Multiple error detecting and correcting system employing Reed-Solomon codes

• US 4,413,339 A
• Filed: 06/24/1981
• Issued: 11/01/1983
• Est. Priority Date: 06/24/1981
• Status: Expired due to Term

• Current Assignee: Maxtor Corporation (Seagate Technology Holdings Plc)
• Original Assignee: Digital Equipment Corporation (HP Inc.)
• Inventors: Weng, Lih-Jyh, Field, Norman A., Riggle, Charles M.
• Primary Examiner(s): Smith, Jerry
• Assistant Examiner(s): Harkcom, Gary V.
• Application Number:
  US06/277,060
• Time in Patent Office: 860 Days
• Field of Search: 371/37-41, 371/45, 371/54, 364/200 MS File, 364/900 MS File
• US Class Current: 714/765
• CPC Class Codes: H03M 13/151 using error location or error correction polynomials

Abstract

An error detecting and correcting system implementing the Reed-Solomon (1023, 1006) code having code words whose symbols are elements in the Galois field GF(2¹⁰) generated by either the primitive polynomial x¹⁰ +x³ +1 or x¹⁰ +x⁷ +1. An original data word is encoded to produce a code word w(x) including a first set of checksum symbols appended thereto. Upon retrieval, the data symbols of the receive code word y(x) are encoded by the same encoder that encodes the original data word to produce a second set of checksum symbols. Both sets of checksum symbols are modulo-two summed to produce a residue R(x) from which error syndromes S_i can be computed and thus enable rapid correction of errors in the received code word y(x). The system also monitors the number of non-zero symbols in the residue R(x) in order to avoid unnecessary computation of error syndromes S_i and other decoding routines, such as when the received code word y(x) is otherwise uncorrectable or when the error exists only in the received checksum symbols, rather than in the data symbols. The distance between code words being (2T+ 2), the error correction routine is bypassed when the number of non-zero symbols in R(x) is less than or equal to T, which indicates that errors only reside in the checksum symbols. When the number of non-zero symbols equals (T+1), the error is uncorrectable. For determining whether a single error exists so that correction can quickly be made, the system also tests whether S_i+1 /S_i is constant for all error syndromes S_i.

Claims

1. A circuit for detecting and correcting errors occurring in a Reed-Solomon code word w(x) constituted by a data word d(x) having k data symbols and a checksum word E(x) having n-k data symbols, the symbols of said checksum word E(x) being elements in the Galois Field GF(2^m), for any integer m greater than 0, said checksum word being derived by encoding said data word d(x) by a predetermined generator polynomial g(x) having roots α

ⁱ wherein α

is a primitive element in the Galois Field GF(2^m), said circuit comprising;

A. receiving means for receiving a word y(x) that was transmitted as the code word w(x), said word y(x) being comprised of a data word c(x) having k symbols and a checksum word E1(x) having n-k symbols,

B. Reed-Solomon encoding means for encoding said data word c(x) by said generator polynomial g(x) for producing another checksum word E2(x), said encoding means comprising successively connected shift register stages and feedback means that enable encoding of the data word c(x) by said generator polynomial g(x) during shifting of the symbols thereof through said shift register,

C. residue generating means connected to said receiving means and to said encoding means for receiving E1(x) and E2(x) for producing a residue R(x) therefrom by modulo-two summing respective symbols thereof,

D. monitoring means connected to said residue generating means for indicating whether errors have occurred in the received word y(x),

E. logic processing means responsive to said monitoring means for computing error syndromes S_i from said residue R(x) thereby to enable the computation of error location signals and error value signals corresponding to the respective locations and values of errors occurring in data word c(x), and

F. correcting means responsive to said error location and error value signals for correcting errors occurring in data word c(x) located in said receiving means.

2. A circuit as recited in claim 1 wherein said predetermined generator polynomial g(x) is characterized by having roots α

ⁱ wherein α

is a primitive element in GF(2¹⁰) generated by the primitive polynomial x¹⁰ +x³ +1, and said Reed-Solomon encoding means comprises;

i. means for encoding said received data word c(x) by said generator polynomial g(x) characterized as such thereby to produce the checksum word E2(x).

3. A circuit as recited in claim 1 wherein said predetermined generator polynomial g(x) is characterized by having roots α

ⁱ wherein α

is a primitive element in GF(2¹⁰) generated by the primitive polynomial x¹⁰ +x⁷ +1, and said Reed-Solomon encoding means comprises;

i. means for encoding said received data word c(x) by said generator polynomial g(x) characterized as such thereby to produce the checksum word E2(x).

4. A circuit as recited in claim 1 for correcting T symbols of the received word y(x) that are in error wherein said predetermined generator polynomial g(x) is further characterized by having (2T+1) factors which are symmetrical about the factor (x+α

⁰), and said Reed-Solomon encoding means comprises;

i. means for encoding said received data word c(x) by said generator polynomial characterized as such thereby rendering symmetrical the feedback divider coefficients of g(x) applied to the shift register stages of said Reed-Solomon encoding means.

5. A circuit as recited in claim 1, 2, 3, or 4 wherein said monitoring means comprises:

i. means for producing a first control signal when the number of non-zero symbols in the residue R(x) is less than or equal to T thereby indicating that c(x) is correct,

ii. means for producing a second control signal when the number of non-zero symbols in the residue R(x) is equal to (T+1) thereby indicating the c(x) is uncorrectable, and

iii. means for producing a third control signal when the number of non-zero symbols in the residue R(x) is greater than or equal to (T+2) thereby indicating that c(x) is erroneous and possibly correctable.

6. A circuit as recited in claim 5 wherein said logic processor means further includes:

i. single error testing means for testing whether S_i+l /S_i is constant for all syndromes S_i thereby to immediately determine whether a single error exists in the data word c(x), and

ii. means responsive to said single error testing means and connected to said correcting means for generating a single error location signal and single error value signal thereby to enable said correcting means to immediately correct said single error without significant delay.

7. A circuit as recited in claim 6 wherein T equals eight and said generator polynomial g(x) is further characterized by:

##EQU2## thereby to provide for the correction of eight 10-bit symbols in y(x) which are in error and the capability of detecting whether nine or more 10-bit symbols in y(x) are in error.

8. In a digital data processing system including data storage means, a circuit for detecting and correcting errors occurring in digital data that is transferred between the data processing system and said data storage means, which circuit comprises:

A. Reed-Solomon encoding means for encoding a data word d(x) having k symbols upon transfer thereof from the data processing system to said data storage means thereby to produce an n-symbol code word w(x) constituted by said k-symbol data word d(x) and a checksum word E(x) having n-k symbols, said encoding means comprising plural successively connected shift register stages and feedback means that enable encoding of the data word d(x) by a generator polynomial g(x) during shifting of the data word through said shift register stages,

B. means for storing said code word w(x) in said data storage means,

C. retrieval means for retrieving from said data storage means a word y(x) that was previously stored as the code word w(x), said word y(x) being constituted by a data word c(x) having k symbols and a checksum word E1(x) having n-k symbols, said retrieval means including;

i. receiving means connected to said data storage means for receiving the data word c(x) and the checksum word E1(x),

ii. means connected to said Reed-Solomon encoding means and the data storage means for producing a second checksum word E2(x) having n-k symbols by encoding the data word c(x) with the same Reed-Solomon encoding means that produced the checksum word E1(x),

iii. residue generating means that receives the symbols of the checksum words E1(x) and E2(x) for producing a residue R(x) therefrom by modulo-two summing respective symbols thereof,

iv. monitoring means that receives said residue R(x) for indicating whether errors have occurred in the word y(x),

v. logic processing means responsive to said monitoring means for computing error syndromes S_i from said residue R(x) thereby to enable the computation of error location signals and error value signals corresponding to the respective locations and values of errors occurring in data word c(x), and

vi. correcting means responsive to said error location and error value signals for correcting errors occurring in said data word c(x) which are located in said receiving means prior to being transferred to said data processing system.

9. A circuit as recited in claim 8 for detecting and correcting errors in digital data further including:

D. connection means for handling the transfer of data between said data processing system and said data storage means, said connection means comprising;

i. input enabling means for enabling the transfer of the data word d(x) from the data processing system or the data word c(x) from the data storage means to said encoding means,

ii. feedback enabling means for enabling the encoding of the data word d(x) or c(x),

iii. clocking means for successively shifting each symbol of the data word d(x) or c(x) into said encoding means,

iv. first control means operative upon transfer of the data word d(x) from the data processing system to said encoding means for

(a) energizing said input enabling means and said feedback enabling means to successively shift and encode the symbols of d(x) under control of said clocking means thereby to produce the checksum word E(x),

(b) de-energizing said input enabling means and said feedback enabling means after k symbols of d(x) have been shifted and encoded by said encoding means, and

(c) for enabling the symbols of said checksum word E(x) then present in said encoding means to be transferred to said data storage means under control of said clocking means, and

v. second control means operative upon transfer of the data word c(x) from the data storage means to said encoding means for

(a) energizing said input means and said feedback means to successively shift and encode the symbols of c(x) under control of said clocking means thereby to produce the checksum word E2(x), and

(b) disabling said feedback means after the transfer of the data word c(x) thereto thereby to provide the continued shifting of the symbols of the checksum word E1(x) from said data storage means into the shift registers of said encoding means for producing said residue R(x) under control of said clocking means.

10. A circuit as recited in claim 8 or 9 wherein said Reed-Solomon encoding means comprises means for encoding said transmitted data words d(x) and said received data word c(x) by a generator polynomial g(x) whose factors are powers of a primitive element α

in the Galois Field (2¹⁰) generated by the primitive polynomial x¹⁰ +x³ +1 thereby to respectively produce the checksum words E1(x) and E2(x) from which said residue R(x) is produced.

11. A circuit as recited in claim 10 for correcting T symbols of the received code word y(x) wherein said generator polynomial g(x) is further characterized by having (2T+1) factors which are symmetrical about the factor (X+α

⁰) thereby rendering symmetrical the feedback divider coefficients of g(x) applied to the shift register stages of said Reed-Solomon encoding means.

12. A circuit as recited in claim 11 wherein said monitoring means comprises:

i. means for producing a first control signal when the number of non-zero symbols in the residue R(x) is less than or equal to T thereby indicating that c(x) is correct,

ii. means for producing a second control signal when the number of non-zero symbols in the residue R(x) is equal to (T+1) thereby indicating the c(x) is uncorrectable, and

iii. means for producing a third control signal when the number of non-zero symbols in the residue R(x) is greater than or equal to (T+2) thereby indicating that c(x) contains errors which may be correctable,

and said logic processor means further includes;

i. first control means responsive to said first control signal for immediately transferring said data word c(x) to said data processing system,

ii. second control means responsive to said second control signal for again retrieving said code word y(x), and

iii. third control means responsive to said third control signal for computing syndromes S_i thereby to enable computation of said error value and error location signals.

13. A circuit as recited in claim 12 wherein said logic processor means further includes:

i. single error testing means for testing whether S_i+l /S_i is constant for all syndromes thereby to immediately indicate that the received code word y(x) contains only a single error, and

ii. means responsive to said single error testing means and connected to said correcting means for generating a single error location signal and single error value signal thereby to enable said correcting means to immediately correct said single error.

14. A circuit as recited in claim 13 wherein T equals eight and said generator polynomial g(x) being further characterized by:

##EQU3## thereby to provide for the correction of eight 10-bit symbols of d(x) which are in error and the capability of detecting whether nine or more 10-bit symbols of y(x) are in error.

15. A circuit as recited in claim 8 or 9 wherein said Reed-Solomon encoding means comprises means for encoding said transmitted data words d(x) and said received data word c(x) by a generator polynomial g(x) whose factors are powers of a primitive element α

in the Galois Field (2¹⁰) generated by the primitive polynomial x¹⁰ +x⁷ +1 thereby to respectively produce the checksum words E1(x) and E2(x) from which said residue R(x) is produced.

16. A circuit as recited in claim 15 for correcting T symbols of the received word y(x) wherein said generator polynomial g(x) is further characterized by having (2T+1) factors which are symmetrical about the factor (X+α

⁰) thereby rendering symmetrical the feedback divider coefficients of g(x) applied to the shift register stages of said Reed-Solomon encoding means.

17. A circuit as recited in claim 16 wherein said monitoring means comprises:

i. means for producing a first control signal when the number of non-zero symbols in the residue R(x) is less than or equal to T thereby indicating that c(x) is correct,

ii. means for producing a second control signal when the number of non-zero symbols in the residue R(x) is equal to (T+1) thereby indicating the c(x) is uncorrectable, and

iii. means for producing a third control signal when the number of non-zero symbols in the residue R(x) is greater than or equal to (T+2) thereby indicating that c(x) contains errors which may be correctable,

and said logic processor means further includes;

i. first control means responsive to said first control signal for immediately transferring said data word c(x) to said data processing system,

ii. second control means responsive to said second control signal for again retrieving said code word y(x), and

iii. third control means responsive to said third control signal for computing syndromes S_i thereby to enable computation of said error value and error location signals.

18. A circuit as recited in claim 17 wherein said logic processor means further includes:

i. single error testing means for testing whether S_i+l /S_i is constant for all syndromes thereby to immediately indicate that the received code word y(x) contains only a single error, and

ii. means responsive to said single error testing means and connected to said correcting means for generating a single error location signal and single error value signal thereby to enable said correcting means to immediately correct said single error.

19. A circuit as recited in claim 18 wherein T equals eight and said generator polynomial g(x) being further characterized by:

##EQU4## thereby to provide for the correction of eight 10-bit symbols of d(x) which are in error and the capability of detecting whether nine or more 10-bit symbols of y(x) are in error.