AU613701B2 - Apparatus for computing multiplicative inverses in data encoding decoding devices - Google Patents

Description

AU-Ai 29065/89' PCT WORLD INTELLjEUA ROPY Nl'ON INTERNATIONAL APPLICATION PUBLISHED UNDER THE PATENT COOPERATION TREATY (PCT) (51) International Patent Classification 4 (11) International Publication Number: WO 89/ 01660 G06F 7/50 Al (43) International Publication Date: 23 February 1989 (23.02.89) (21) International Application Number: PCT/US88/02160 (22) International Filing Date: (31) Priority Application Number: (32) Priority Date: (33) Priority Country: 24 June 1988 (24.06.88) 067,712 26 June 1987 (26.06.87) (71) Applicant: DIGITAL EQUIPMENT CORPORATION [US/US]; 146 Main Street, Maynard, MA 01754 (US).

(72) Inventor: WENG, Lih-Jyh 6 Bicentennial Drive, Lexington, MA 02173 (US).

(74) Agents: SHEEHAN, Patricia, A. et al.; Nutter, McClennen Fish, One International Place, Boston, MA 02110 (US).

(81) Designated States: AT (European patent), AU, BE (European patent), CH (European patent), DE (European patent), FR (European patent), GB (European patent), IT (European patent), JP, KR, LU (European patent), NL (European patent), SE (European patent).

Published With international'search report.

Before the expiration of the time limitfor amending the claims and to be republished in the event of the receipt of amendments, A.D.J.P. 27 APR 198

AUSTRALIAN

9 MAR 1989 PATENT OFFICE A GALOIS FIELD GF(2QM)".

SDECODER

(54)Title: "APPARATUS FOR DIVIDING ELEMENTS OF (57) Abstract The invention is an apparatus and/or method which enables one to divide two elements, A and B, of GF(221), that is, perform the operation B/A, by finding the multiplicative inverse of the divisor A, and then multiplying the inverse by the numerator, B. The multiplicative inverse, A 1 of A if found by computing a conversion factor. D, (100) and then multi.

plying A by D to convert it to an element C, (102) where C is also an element ofa smaller Galois Field, which is a subfield of GF(221). Specifically, C is equal to A2+1, or A- in the field GF(22I). Next the multiplicative inverse,

C

1 ofC in GF(2% 1 is found by appropriately entering a stored look-up table containing the 2 elements of GF(2M) (104) The multiplicative inverse, Ct, of C is thereafter converted, by multiplying it by the conversion factor D calculated above, to the element of GF(22M) which is the multiplicative inverse, A- of the original divisor, A, (106). The multiplicative in verse, A- 1 of A is then multiplied by B to calculate the quotient, B/A, (108).

,nn ii i 1- I 11 in i [i iiB ii- .i i. i ii~t~ Tr 1 inm ii^ r- iit-i-; ii ii-iiiiii i M.i. M~i WO 9/01660 PCT/US88/02i60 -1- "APPARATUS FOR DIVIDING ELEMENTS OF A GALOIS FIELD GF(2QM) This invention relates to data error correction decoding, data encryption and decryption mechanisms and signal processing systems and more particularly to such systems which employ Galois Field division operations.

Doccription of Prior Art The importance of error correction coding of data in digital computer systems has increased greatly as the density of the data recorded on mass storage media, more particularly magnetic disks, has increased. With higher recording densities, a tiny imperfection in the recording surface of a disk can corrupt a large amount of data. In order to avoid losing that data, error correction codes are employed to, as the name implies, correct the erroneous data.

Before a string of data symbols is recorded on a disk, it is mathematically encoded to form ECC symbols. The ECC symbols are then appended to the data string to form code words data symbols plus ECC symbols and the code words are then stored on the disk. When the stored data is to be accessed from the f disk, the code words containing the data symbols are retrieved from the disk and mathematically decoded. During decoding any errors in the data are detected and, if possible, corrected through manipulation of the ECC symbols [For a detailed description of decoding see Peterson and Weldon, Error Correction Codes, 2d Edition, MIT Press, 1972].

Stored digital data can contain multiple errors. One of the most effective types of ECC used for the correction of multiple errors is a Reed-Solomon code (For a detailed description of Reed-Solomon codes, see Peterson and Weldon, Error Correction Codes]. Error detection and correction techniques for Reed-Solomon ECC's are well known. Id. One such a a M!~I!T WO 89/01660 PCT/US88/02160 -2techniique begins with again encoding the code word data to generate ECC symbols and then comparing these ECC symbols with the ECC symbols in the code word, i.e. the ECC symbols generated by the pre-storage encoding of the data; to detect any errors in the retrieved data. ([For a detailed discussion of this error detection technique, see United States Patent 4,413,339 issued to Riggle and Weng].

if errors are detected in the retrieved data, Galois Field division is usually one of the necessary operations performed in correcting the errors. Galois Field division is a time consuming operation which significantly lengthens the ECC decoding process. The time it takes to perform error correction adversely affects the rate at which data can be retrieved from a storage device.

As the potential for errors in stored data increases with the use of higher recording densities, the effect of slow error correction has a material effect on the average speed with which the data can be retrieved. The increased data retrieval time in turn limits the speed of all computer applications that require the retrieval of the stored data. This speed limitation occurs at a time when computer system technology advances have otherwise made possible faster data transfer operations. Thus, there is a need for a faster apparatus to perform Galois Field division, which will effectively speed up ECC decoding. Such a method would allow computer systems to take advantage of the faster data transfer rates possible with advances in the state of the art computer technology.

The importance of data encryption has increased as the use of computer communications systems, more particularly communication systems involving the transfer of data over telephone lines, has increased. One important encryption WO 89/01660 PCT/US88/02160 -3method involves encoding the data over a Galois Field.

Encrypting the data and later decrypting the data involves division over the Galois Field. The speed with which data can be encrypted and decrypted directly affects the speed with which data can be transferred and processed.

Recently, Galois Fields have been employed in computer controlled signal processing. Specifically, the mathematical transformations of the signal required in signal processing are now often performed over Galois Fields so that the properties of finite, cyclic fields can be utilized. The manipulation of the signal transformations requires Galois Field division.

Also, the use of Galois Field operations in data compression and data expansion apparatus is under development. Any such apparatus will perform division over a Galois Field. Thus the speed with which data operations can be performed in either signal processing apparatus or data compression and/or expansion apparatus is significantly affected by the speed with which Galois Field division can be performed.

i ftuec'r WO 89/01660 PCT/US88/02160 -4- I| iSu-a-ry of Invention -a n b The invention enablescone to divide two elements, A and B, of GF(2 that is, perform the operation B/A, by quickly finding the multiplicative inverse of the divisor A, and then multiplying the inverse by the numerator, B. The multiplicative inverse, A- of A is found by computing a conversion factor, D, and then multiplying A by D to convert it to an element C, where C is also an element of a smaller Galois Field, GF(2 which is a subfield of GF(22 Specifically, C is equal to A 1 or A 2 in the field GF(22). Next, the multiplicative inverse, C 1 of C in GF(2

M

is found quickly by appropriately entering a stored look-up table containing the 2" elements of GF(2").

'The multiplicative inverse, C of C is then converted, by multiplying it by the conversion factor D calculated above, to the element of GF(22M) which is the multiplicative inverse, A of the original divisor, A. The multiplicative inverse, -1 A of A is then multiplied by B to calculate the quotient,

B/A.

In a Galois Field of characteristic two, the operation of raising an element to a power 2 1 that is, computing A 2 has the same degree of simplicity as the multiplication operation.

Thus, while five operations, that is, calculating the conversion factor D; computing C; entering the 2 m look-up table to retrieve C-1; multiplying C-1 by the conversion factor D to produce and multiplying B*A 1 are involved in computing the quotient, B/A, the new division process is faster than conventional Galois Field division. Conventional Galois Field division requires that the multiplicative inverse of the divisor, A, be found in the larger 2" element Galois Field.

If a look-up table is used, the table will contain 22" elements. Finding a multiplicative inverse in the larger table cliptRTITUTS WIMET~=- WO 89/01660 PCT/US88/02160 is significantly slower then performing the same operation in the smaller field.

In the more general case of GF(2 0 the conversion factor D is equal to 2(Q- 1 )M +2(Q-2)M

A

Thus the element C, which is A*D, is equal to z(0-1)M (Q-2)M M A A 12 +2 +2 +11 and the new division process requires Q+3 operations, that is, Q operations for calculating the conversion factor D and Scomputing C and three operations for retrieving C from the 2' element look-up table and calculating A- and By a proper choice of the factors, Q and M, of the exponent of I it is possible to obtain a best or an optimum implementation of the new division process. For example, if the exponent of the field is 12, there are several choices for the factors Q and M. One set of factors will be chosen for the new division process, depending on the speed with which the Q+3 operations can be performed and the speed with which a 2" look-up table can be entered. However, the new division process, with its Q+3 operations, is still faster than conventional Galois Field division operations.

Bricef Deseription of th rawi- The invention will be pointed out with particul y in the appended claims. The above and other adv e of the invention may be better underst y referring to the following description n in conjunction with the accompanyin wings, in which: ,tg r~rr @LIFFT According to one aspect of the present invention there is provided an encoding system for encoding data symbols to form data code words which include redundancy symbols, said system including: means for receiving data symbols, said symbols being elements of Galois Field GF( 2

QM);

J manipulating means for manipulating the data symbols in accordance with an error correcting code, said I manipulating means including: S 10 a Galois Field divider for computing the quotient, B/A, of two elements of Galois Field i 2Q

M

GF(

2 QM), said divider comprising: S a first Galois Field multiplier for converting the divisor, A, into an element C of GF( 2

QM

where C is A and i P equals 2 (Q-1)M 2

(Q-

2 M i 2 the element C being also an

M

j element of a smaller Galois Field GF(2

M

i which is a sub-field of GF(2QM), said multiplier multiplying A by itself P-1 0 times;

M

a 2 element look-up table containing the multiplicative inverses for all elements of the smaller Galois Field S 25 GF(2M), said look-up table being entered using C and providing the multiplicative inverse,

C

1 a second Galois Field multiplier for converting the multiplicative inverse, S 30 C 1 into an element of GF( 2

QM

which is the multiplicative inverse, A said multiplier multiplying C-1 by AS where S equals 2 2 (Q-2)M 2M; and a third Galois Field multiplier for multiplying the element B by the multiplicative inverse A-1; and (ii) means for using the quotients to determine S/ the values of redundancy symbols corresponding -Sa- I Ii to a predetermined number of received data symbols; means for concatenating the received data symbols and the associated redundancy symbols to form data code words; and transmitting means for transmitting the data code words.

According to a further aspect of the present j invention there is provided a decoding system for decoding S 10 data symbols which have been encoded to form data code Swords which include the data symbols and associated i redundancy symbols, the decding system including: means for receiving the data code words; a Galois Field divider for computing the .5 quotient, B/A, of two elements of Galois Field GF(2QM), said divider comprising: a first Galois Field multiplier for converting the divisor, A, into an element C of GF( 2 QM), where C is A and P ecuals 2

(Q-

1 2 (Q-2)M the element C being also an element of a smaller Galois Field GF(2 i which is a sub-field of GF( 2 QM), said Smultiplier multiplying A by itself P-l times; a 2 element look-up table Scontaining the multiplicative inverses for all elements of the smaller Galois Field GF(2M), said look-up table being entered S 30 using C and providing the multiplicative -1.

inverse, C-; a second Galois Field multiplier for converting the multiplicative inverse, C"1, into an element of GF 2

QM

which is the multiplicative inverse, A-1, said multiplier multiplying C-1 by AS where S equals 2 (Q 2 (Q2)M 2M; and 39 a third Galois Field multiplier 1 b.

C.

C

C.

S

for multiplying the element B by the -1 multiplicative inverse A- and means for using the quotients to determine the locations and values of errors in the received data code words; means for correcting the errors; and means for transmitting the error-free data symbols.

The above and other advantage of the invention may be better understood by referring to the following S description taken in conjunction with the accompanying S drawings, in which: LS\ :-c WO 89/01660 PCr/US88/02160 Figure 1 is a flow chart of the steps involved in the operation of the preferred embodiment; and Figure 2 is a functional block diagram of a decoder including means for determining the quotient B/A constructed in accordance with the invention.

Detailed Description It should be understood that all addition and multiplication operations performed during the new division process are Galois Field operations.

With reference to Figures 1 and 2, the new Galois Field division process is performed as part of an encoding or decoding process, where the encoding and/or decoding can be for error correction, data encryption or decryption or signal processing. The Galois Field division is performed by first converting the divisor, A, which is a non-zero element of GF(22 to an element, C, which is also an element of a smaller Galois Field, GF(2M) (Steps 10-12). The conversion is accomplished by first calculating a conversion factor, D, which is equal to:

A

(Step 10) in a Galois Field calculator 100. Calculating A' in a Galois Field of characteristic two is a relatively simple operation.

Next, the conversion factor, D, is multiplied by A (Step 12) in a Galois Field multiplier 102 to produce C: 011"Oorl"I"MI" WO 89/01660 PCT/US88/02160 -7- M M

A

2 *A A 2 +1 C

M

where A 2 1 is an element of the smaller field GF(2x). Thus for every A which is an element of GF(2 2 there is an element C which is also an element of GF(2n). In general, for any Galois Field GF(2QM), that is, a 'Galois Field characterized in part by an exponent, QM, which can be factored, there exists a subf.eld GF(2").

Next the multiplicative inverse, C- 1 of C is determined (Step 14) by entering a look-up table 104 consisting of the 2" elements in GF(2 m The look-up table 104 is entered according to the value of element C, and the unique multiplicative inverse, C- of C in which can be written in the form

A-

1 2"+1 is retrieved from the table.

The multiplicative inverse, C 1 of C in GF(2 is then multiplied by the conversion factor, D, which was calculated earlier in converting A to C, in a Galois Field multiplier 106 (Step 16) to convert it to the element of GF(2 2) which is the multiplicative inverse, A- of A: A- 1 *A(2I

A"

1 The quotient B/A can then be readily obtained by multiplying B*A" (Step 18) in a Galois Field multiplier 108.

The size of the look-up table used in the new division process is 2" elements. If, for example, the larger Galois Field is that is, GF(2 2 5 the look-up table will WO 89/01660 PCT/US88/02160' have bnly 2s or 32 elements. The multiplicative inverse can be quickly obtained from a 32-element table.

Conventional Galois Field division requires selecting the multiplicative inverse of the divisor, A, from a 22 element table. Using GF(21 0 the look-up table would have 210 or 1024 elements. Finding the multiplicative inverse in this 22M element table is materially slower than using the 2 element look-up table.

The foregoing description is limited to a specific embodiment of this invention. It will be apparent, however, that this invention can be practiced in systems having diverse basic construction or using different internal circuitry than is described in the specification with the attainment of some or all of the advantages of this invention. Therefore, it is the object of the appended claims to cover all such variations as come within the true spirit and scope of this invention.

What is claimed as new and desired to be secu Patent of theUnitd i SUBSTITUTE

SHEET

Claims (6)

1. An encoding system for encoding data symbols to form j data code words which include redundancy symbols, said system including: i means for receiving data symbols, said symbols i being elements of Galois Field GF( 2 QM); manipulating means for manipulating the data symbols in accordance with an error correcting code, said manipulating means including: S(i) a Galois Field divider for computing the S quotient, B/A, of two elements of Galois Field "Q M GF( 2 QM), said divider comprising: S* a first Galois Field multiplier for converting the divisor, A, into an element C of GF( 2 QM where C is AP and (Q-2)M P equals 2 Q-) M 2 Q- 2 2 the element C being also an element of a smaller Galois Field GF(2 M which is a sub-field of GF(2QM), said multiplier multiplying A by itself P-l times; a 2 element look-up table containing the multiplicative inverses for all elements of the smaller Galois Field GF(2M), said look-up table being entered using C and providing the multiplicative -1. inverse, C- a second Galois Field multiplier for converting the multiplicative inverse, into an element of GF( 2 QM which is the multiplicative inverse, A 1 said multiplier multiplying C by A where S equals 2 Q 1 M 2 (Q-2) M 2 M; and a third Galois Field multiplier for multiplying the element B by the R multiplicative inverse A- 1 and S39 T (ii) means for using the quotients to determine LS vr -9- the values of redundancy symbols corresponding to a predetermined number of received data symbols; means for concatenating the received data symbols and the associated redundancy symbols to form data code words; and transmitting means for transmitting the data code words.

2. A system according to claim 1 wherein said first Galois Field multiplier includes: the form A said calculating means multiplying A by itself S-1 times; and j* means for multiplying said conversion factor by the element A; S. said calculating means supplying said conversion factor to said second Galois Field multiplier.

3. A decoding system for decoding data symbols which have been encoded to form data code words which include the data symbols and associated redundancy symbols, the S decoding system including: means for receiving the data code words; a Galois Field divider for computing the quotient, B/A, of two elements of Galois Field GF( 2 QM), said divider comprising: a first Galois Field multiplier for converting the divisor, A, into an element C of GF(2QM), where C is A and P equals 2 (Q- 1 2 Q 2 )M 2 the element C being also an element of a smaller Galois Field GF(2 M which is a sub-field of CF( 2 QM), said multiplier multiplying A by itself P-l times; a 2 M element look-up table containing the multiplicative inverses for all elements of the smaller Galois Field GF(2M), said look-up table being entered 39/T/ using C and providing the multiplicative P *9S P. P PPf P. P -1 inverse, C- a second Galois Field multiplier for converting the multiplicative inverse, C- 1 into an element of GF (2 Q which is the multiplicative inverse, A-1, said multiplier multiplying C 1 by A s where S equals 2 2 (Q-2)M 2M; and a third Galois Field multiplier for .ultiplying the element B by the multiplicative inverse A- 1 and means for using the quotients to determine the locations and values of errors in the received data code words; means for correcting the errors; and means for transmitting the error-free data symbols.

4. A system according to claim 3 wherein said first Galois Field multiplier includes: calculating means for calculating a conversion factor of the form AS, said calculating means multiplying A by itself S-1 times; and means for multiplying said conversion factor by the element A; said calculating means supplying said conversion factor to said second Galois Field multiplier.

5. An encoding system according to claim 1 substantially as herein described with reference to the accompanying drawings.

6. A decoding system according to claim 3 substantially as herein described with reference to the accompanying drawings. Dated: 24 May 1991 PHILLIPS ORMONDE FITZPATRICK Attorneys for: DIGITAL EQUIPMENT CORPORATION 39':82I\ ,tfj 1-