Abstract: In one embodiment, a multistage interconnection network (MIN) has two or more configurable stages, each stage having a plurality of switches. The network has one or more unused input terminals, each mapped using fixed switch connections to an unused output terminal. The network also has a set of used input terminals that are selectively mapped to a set of used output terminals based on values of control signals supplied to the stages. Each stage receives a different control signal, and each control signal is generated by cyclically shifting a control seed by a corresponding cyclic-shift value. Fixing the mappings of the unused terminals ensures that the used input terminals are not mapped to any unused output terminals. By storing only the control seed, memory requirements are reduced over networks that explicitly store individual control signals for all of the stages.

Abstract: In one embodiment, a multistage interconnection network (MIN) has two or more configurable stages, each stage having a plurality of switches. The network has one or more unused input terminals, each mapped using fixed switch connections to an unused output terminal. The network also has a set of used input terminals that are selectively mapped to a set of used output terminals based on values of control signals supplied to the stages. Each stage receives a different control signal, and each control signal is generated by cyclically shifting a control seed by a corresponding cyclic-shift value. Fixing the mappings of the unused terminals ensures that the used input terminals are not mapped to any unused output terminals. By storing only the control seed, memory requirements are reduced over networks that explicitly store individual control signals for all of the stages.

Abstract: In one embodiment, a configurable communications system accommodates a plurality of different transmission word sizes. In a transmit path, the system inserts a number of padding bits corresponding to missing user-data bits onto the end of an input data sequence to generate a set of data having N bits. The N bits are interleaved and error-correction (EC) encoded to generate parity bits corresponding to an EC codeword. The parity bits are de-interleaved and multiplexed with the input data stream to generate a transmission word. In a receive path, a channel detector recovers channel values corresponding to the transmission word. Padding values, corresponding to the missing-bit locations, are inserted among the channel values. The resulting channel values are interleaved and EC decoded to recover the EC codeword. The data bits of the codeword are de-interleaved, and the padding bits corresponding to the missing channel values are discarded.

Abstract: In one embodiment, a communications system has a write path and a read path. In the write path, a local/global interleaver interleaves a user data stream, and an error-correction (EC) encoder encodes the user data stream to generate an EC codeword. A local/global de-interleaver de-interleaves the parity bits of the EC codeword, and both the original un-interleaved user data and the de-interleaved parity bits are transmitted via a noisy channel. In the read path, a channel detector recovers channel soft-output values corresponding to the codeword. A local/global interleaver interleaves the channel values, and an EC decoder decodes the interleaved values to recover the original codeword generated in the write path. A de-multiplexer de-multiplexes the user data from the parity bits. Then, a local/global de-interleaver de-interleaves the user data to obtain the original sequence of user data that was originally received at the write path.

Abstract: Executed when a channel input (e.g., LDPC) codeword is written to a storage medium, a write-verification method (i) compares the channel input codeword to the written codeword, (ii) identifies any erroneous too bits, and (iii) stores the erroneous-bit indices to a record in a table. At some later time, the written codeword is read and sent to a decoder. If the decoder fails with a near codeword, a write-error recovery process searches the table and retrieves the erroneous-bit information. The codeword bits at those indices are adjusted, and the modified codeword is submitted to further processing.

Abstract: In one embodiment, a de-interleaver receives soft-output values corresponding to bits of an LDPC-encoded codeword. The de-interleaver has scratch pad memory that provides sets of the soft-output values to a local de-interleaver. The number of values in each set equals the number of columns in a block column of the LDPC H-matrix. Each set has at least two subsets of soft-output values corresponding to at least two different block columns of the LDPC H-matrix, where the individual soft-output values of the at least two subsets are interleaved with one another. Local de-interleaving is performed on each set such that the soft-output values of each subset are grouped together. Global de-interleaving is then performed on the subsets such that the subsets corresponding to the same block columns of the H-matrix are arranged together. In another embodiment, an interleaver performs global then local interleaving to perform the inverse of the de-interleaver processing.

Abstract: A communication system (e.g., a hard drive) having a random-access memory (RAM) for storing trapping-set (TS) information that the communication system generates on-line during a special operating mode, in which low-density parity-check (LDPC)-encoded test codewords are written to a storage medium and then read and decoded to discover trapping sets that appear in candidate codewords produced by an LDPC decoder during decoding iterations. The discovered trapping sets are filtered to select a subset of trapping sets that satisfy specified criteria. The discovery and filtering of trapping sets is performed based on error vectors that are calculated using the a priori knowledge of original test codewords. The TS information corresponding to the selected subset is stored in the RAM and accessed as may be necessary to break the trapping sets that appear in candidate codewords produced by the LDPC decoder during normal operation of the communication system.

Abstract: In one embodiment, an LDPC decoder performs a targeted bit adjustment method to recover a valid codeword after the decoder has failed. In a first stage, a post processor initializes the decoder by saturating LLR values output by the decoder during the last (i.e., failed) iteration to a relatively small value. Then, two-bit trials are performed, wherein LLR values corresponding to two bits of the codeword are adjusted in each trial. Decoding is performed with the adjusted values, and if the number of unsatisfied check nodes exceeds a specified threshold, then a second stage is performed. The post processor initializes the decoder by saturating the LLR values output by the decoder during the last (i.e., failed) iteration of the first stage to a relatively small value. The second stage then performs single-bit adjustment trials, wherein one LLR value corresponding to one bit of the codeword is adjusted in each trial.

Abstract: In one embodiment, a turbo equalizer is selectively operable in either first or second modes. In the first mode, layered (low-density parity-check (LDPC)) decoding is performed on soft-output values generated by a channel detector, where, for each full local decoder iteration, the updates of one or more layers of the corresponding H-matrix are skipped. If decoding fails to converge on a valid LDPC-encoded codeword and a specified condition is met, then LDPC decoding is performed in a second mode, where the updates of all of the layers of the H-matrix are performed for each full local decoder iteration, including the one or more layers that were previously skipped in the first mode. Skipping one or more layers in the first mode increases throughput of the decoder, while updating all layers in the second mode increases error correction capabilities of the decoder.

Abstract: Low-latency programmable encoders, and more particularly, low-latency programmable encoders which use low-density parity check (LDPC) codes in combination with an outer systematic code. The LDPC encoder is programmable for any irregular circulant-based LDPC code. The code profile, block length, number of block rows, and number of block columns can vary. The LDPC encoding and the outer systematic code encoding can proceed in a parallel manner (e.g., simultaneously) instead of in a serial manner.

Abstract: Various embodiments of the present invention provide systems and methods for encoding data. As an example, a data encoding circuit is disclosed that includes a first stage data encoder circuit and a second stage data encoder circuit. The first stage data encoder circuit is operable to provide a first stage output. The first stage data encoder circuit includes a first vector multiplier circuit operable to receive a data input and to multiply the data input by a first sparse matrix to yield a first interim value. The second stage encoder circuit includes a second vector multiplier circuit operable to multiply the first stage output by a second sparse matrix to yield a second interim value.

Abstract: Various embodiments of the present invention provide systems and methods for decoding data. As an example, a data processing circuit is disclosed that includes a multi-tier decoding circuit having a first tier decoding circuit operable to decode portions of an encoded data set exhibiting low row weight, and a second tier decoding circuit operable to decode portions of an encoded data set exhibiting high row weight.

Abstract: In one embodiment, an LDPC decoder has a controller and an extrinsic log-likelihood (LLR) value generator. The extrinsic LLR value generator is selectively configurable to operate in either (i) a non-averaging mode that updates extrinsic LLR values without averaging or (ii) an averaging mode that updates extrinsic LLR values using averaging. Initially, the extrinsic LLR value generator is configured to generate non-averaged extrinsic LLR values, and the decoder attempts to recover an LDPC-encoded codeword using the non-averaged extrinsic LLR values. If the decoder is unable to recover the correct codeword, then (i) the controller selects the averaging mode, (ii) the extrinsic LLR value generator is configured to generate average extrinsic LLR values, and (iii) the decoder attempts to recover the correct codeword using the average extrinsic LLR values. Averaging the extrinsic LLR values may slow down the propagation of erroneous messages that lead the decoder to convergence on trapping sets.

Abstract: A plurality of “local” interleavers replaces a single global interleaver for processing encoded data. If the encoded data may be represented as a matrix of data blocks, or “circulants,” each local interleaver can be the size of one or a small number of circulants. Thus, for example, if the matrix has a certain number of rows and columns, the number of local interleavers may be equal to the number of columns. Each local interleaver is small so latency is low.

Abstract: Various embodiments of the present invention provide systems and methods for data processing. For example, a data processing system is disclosed that includes a channel detector circuit. The channel detector circuit includes a branch metric calculator circuit that is operable to receive a number of violated checks from a preceding stage, and to scale an intrinsic branch metric using a scalar selected based at least in part on the number of violated checks to yield a scaled intrinsic branch metric.

Abstract: Certain embodiments of the present invention are efficient run-time methods for creating and updating a RAM list of dominant trapping-set profiles for use in (LDPC) list decoding. A decoded correct codeword is compared to a near codeword to generate a new trapping-set profile, and the profile written to RAM. Record is kept of how many times RAM has been searched since a profile was last matched. Profiles that have not been matched within a specified number of searches are purge-eligible. Purge-eligible profiles are further ranked on other factors, e.g., number of times a profile has been matched since it was added, number of unsatisfied check nodes, number of erroneous bit nodes. If there is insufficient free space in RAM to store a newly-discovered profile, then purge-eligible profiles are deleted, beginning with the lowest-ranked profiles, until either (i) sufficient free space is created or (ii) there are no more purge-eligible profiles.

Abstract: In one embodiment, a matrix-vector multiplication (MVM) component generates a product vector based on (i) an input matrix and (ii) an input vector. The MVM component has a permuter, memory, and an XOR gate array. The permuter permutates, for each input sub-vector of the input vector, the input sub-vector based on a set of permutation coefficients to generate a set of permuted input sub-vectors. The memory stores a set of intermediate product sub-vectors corresponding to the product vector. The XOR gate array performs, for each input sub-vector, exclusive disjunction on (i) the set of permuted input sub-vectors and (ii) the set of intermediate product sub-vectors to update the set of intermediate product subvectors, such that all of the intermediate product sub-vectors in the set are updated based on a current input sub-vector before updating any of the intermediate product sub-vectors in the set based on a subsequent input sub-vector.

Abstract: In one embodiment, a forward substitution component performs forward substitution based on a lower-triangular matrix and an input vector to generate an output vector. The forward substitution component has memory, a first permuter, an XOR gate array, and a second permuter. The memory stores output sub-vectors of the output vector. The first permuter permutates one or more previously generated output sub-vectors stored in the memory based on one or more permutation coefficients corresponding to a current block row of the lower-triangular matrix to generate one or more permuted sub-vectors. The XOR gate array performs exclusive disjunction on (i) the one or more permuted sub-vectors and (ii) a current input sub-vector of the input vector to generate an intermediate sub-vector. The second permuter permutates the intermediate sub-vector based on a permutation coefficient corresponding to another block in the current block row to generate a current output sub-vector of the output vector.

Abstract: Certain embodiments of the present invention are methods for the organization of trapping-set profiles in ROM and for the searching of those profiles during (LDPC) list decoding. Profiles are ranked by dominance, i.e., by their impact on the error-floor characteristics of a decoder. More-dominant trapping-set profiles contain information about both unsatisfied check nodes (USCs) and mis-satisfied check nodes (MSCs), while less-dominant trapping-set profiles contain information about only USCs. Trapping-set profile information is organized into a number of linked, hierarchical data tables which allow for the rapid location and retrieval of most-dominant matching trapping-set profiles using a pointer-chase search.

Abstract: In a communications system that demultiplexes user data words into multiple sub-words for encoding and decoding within different subword-processing paths, the minimum distance between bit errors in an extrinsic codeword can be increased by having corresponding interleavers/deinterleavers in the different subword-processing paths use different interleaving/deinterleaving algorithms.