuitry selects a subset of the codewords, and sends them to erasure decoder circuitry. Erasure decoder circuitry receives the selected codewords and decodes them.

Systems and methods are disclosed for full utilization of a data path's dynamic range. In certain embodiments, an apparatus may comprise a circuit including a first filter to digitally filter and output a first signal, a second filter to digitally filter and output a second signal, a summing node, and a first adaptation circuit. The summing node combine the first signal and the second signal to generate a combined signal at a summing node output. The first adaptation circuit may be configured to receive the combined signal, and filter the first signal and the second signal to set a dynamic amplitude range of the combined signal at the summing node output by modifying a first coefficient of the first filter and a second coefficient of the second filter based on the combined signal.

A symbol-determining device according to an embodiment includes: a transmission line shortening unit that multiplies each symbol value of a symbol array that is part of an input signal by a tap gain of a linear digital filter and outputs a symbol array representing a sum of values acquired through the multiplication; a transmission line estimating unit that estimates a transfer function of a transmission line using an adaptive nonlinear digital filter on the basis of a symbol array representing a state of the transmission line; an addition comparison processing unit that calculates a minimum value of a distance function in a Viterbi algorithm on the basis of a metric that is calculated on the basis of the output of the transmission line shortening unit and the transfer function; and a path tracing-back determination unit that performs symbol determination by tracing back a trellis path in the Viterbi algorithm on the basis of the minimum value of the distance function.

Methods, devices, non-transitory computer-readable medium, and systems are described for compressing audio data. The techniques involve obtaining a sequence of digitized samples of an audio signal, performing a transform using the sequence of digitized samples, to generate a plurality of spectral lines, obtaining a group of spectral lines from the plurality of spectral lines, and quantizing the group of spectral lines to generate a group of quantized values. Quantizing the group of spectral lines to generate the group of quantized values may comprise performing a specialized rounding operation on a spectral line selected from the group of spectral lines and using the specialized rounding operation to force a group parity value, computed for the group of quantized values, to a predetermined parity value. One or more data frames based on the group of quantized values may be outputted.

A data processing method includes performing first equalization processing on a data stream that comprises a plurality of sub-data stream segments, performing segment de-interleaving on the data stream, separately performing first forward error correction (FEC) decoding on each sub-data stream segment in a data stream, performing, according to an equalization termination state of each sub-data stream segment, second equalization processing on each sub-data stream segment, performing second FEC decoding on the data stream, and outputting the data stream obtained according to the second FEC decoding in response to a preset iteration termination condition being met, or performing, in response to the preset iteration termination condition not being met, according to the equalization termination state of each sub-data stream segment obtained according to the first equalization, the second equalization processing on each sub-data stream segment obtained according to the second FEC decoding.

A decoding apparatus includes a multi-input branch metric calculation unit configured to calculate, by using a branch label corresponding to a path extending toward a state S at a time point N in a trellis diagram and a plurality of reception signal sequences, a branch metric in the state S, a path metric calculation unit configured to calculate a path metric in the state S at the time point N, and a surviving path list memory configured to store path labels corresponding to L path metrics among a plurality of calculated path metrics. The path metric calculation unit generates a path label in the state S at the time point N by combining the branch label with a path label in each of the states at the time point N−1 and the surviving path list memory outputs path labels corresponding to L path metrics.

A decoding apparatus includes a multi-input branch metric calculation unit configured to calculate, by using a branch label corresponding to a path extending toward a state S at a time point N in a trellis diagram and a plurality of reception signal sequences, a branch metric in the state S, a path metric calculation unit configured to calculate a path metric in the state S at the time point N, and a surviving path list memory configured to store path labels corresponding to L path metrics among a plurality of calculated path metrics. The path metric calculation unit generates a path label in the state S at the time point N by combining the branch label with a path label in each of the states at the time point N−1 and the surviving path list memory outputs path labels corresponding to L path metrics.

Provided is an optical transmission system that includes an optical transmitter, and an optical receiver. The optical transmitter includes: a signal coding unit configured to perform nonlinear trellis coding, which corresponds to nonlinear calculation, on a symbol sequence; and a modulator configured to modulate the symbol sequence subjected to the nonlinear trellis coding by the signal coding unit, and transmit the modulated symbol sequence to the optical receiver. The optical receiver includes: a light receiving unit configured to receive an optical signal transmitted from the optical transmitter, and convert the received optical signal into an electrical signal; and a digital signal processing unit configured to perform digital signal processing on the electrical signal to reconstruct the symbol sequence.