# US 2017/0257122 A1 · Memory controller and decoding method

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2017/0257122 A1
- Document ID: US20170257122A1
- Status: Published application
- Kind: A1
- Application number: 15/259,690
- Filed: 2016-09-08
- Published: 2017-09-07
- Earliest claimed: 2016-03-02
- Assignees: Kabushiki Kaisha Toshiba
- Applicants: Kabushiki Kaisha Toshiba
- Inventors: Osamu TORII; Haruka OBATA
- Web page: https://patentgrep.com/patent/US20170257122A1

## CPC

- G06F 11/1012: {using codes or arrangements adapted for a specific type of error (G06F11/1048 takes precedence)}
- H03M 13/2909: {Product codes}
- H03M 13/2927: {Decoding strategies}
- H03M 13/3707: {Adaptive decoding and hybrid decoding, e.g. decoding methods or techniques providing more than one decoding algorithm for one code}
- H03M 13/3738: {with judging correct decoding}
- H03M 13/3746: {with iterative decoding}
- H03M 13/3753: {using iteration stopping criteria}
- H03M 13/09: Error detection only, e.g. using cyclic redundancy check [CRC] codes or single parity bit
- H03M 13/1515: {Reed-Solomon codes}
- H03M 13/152: {Bose-Chaudhuri-Hocquenghem [BCH] codes}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H03M: CODING; DECODING; CODE CONVERSION IN GENERAL (using fluidic means F15C4/00; optical analogue/digital converters G02F7/00; coding, decoding or code conversion, specially adapted for particular applications, see the relevant subclasses, e.g. G01D, G01R, G06F, G06T, G09G, G10L, G11B, G11C, H04B, H04L, H04M, H04N; ciphering or deciphering for cryptography or other purposes involving the need for secrecy G09C)

## Abstract

A memory controller according to an embodiment includes a first decoder which calculates first extrinsic value based on a decoding success rate specified using a first table showing a correspondence between first distance information indicating a square Euclidean distance between a first decode word and a first soft input value and a first decoding success rate indicating a probability of a decoding result that the first decode word is correct, and a second decoder which calculates second extrinsic value based on a decoding success rate specified using the second table showing a correspondence second distance information indicating a square Euclidean distance between a second decode word and a second soft input value and a second decoding success rate indicating a probability of a decoding result that the second decode word is correct.

## Claims

### Claim 1

A memory controller comprising:
a soft decision decoder configured to include
a first decoder which stores a first table, the first table showing a correspondence between first distance information indicating a square Euclidean distance between a first decode word and a first soft input value and a first decoding success rate indicating a probability of a decoding result that the first decode word is correct, calculates the first distance information based on a first word read as a soft decision value from nonvolatile memory, specifies the first decoding success rate associated to the first distance information from the first table, calculates first extrinsic value based on the first decoding success rate, and outputs the first extrinsic value, and
a second decoder which stores a second table, the second table showing a correspondence between second distance information indicating a square Euclidean distance between a second decode word and a second soft input value and a second decoding success rate indicating a probability of a decoding result that the second decode word is correct, calculates the second distance information based on an addition result of a second word read as a soft decision value from the nonvolatile memory and the first extrinsic value, specifies the second decoding success rate associated to the second distance information from the second table, calculates second extrinsic value based on the second decoding success rate, and outputs the second extrinsic value,
wherein a minimum value of the first decoding success rate registered in the first table is a first non-zero value, and
wherein a minimum value of the second decoding success rate registered in the second table is a second non-zero value.

### Claim 2 (depends on 1)

The memory controller according to claim 1,
wherein the first distance information Dist1 is a value calculated by the following equation,
Dist


1
=
∑
r
i

d
i
<
0

|
r
i
|
(where, ri is an i-th element of a vector having a value obtained by adding the first soft input value or the first soft input value and a first certain value as an element, and di is an i-th element of a decode word vector which is a vector expression of the first decode word), and
wherein the second distance information Dist2 is a value calculated by the following equation,
Dist


2
=
∑
r
j

d
j
<
0

|
r
j
|
(where, rj is a j-th element of a vector having a value obtained by adding the second soft input value or the second soft input value and a second certain value as an element, and dj is a j-th element of a decode word vector which is a vector expression of the second decode word).

### Claim 3 (depends on 2)

The memory controller according to claim 2,
wherein the second decoder calculates the second distance information based on the addition result obtained by adding the second word and the first extrinsic value using the first extrinsic value as first a priori value, and
wherein, in a case where the second extrinsic value is calculated by the second decoder, the first decoder calculates the first distance information based on an addition result obtained by adding the first word and the second extrinsic value using the second extrinsic value as second a priori value.

### Claim 4 (depends on 1)

The memory controller according to claim 1,
wherein the first decoding success rate registered in the first table is a value which is obtained by normalizing a first quantized value using a first maximum value of a first quantization bit width,
wherein the second decoding success rate registered in the second table is a value which is obtained by normalizing a second quantized value using a second maximum value of a second quantization bit width,
wherein the first value is a value which is obtained by normalizing a minimum unit of the first quantization using the first maximum value, and
wherein the second value is a value which is obtained by normalizing a minimum unit of the second quantization using the second maximum value.

### Claim 5 (depends on 1)

The memory controller according to claim 1,
wherein the first decoder performs a first decoding using a first reception word corresponding to the first word read from the nonvolatile memory as the first soft input value, calculates the first distance information based on a first decode word obtained by the first decoding, specifies the first decoding success rate associated to the first distance information from the first table, calculates the first extrinsic value based on the first decoding success rate, and outputs the first extrinsic value, and
wherein the second decoder performs a second decoding using an addition result of a second reception word corresponding to the second word read from the nonvolatile memory and the first extrinsic value as the second soft input value, calculates the second distance information based on a second decode word obtained by the second decoding, specifies the second decoding success rate associated to the second distance information from the second table, calculates the second extrinsic value based on the second decoding success rate, and outputs the second extrinsic value.

### Claim 6 (depends on 5)

The memory controller according to claim 5,
wherein the soft decision decoder includes a completion determinator which determines whether the decoding is completed based on a decoding result of the second decoding,
wherein the first decoder iterates the first decoding until the completion determinator determines that the decoding is completed,
wherein the second decoder iterates the second decoding until the completion determinator determines that the decoding is completed, and
wherein, in a case where the second extrinsic value is calculated, the first decoder uses an addition result of the first reception word and the second extrinsic value as the first soft input value.

### Claim 7 (depends on 6)

The memory controller according to claim 6,
wherein, in a case where the completion determinator determines that the decoding is completed as a result of iterating the first decoding of the first decoder and the second decoding of the second decoder a certain number of times, the soft decision decoder sets the first distance information and the second distance information to zero.

### Claim 8 (depends on 1)

The memory controller according to claim 1,
wherein the first word is a first-dimensional code word which configures a product code, and
wherein the second word is a second-dimensional code word which configures the product code.

### Claim 9

A decoding method performed in a memory controller which includes a soft decision decoder, the soft decision decoder performing a decoding on a code word read as a soft decision value from nonvolatile memory,
wherein the soft decision decoder is configured to
calculate first distance information based on a first word read as the soft decision value from the nonvolatile memory,
specify a first decoding success rate associated to the first distance information from a first table, the first table showing a correspondence between the first distance information indicating a square Euclidean distance between a first decode word and a first soft input value and a first decoding success rate indicating a probability of a decoding result that the first decode word is correct,
calculate first extrinsic value based on the first decoding success rate and output the first extrinsic value,
calculate second distance information based on an addition result of a second word read as the soft decision value from the nonvolatile memory and the first extrinsic value,
specify a second decoding success rate associated to the second distance information from a second table, the second table showing a correspondence between the second distance information indicating a square Euclidean distance between a second decode word and a second soft input value and a second decoding success rate indicating a probability of a decoding result that the second decode word is correct, and
calculate second extrinsic value based on the second decoding success rate and output the second extrinsic value,
wherein a minimum value of the first decoding success rate registered in the first table is a first non-zero value, and
wherein a minimum value of the second decoding success rate registered in the second table is a second non-zero value.

### Claim 10 (depends on 9)

The decoding method according to claim 9,
wherein the first distance information Dist1 is a value calculated by the following equation,
Dist


1
=
∑
r
i

d
i
<
0

|
r
i
|
(where, ri is an i-th element of a vector having a value obtained by adding the first soft input value or the first soft input value and a first certain value as an element, and di is an i-th element of a decode word vector which is a vector expression of the first decode word), and
wherein the second distance information Dist2 is a value calculated by the following equation,
Dist


2
=
∑
r
j

d
j
<
0

|
r
j
|
(where, rj is a j-th element of a vector having a value obtained by adding the second soft input value or the second soft input value and a second certain value as an element, and dj is a j-th element of a decode word vector which is a vector expression of the second decode word).

### Claim 11 (depends on 10)

The decoding method according to claim 10,
wherein the soft decision decoder is configured to
calculate the second distance information based on the addition result obtained by adding the second word and the first extrinsic value using the first extrinsic value as first a priori value, and
in a case where the second extrinsic value is calculated by the second decoder, calculate the first distance information based on an addition result obtained by adding the first word and the second extrinsic value using the second extrinsic value as second a priori value.

### Claim 12 (depends on 9)

The decoding method according to claim 9,
wherein the first decoding success rate registered in the first table is a value which is obtained by normalizing a first quantized value using a first maximum value of a first quantization bit width,
wherein the second decoding success rate registered in the second table is a value which is obtained by normalizing a second quantized value using a second maximum value of a second quantization bit width,
wherein the first value is a value which is obtained by normalizing a minimum unit of the first quantization using the first maximum value, and
wherein the second value is a value which is obtained by normalizing a minimum unit of the second quantization using the second maximum value.

### Claim 13 (depends on 9)

The decoding method according to claim 9,
wherein the soft decision decoder is configured to
perform a first decoding using a first reception word corresponding to the first word read from the nonvolatile memory as the first soft input value to calculate the first distance information based on a first decode word obtained by the first decoding, and
perform a second decoding using an addition result of a second reception word corresponding to the second word read from the nonvolatile memory and the first extrinsic value as the second soft input value to calculate the second distance information based on a second decode word obtained by the second decoding.

### Claim 14 (depends on 13)

The decoding method according to claim 13,
wherein the soft decision decoder is configured to
determine whether the decoding is completed based on a decoding result of the second decoding,
iterate the first decoding and the second decoding until it is determined that the decoding is completed, and
in a case where the second extrinsic value is calculated, use an addition result of the first reception word and the second extrinsic value as the first soft input value.

### Claim 15 (depends on 14)

The decoding method according to claim 14,
wherein, in a case where it is determined that the decoding is completed as a result of iterating the first decoding and the second decoding a certain number of times, the soft decision decoder sets the first distance information and the second distance information to zero.

### Claim 16 (depends on 9)

The decoding method according to claim 9,
wherein the first word is a first-dimensional code word which configures a product code, and
wherein the second word is a second-dimensional code word which configures the product code.
