GB2324390A

GB2324390A - Decoding Reed-Solomon codes for error correction

Info

Publication number: GB2324390A
Application number: GB9707768A
Authority: GB
Inventors: Wei-Hung Huang
Original assignee: United Microelectronics Corp
Current assignee: United Microelectronics Corp
Priority date: 1997-04-17
Filing date: 1997-04-17
Publication date: 1998-10-21
Anticipated expiration: 2017-04-17
Also published as: GB9716661D0; GB2324390B; GB9707768D0; GB2324391B; GB2324391A

Abstract

A method and an apparatus perform error decoding on digital data transmitted using a Reed-Solomon code. The method decodes the value and location of errors and erasures for digital data, defined as RS m (n,n-4), by a series of arithmetic operations using multiplication-addition (234), inversion (232), shift-logarithm operations (236), and location determinations (238), but without operations to determine powers. Thus, the decoder requires less hardware. By these calculations the values and locations of errors and erasures for digital data are determined correctly and rapidly, avoiding Chien's search.

Description

ERROR DECODING METHOD AND APPARATUS FOR REED-SOLOMON CODES BACKGROUND OF THE INVENTION Field of the Invention The invention relates generally to a method and an apparatus for performing error decoding of digital data using Reed-Solomon codes. More particularly, the invention relates to a method and an apparatus for decoding the value and location of error words in digital data using Reed-Solomon codes by a series of product sum arithmetic operations, without a calculation of power.

Technical Background Reed-Solomon (hereinbelow, abbreviated as RS) codes are very important in communication for the preservation of data, operating to continuously correct errors in bytes of digital data.

There are generally two different circuit types in the design of RS code decoders.

One is the pipe-line type and the other is the processor-based program control type.

Pipe-line type RS code decoders are typically used in systems that must work at highspeed and in real-time. Correspondingly, the circuitry for the pipe-line type RS code decoder is complex and expensive which is wasteful of resources and money in a system demanding only moderate speed. The RS code decoder of the processor-based program control type performs repeated calculations using an arithmetic logic unit (ALU), under the program control, which reduces the complexity and cost of the circuit of the RS code decoder. There are apparent differences between the pipe-line type and the processorbased program control type RS code decoders.

For RS codes, the calculations for decoding, when there are more than two errors among the transmitted words, are too complex to appropriately use the processor-based program control type of RS code decoder. However, the conventional system of this type for decoding two error words is used and is illustrated hereinafter: An RS code can be defined as RSm(n, k). The input terminal of the RS code decoder receives n symbols, with each symbol consisting of m bits. After error correction, k data symbols are transmitted from the decoder.

RSm(n,k) can be defined as: m the number of bits of each imported symbol; nS(2m-1) the code length of the RS codes; k the number of information words in RS codes; d= n-k+l the Hamming distance in RS codes; and

the maximum number of error words decoded by the decoder.

Therefore, in an RS code decoder which can detect two error words, t=2, d=2t+l=5, k-n+l-5=nX, and RSm(n,k) can be modified as RSm(n,n-4). For RSm(n,n-4) codes, the generating polynomial can be represented as: g(x)=(x-1)(x-α)(x-α)(x-α ) It may be assumed that in the n symbols there are v error words which are located at ll, 12, 13, 14 , lv and whose values are respectively Y1,Y2,Y3,Y4,............, YV. In order to explain the principle of decoding the RS codes hereinafter, if the location li is known before the n symbols are decoded, the error word is defined as an 'erasure', and if the location Ii is unknown before the n symbols are decoded, the error word is defined as a error.

With RS codes having a Hamming distance (d=5), p error words and q erasure words can be detected ifp and q satisfy the following formula: 2p+q < d-l=5-1=4 (I) And if Xi=αJi 1#i#v, then the syndromes are:

that is, the syndromes So, Si, S2, S3 can be obtained with the respective combinations ofXj and Yj where: v: the number of erasure words and error words, v-p+q, p: the number of error words, q: the number of erasure words, Xj: Xj=αJj and lj is the location of the jth error Sk: syndrome, Yj: jth error value, and d: Hamming distance of the code.

According to several U.S. patents, for example, Ozaki (4,852,099) and Okamoto et al. (4,677,622), we know the following: If p=0 and q=4, then:

If q=3 and p=O, then: S2 + (X1 + X2)S1 + X1X2S0 Y3 = (4) (X3 + X1)(X3 + X2) If q=2 and p=0, then: S1 + X1S0 Y2 = (5) X2 + X1 If q=l and p=O, then: 4 = S0 (6) When the number of erasure words q is from one to three, the number of error words p that can be accommodated becomes zero or one in accordance with Equation (1).

To detemline whether the actual value of p is zero, or one, or larger than one (in the situation that p is larger than one, the decoder cannot detect any error word from n symbols), reference must be made to a set of Forney Syndromes.

If q is equal to one, then the Forney Syndromes are defined as: T1=S1+S0X1 T2=S2+S1X2 (7) T3=S3+S2X3 If q is equal to two, then the Forney Syndromes are defined as: T1=S2+(X1+X2)S1+X1X2S0 (8) T2=S3+(X1+X2)S2+X1X2S1 If q is equal to three, then the Forney Syndromes are defined as: T1=S3+(X1+X2+X3)S2+(X1X2+X2X3+X3X1)S1+X1X2X3S0 (9) In the condition that q is equal to one or two, the value of p can be determined according to the rules below: (1) If the Forney Syndromes are all equal to zero, the value of p is zero.

(2) If the Forney Syndromes are all not equal to zero, the value of p is one.

(3) If at least one of the Forney Syndromes is equal to zero and at least one of the Forney Syndromes is not equal to zero, the decoder cannot correct any error word from n symbols.

Further, when q is equal to three, if the Forney Syndrome is equal to zero, then the decoder can correct error words from n symbols. When q is equal to three, and if the Forney Syndrome is not equal to zero, then the decoder cannot correct errors from n symbots.

When p is equal to one, the locations of unknown error words can be determined with the equations below: if q=1 and p=1, then S2 + S1X1 X2 = (10) S1 + S0X1 if q=2 and p=l, then

From the above Equations (3) to (11), the value of p can be determined only when the value of q is not equal to zero. In the situation that q is equal to zero, the value of p is determined by evaluating the following quantities: SA = S1S1+S0S2 SB = S2S2+S,S3 (12) SC = S0S3+S1S2 (1) If the three quantities are all equal to zero, then p is equal to one.

(2) If three quantities are all not equal to zero, then p is equal to two.

(3) If one or more of the quantities are equal to zero, and one or more of the quantities are not equal to zero, the decoder cannot correct the errors in n symbols.

In situation (1), that is q=O and p=l: X, = S' (13) So In situation (2), that is q=O and p=2:

From Equation (3) to Equation (13), X; and Yi can be determined by three types of calculations, such as adding, multiplying and inverting, except in the case of (p,q)=(2,0).

( When q is not equal to zero, under the conditionthat i < q , Xi can be obtained while the n symbols are being imported, and the values of Yi and 1; can be obtained with the additional calculation of a logarithm.) Under the condition that p=2 and q=O, there are several ways to solve for the values of Xi and X2 with Equation (14) and Equation (15). For example, the conventional method is to perform a "Chien's Search". In the method using a Chien's Search, at first, Xis set to a predetermined value, and the value of X,+X2 is set as SUM. Thus, X2=SUM-X, is obtained from Equation (14). Thereafter, X, and X2 are substituted in Equation (15). If the value of the left side of the equation is not equal to the value of the right side, Xis set to another predetermined value, referring to the former value. By repeatedly setting the value of X1, the value of X2 can be obtained. But the speed of decoding is very slow for several repetitions of the calculation in which different values of X, and X2 are substituted in the processor-based program control decoder. The method therefore has a negative influence on the whole system, and is not practicable for calculating the actual values of X, and X2.

SUMMARY OF THE INVENTION It is therefore an object of the invention to provide a method for performing error decoding of digital data using a Reed-Solomon code, that is useful to correctly and rapidly decode the values and locations of two errors in digital data defined as RSm(fl"1 4) of the Reed-Solomon code, by a series of product sum arithmetic operations, using addition, multiplication, division, shift-logarithm, and location.

It is another object of the invention to provide an apparatus for performing such a method of decoding.

This is accomplished in accordance with the invention. by providing an error decoding method for decoding syndrome signals using Reed-Solomon codes, the Reed Solomon codes being defined as RSm(n,k), wherein m is a number of bits of each syndrome, n is a length of Reed-Solomon codes, and k is a length of information words in the Reed-Solomon codes, wherein the Reed-Solomon codes comprise p error words and q erasure words, n1(2"-1) and 2p+qSd- l, and wherein d is the Hamming distance. The error decoding method includes decoding the error words and the erasure words existing in every syndrome signal and determining the numbers p and q.

Two equations X,2= a24 and X22= a212 are evaluated for a calculation of location when there are two error words and no erasure words in every syndrome signal, wherein the syndrome signals having two errors are defined as:

wherein X1=αl1, X2=αl2, l1 and l2 define the locations of two error words, and Y and Y2 are values of the two error words: l1 and 12 are evaluated by a calculation of the logarithm of a2l1 and a212 and a calculation of shifting; and Y1and Y2 are obtained by a series of product/sum arithmetic calculations according to X1=α2l1, X2=α2l2.

In the above-mentioned error decoding method, when there are no erasure words in syndrome signals, the number p of error words can be determined by the three quantities..

S,SI + S0S2, S2S2 + S1S3 and S0S3 + S,S2 When none of the three quantities are equal two 0, thenp is equal to 1. When all three quantities are equal to 0, then p is equalto 1. When at least one of the three quantities is equal to 0 and at least one of the three quantities are not equal to 0, then p is not determinable.

Furthermore, in the above-mentioned error decoding method, the calculation of location is performed by evaluating the following equation: αi+ α#αi 1 #i #2m-1 When there are two error words and no erasure words in every syndrome signal, then: S0S3 + S1S2 X1 + X2 = S1S1 + S0S2 S2S2 + S1S3 X1 . X2 = S1S1 + S0S2 Therefore, locations l1 and l2 of the error words can be determined by calculation of location according to the following equations: 2 = a X22 = a212 and 1 and 12 determined as follows: if not shift ai ifshift and i is even 2 i+(2m - 1) i - 1 = 2m-1 + if shit and i is odd 2 2 Furthermore, the error decoding method according to the invention can be executed by providing an apparatus for decoding error words and erasure words existing in transmitted word signals, using Reed-Solomon codes provided from an external system.

Such an apparatus would include a syndrome generator, to be coupledto the external. system, for receiving the transmitted word signals therefrom and generating syndrome signals.

An erasure counter is to be coupled to the external system, for receiving the erasure words. An erasure counting value stored in the erasure counter is incremented by one each time that an erasure word is detected.

Registers are coupled to the syndrome generator, and to be coupled to the external system, for receiving the syndrome signals and the erasure words, and for storing the syndrome signals and the erasure words at predetermined locations. An arithmetic logic circuit (ALU) is coupled to the registers for reading the syndrome signals and the erasure words stored in the registers, and performing a predetermined operation in order to provide a feedback signal to the registers, and generate a deductive signal.

A comparator is coupled to the arithmetic logic circuit for receiving the deductive signal, performing a comparing operation, and then outputting a comparison signal. A program controller is coupled to the comparator and the erasure counter, for receiving the comparison signal from the comparator, reading the erasure counting value stored in the erasure counter, and outputting a program control signal. A program counter is coupled to the program controller for receiving the program control signal, in response to which it generates a program counting signal, and also outputs a feedback signal to the program controller. In response to receipt of the feedback signal, the program controller changes the program control signal.

An interpreter is coupled to the program counter to receive the program counting signal, and in response outputs a register address signal and an ALU enable signal. The arithmetic logic circuit (ALU) reads the syndrome signals and also reads the erasure words stored in the registers, according to the register address signal, in order to decode the error words in the transmitted word signals.

Moreover, the arithmetic logic circuit includes an inverter, a multiplier-adder circuit, a shift-logarithm circuit for performing shift and logarithm calculations and a locator for performing a location calculation. The inverter performs an inverting calculation on the syndrome signals and the multiplier-adder circuit performs an add calculation and a multiplication calculation on the syndrome signals. The shift-logarithm circuit performs a logarithmic calculation of the syndrome signals, and a shift calculation on the result of the logarithmic calculation, according to the ALU enable signal. The locator performs a calculation of location.

The method is executed selectively in the arithmetic logic circuit by operating the inverter, the multiplier-adder circuit, the shift-logarithm circuit, and the locator, in order to generate the feedback signal and the deductive signal.

BRIEF DESCRIPTION OF THE DRAWINGS Other objects, features, and advantages of the invention will become apparent from the following detailed description of the preferred but non-limiting embodiment, with reference to the accompanying drawings in which: FIG. 1 is a circuit block diagram of the preferred embodiment of decoding apparatus according to invention; FIG. 2 is a circuit block diagram of the shift-logarithm circuit in the preferred embodiment; and FIG. 3 is a flow chart ofthe preferred embodiment of decoding method according to invention.

DETAILED DESCRIPTION OF THE INVENTION Referring to FIG. 1, a decoding apparatus 200 according to the preferred embodiment of the invention comprises syndrome generator 210, registers 220, an arithmetic logic circuit 230, an interpreter 240, a comparator 250, a program counter 260, a program controller 270, and an erasure counter 280. The arithmetic logic circuit 230 includes an inverter t32, a multiplier-adder circuit 234, a shift-logarithm circuit 236, and a locator circuit 238.

When an external system 100 begins to transmit signals, transmitted word signals are transmitted to the syndrome generator 210 through a line 10, and erasure word signals are transmitted to the registers 220 and the erasure counter 280 through a line 32. The syndrome generator 210 generates syndrome signals from the transmitted word signals according to the format of RS codes. Syndrome signals are output to the registers 220 through a line 12. The registers 220 temporarily store every syndrome signal and erasure word signal in corresponding addresses in predetermined registers therein. Each time that the erasure counter 280 receives an erasure word signal via the line 32, an erasure counting value stored in the erasure counter 280 is incremented by one in order to count the sum of all erasure word signals. The erasure counting value is transmitted to the program controller 270 through a line 34.

When the apparatus 200 is enabled, the arithmetic logic circuit 230 receives syndrome signals and erasure word signals, which are stored in the registers 220, through the respective line 13, line 14, and line 15. After a series of selective calculations in the arithmetic logic circuit 230, a feedback signal is output to the registers 220 on line 16 where it is stored at a predetermined location. At the same time a deductive signal is output to the comparator 250 for a comparison calculation. Thereafter, a comparison signal is output to a program controller 270 on line 22.

Based on the output of the comparator 250, the erasure counting value from the erasure counter 280, and the program counting value stored in program counter 260, the program controller 270 determines whether the program address should jump or not, and if it should, to which address it should jump. Based on the determination, a program control signal is output from the program controller 270 to the program counter 260 through a line 28. A program counting signal is outputted to the interpreter 240 through a line 24 when the program counter 260 receives the program control signal. At the same time, the program counter 260 changes the program counting value, and then outputs the modified value as feedback to the program controller 270 on line 26 in order serve as a reference for the shifting of the program controller 270.

The interpreter 240 outputs register address signals on line 20 and an enable signal on line 30 respectively to the registers 220 and the arithmetic logic circuit 230 while receiving the program counting signal from the program counter 260. The arithmetic logic circuit 230 reads the values stored in correspondence with the register address signal, while receiving the enable signal. By the deductive function performed in the arithmetic logic circuit 230, the value and location of error words existing in transmitted word signals can be deterrnined.

For the purpose of performing the deductive function, the. arithmetic logic circuit 230 comprises the inverter 232, the multiplier-adder circuit 234, the shift-logarithm circuit 236 and the locator 238. The inverter 232 performs an inverting calculation on the ALU input value. The multiplier-adder circuit 234 performs an addition and multiplication calculation on the ALU input values. The shift-logarithm circuit 236 performs a logarithmic calculation on the ALU input values, and performs a shift calculation on the ALU input values based on the result of logarithm calculation, according to the enable signal. The locator 238, performs a calculation of location.

Referring to FIG. 2, the shift-logarithm circuit 236 includes a logarithm calculator 410 and a multiplexer 420. When calculating data, for example, ALU input value is transmitted to the logarithm calculator 410 through a line 402, and shifted syndrome signals and unshifted syndrome signals are respectively output therefrom to the multiplexer 420 as, for example, b0,b2m-1,b2m-2,......,b1 and b2m-1,b2m-2,.. ...,b0 through lines 412 and 414. When the multiplexer 420 receives these signals and is enabled by the enable signal (shift) through line 422, the multiplexer 420 outputs the calculation result through line 424. The shift-logarithm performed in the shift-logarithm circuit 236, is specified below: i ifnotshift α i/2 if shift and i is even (16) i+(2m-1) i-1 = 2m-1+ if shift and i is odd 2 2 The calculating method of location is: αi+α-1

αi 1 # i # 2m-1 (17) When word signals are transmitted, the apparatus 200 for decoding error words and erasure words performs some initial actions. That means, at first, the values stored in two of the registers 220 are respectively set as an and 6', and these two registers are defined as r, and r2. Therefore, the value of register r1 is a" and the value of register r2 is a''. When the first transmitted word signal is transmitted to the apparatus 200, the value of register r, is multiplied by α-1 ,and then the value of register r, is replaced with r1xa'.

That means that if the syndrome signal is the Lth signal in transmitted data, the value of register rl is set to a" xa-L Whenever the erasure word signal is not logic "0", the value of r, is saved in some proper register of the registers 220. Then the definition Xj=(Xli can be executed without the calculation of multi-powers, i.e., powers higher than 2. This procedure can reduce the complexity of the circuit.

Previously, an RS decoder was able to correct two errors with a processor-based program control structure, as noted above. However, a processor-based program control structure is not practical because the calculating formulas become very complex when an RS decoder must solve more than two errors. Therefore, a decoder which can correct two errors existing in RS codes accords with the real demands.

Errors and erasures can be corrected by the above Equations (3) to (10) when (p,q)=(0,l), (0,2), (0,3), (0,4), (1,1), or (1,2), but not for the situation p=2 and q=0. The errors also can be determined by Equation (13) when (p,q)=(l,0). The value of (Xj,Y;) can be determined by calculations above-mentioned of addition, subtraction and inversion and the value of location li of Y, can also be determined.

The Equations (14) and (15) can also be solved when p=2 and q=0:

According to conventional calculating methods, for example, a Chien Search is conducted, in which the calculating formulas are very complex and time consuming. The invention can decode the errors with the new calculating methods of location and shiftlogarithm. Equation (14) divided by Equation (15) provides an Equation (17) as follows: (X1 + X2) X1 X2 = + = α(l1-l2) + α-(l1-l2) (18) X1 . X2 X2 X1 2 1 The calculating method of location is: αi + α-i # αi 1 # i # 2m-1 (19) The following equation is obtained by the method of location: X1 αl1-l2 = 1 # (l1-l2)mod(2m-1) # 2m-1 (19) X2 Multiplying Equation (15) by Equation (19) provides the following equation: X1.X2.X2/X1 = X2 = α2/2 In the same way as by the shift-logarithm calculation of Equation (16), the value of l2 is obtained.

With this method the values of X12 and X22 are obtained. In order to solve for the value of Y2 by Equation (5), it is necessary to calculate according to the following: i # αi in order to find out the values of X1 and X2.

Equation (5) is redefined in order to avoid the calculation of αi from i and solve for the value Y2: S2 + X1 .S0 Y2 = (20) X1 + X2 Referring to the flow chart ofFIG.3, when all transmitted words have been input into the decoder, if all syndrome signals are zero as determined in step 310, this of course means that the transmitted words have no error words, and in the following step, step 340, the decoding action is terminated. If all syndrome signals are not zero, then in a step 312 the erasure counting value stored in the erasure counter 280 is tested. If the erasure counting value is 1, 2 or 3, the flow branches to the left where a determination is made as to whether there is any error word or not using the Forney Syndrome formulas separately according to step 314, step 316, or step 318, respectively.

For example, in step 318, when the erasure counting value is equal to l, the Fomey Syndrome formulas of Equation (7) are evaluated. In step 316, when the erasure counting value is equal to 2, the Fomey Syndrome formulas of Equation (8) are evaluated. In step 314, when erasure counting value is equal to 3, the Forney Syndrome formula of Equation (9) is evaluated. After one of the above-mentioned steps, the values ofXi, such as X3 in step 320 and X2 in step 322 are obtained.

When the erasure counting value is equal to 4, then in step 324 the value of Y4 is obtained according to Equation (3) and the values of S2 ,i ,and S0 are updated according to the value of Y4. In step 326 the value of Y3 is directly obtained according to Equation (4), and based on this value of Y3 the values of S2 and S0 are updated. And then, in step 328, the value of Y2 is obtained by Equation (20), and based on this value of Y2 the value of So is updated. Then in step 330, the value of Y, is obtained with the Equation (6).

The method of calculating the value of Yi is initiated by substitution of the maximum value 'i'. The calculated value Yj(m,", may be taken as a known error word and that error word may be regarded as removed from the transmitted words. New values of Si are obtained by a method supplemented by Equation (2).

For example, if the value of Y4 is known, then the new values of Si are: So # So + Sl t 5o + Y4 X4 S2 # S2 + Y4X4 Every value of Y can be obtained corresponding to every X;.

In the same way, when the erasure counting value is equal to 3, then when in step 314 the Fomey Syndrome T, according to Equation (9) is computed and checked, if the result is not zero, which means that the errors cannot be decoded, this causes the termination of decoding in step 342. If the Syndrome T, is zero, then flow branches from decision step 314 and the values of Y1, Y2, and Y3 are computed separately in step 326, step 328, and step 330 as already described.

When the erasure counting value is equal to 2, then in decision step 316 the values of the Fomey Syndromes T, and T2are computed and checked according to Equation (8). If T1 orT2, but not both, are equal to zero, the flow is terminated in step 342. If T, and T2 are both equal to zero, then the process flow branches and the values of Y1 and Y2 are obtained separately in step 328 and step 330, as previously described.

When the erasure counting value is equal to 1, decision step 318 is performed to compute and check Fomey Syndrome Equations (7) forT1, T2andT3. If at least one, but not all of T1, T2 and T3 are equal to zero, decoding cannot be performed, causing the termination of decoding through step 342. If Tl, T2 and T3 are all equal to zero, then the value of Y, is obtained in step 330 as previously described.

When the values Y according to the series of steps above described are determined, the corresponding locations li corresponding to Yi (the values of Ii are obtained by the shift-logarithm operation on X; as described above). In the following step 332, ajudgment is made as to whether li is in the correct range or not. Because the sum of the inputted syndrome signals is n, the value of lj must be in the range from 0 to n-l. lf l is outside of the correct range decoding is not possible and is stopped at block 342.

When the erasure counting value is equal to 0, decoding can be completed in a manner similar to the steps described hereinbefore. Flow branches from decision block 312 to step 334 where the number of error words is computed according to Equation (12).

When some, i.e., one or more, but not all of the quantities computed in equation (12), are equal to zero, the decoding is insolvable and the decoding is terminated at block 340.

When none of the quantities computed with Equation (12) are equal to zero, that is p=2, the values oft12 and X22 according to equations 14-19, and then Y2are computed and Sois updated, in steps 336 and 328 respectively. When the quantities computed with equation (12) are all equal to zero, that is p=l, the value of X1 is computed in step 338, and the value of Y, is computed in step 330. The test in step 332 is then performed to check for the correct range, as previously described.

In conclusion, the invention provides a new method of decoding of RS codes whose Hamming distance is equal to 5, by using the program to control sharing of information in the processing-based structure, and by using the calculation of location to avoid the redundant calculation of a Chien's Search. It is important that the arithmetic logic circuit thus becomes very simple without the need for the calculation of powers.

Only the following four kinds of calculations are required: Inverting calculation: a -4 - a Multiplication-addition calculation: a, b, o a b + c Shift-logarithm calculation: referring to Equation (16) Location calculation: referring to Equation (17) With regard to a shortened RS code, n # 2, the hardware of the shift-logarithm circuit and locator in the arithmetic logic circuit 230 can be simplified even further. At first the shift-logarithm circuit can calculate the value of li using & and Xj2 This is meaningful only when li is in the range from 0 to n-l, and decoding cannot be completed when li is outside of the range. Therefore, the logarithm table (not shown) in the shiftlogarithm circuit 236 of FIG. 1 corresponds only from 0 to 21-'. If the inputted value is outside of the valid range, the shift-logarithm circuit outputs an error flag.

Regarding the locator 238, because (l, - 12) is in the range from 1 to n-l (when li > 12), according to Equation (17), the output value of the locator 238 must be in the range from 1 to 2m-'. Therefore, the calculation of location can be executed in the range from I to Max (2m.i, n-l). If the value is outside of this range, an error flag is output to show that the decoder cannot correct any error words, that is, the code is insolvable.

Therefore, in the preferred embodiment with the decoder detecting RS codes whose Hamming distance is 5, the calculation of powers is not required. This permits a simplification of the arithmetic logic circuit. It does not have to repeatedly execute complex calculations when there are two error words and there is no erasure word.

Therefore, the decoding speed is increased. Moreover, the invention provides a great improvement in fields concerning the decoding of RS codes.

While the invention has been described by way of example and in terms of a preferred embodiment, it is to be understood that the invention is not limited to the disclosed embodiment. To the contrary, it is intended to cover various modifications and similar arrangements. Therefore, the scope of the appended claims should be accorded the broadest interpretation so as to encompass all such modifications and similar arrangements.

Claims

1. An error decoding method for decoding syndrome signals using Reed-Solomon codes, the Reed-Solomon codes being defined as RSm(n,k), wherein m is a number of bits of each syndrome signal, n is a length of the Reed-Solomon codes, k is a length of information words in the Reed-Solomon codes, and the Reed-Solomon codes comprise p error words and q erasure words, and wherein nS(2m-l) and 2p+q#d-1, and d is a Hamming distance of the Reed-Solomon codes, the error decoding method comprising the steps of: (a) decoding the syndromes, obtaining therefrom the error words and the erasure words existing in each of the syndromes signal and determining the number of p and q for each of the syndromes; (b) obtaining first and second error locations from the equations X,2= a2'' and X22= a22 by a calculation of location when there are two error words and no erasure words in every one of the syndromes, wherein the syndromes having two errors are defined as:

wherein X1=αl1 , X2=αl2, land 12 define the locations of the two error words, I; and Y2 are values of the two error words; (c) obtaining 1 and 12 according to a2' and a212 by performing logarithm and shift operations; and (d) obtaining t and Y2 by performing a series of product sum arithmetic operations according to X2,= a24 , X22= a212

2. The error decoding method as claimed in claim 1, further comprising: determining values SA, SB and 5c by evaluating the following three equalities when there are no erasure words in the syndromes: SA = S1S1 + S0S1 SB = S2S2 + S1S3 SC = S3S3 + S1S2 and then: determining p to be equal to 1 when SA, SB, SC are all equal to 0, determining p to be equal to 2 when none of SA, SB, SC are equal to 0, and determining that p is not determinable when only one or two of SA, SB, SC are equal to 0.

3. The error decoding method as claimed in claim 1, wherein step (b) includes calculating location according to the following: αi+α-i # i 1 # i # 2m-1

4. The error decoding method as claimed in claim 3, wherein when p=2 and q=0, step (b) includes evaluating:

5. The error decoding method as claimed in claim 4, wherein step (c) includes obtaining 1 and 12 to determine the locations of the error words by evaluating the following: X1 = α2l1 X2 = α2l2 6. The error decoding method as claimed in claim 5, wherein step (c) further includes evaluating the following: if not shift α # i/2 if shift and i is even i + (2m - 1) i - 1 =2m-1 + if shift and i is odd.

2 2

7. The error decoding method as claimed in claim 6, wherein step (d) includes obtaining the values of the two error words Y, and Y2 by evaluating the following: S2 + X1 . S0 X1 + X2 S0 # S0 + Y1 Y1 = S0 8. An apparatus for decoding error words and erasure words existing in transmitted word signals using Reed-Solomon codes from an external system, comprising: (a) a syndrome generator, for receiving the transmitted word signals from the external system and generating syndrome signals; (b) an erasure counter for receiving the erasure words from the external system and incrementing by one an erasure counting value stored in the erasure counter each time an erasure word is received; (c) register means, coupled to the syndrome generator for receiving syndrome signals from the syndrome generator and having an input for receiving the erasure bits from the external system, for storing the syndrome signals and the erasure bits in predetermined locations thereof; (d) arithmetic logic means, coupled to the register means, for reading the values stored in the registers means, performing a predetermined operation to generate a first feedback signal and a deductive signal, and providing the first feedback signal to the register means; (e) a comparator, coupled to the arithmetic logic means, for receiving the deductive signal and performing a comparison operation to generate a comparative signal; (f) a program controller, coupled to the comparator and the erasure counter,: for receiving the comparative signal from the comparator and reading the erasure counting- value stored in the erasure counter, and for outputting a program control signal; (g) a program counter, coupled to the program controller, and responsive to the program control signal, for generating a program counting signal and a second feedback signal, and outputting the second feedback signal to the program controller, whereby the program controller is responsive to the second feedback signal to selectively change the program control signal; and (h)an interpreter, coupled to the program counter and responsive to the program counting signal, for generating a register address signal and an enable signal, the interpreter being coupled to the register means for providing the register address signal thereto, and coupled to the arithmetic logic means for providing the enable signal thereto, the arithmetic logic means being responsive to the enable signal for reading the syndrome signals and the erasure words stored in the register means, according to the register address signal, in order to decode the error words in the transmitted word signals.

9. The apparatus as claimed in claim 8, wherein the arithmetic logic means includes: an inverter for performing an inverting calculation on the syndrome signals; multiplier-adder means, for performing an add calculation and a multiplication calculation on the syndrome signals; shift-logarithm means, for performing a logarithmic calculation on the syndrome signal, and for performing a shift calculation on a result of the logarithmic calculation, according to the enable signal; and locator means for performing a calculation of location.

10. The apparatus as claimed in claim 8, wherein the predetermined operation is executed selectively by operating the inverter, the multiplier-adder means, the shiftlogarithm means, and the locator means, in the arithmetic logic means in order to generate the first feedback algorithm signal and the deductive signal.

11. An error decoding method for decoding syndrome signals using Reed-Solomon codes substantially as hereinbefore described with reference to and/or as illustrated in any one of or any combination of the accompanying drawings.

12. Apparatus for decoding error words and erasure words existing in transmitted word signals using Reed-Solomon codes substantially as hereinbefore described with reference to and/or as illustrated in any one of or any combination of the accompanying drawings.

Amendments to the claims have been filed as follows @. An error decoding method for decoding syndrome signals using Reed- Solomon codes, the Reed-Solomon codes being defined as RSm(n,k), wherein m is a number of bits of each syndrome signal, n is a length of the Reed-Solomon codes, k is a length of information words in the Reed-Solomon codes, and the Reed-Solomon codes comprise p error words and q erasure words, and wherein n < (2m-1) and 2p+q#d-1, and d is a Harming distance of the Reed-Solomon codes, the error decoding method comprising the steps ot: (a) decoding the syndromes, obtaining therefrom the error words and the erasure words existing in each of the syndromes signal and determining the number of p and q for each of the syndromes; (b)obtaining first and second error locations from the equations X,Z= a211 and X2=α2l2 by a calculation of location when there are two error words and no erasure words in every one of the syndromes; wherein the syndromes having two errors are defined as:

wherein X1=αl1, X2=αl2, l1 and l2 define the locations of the two error words, Y; and Y2 are values of the two error words; (c) obtaining l, and l2 according to a211 and a21. by performing logarithm and shift operations; and (d) obtaining Y1 and Y2 by performing a series of product sum arithmetic operations according to X1=α2l1, X2=α2l2 2. The error decoding method as claimed in claim 1, further comprising: determining values SAÇ S9 and SC by evaluating the following three equalities when there are no erasure words in the syndromes: SA = S1S1 + S0S1 SB = S2S2 + S1S3 SC = S3S3 + S1S2 and then: determining p to be equal to 1 when SA, SB, SC are all equal to 0, determining p to be equal to 2 when none of SA, S3, SC are equal to 0, and determining that p is not determinable when only one or two of SA, SB, SC are equal to 0.

3. The error decoding method as claimed in claim 1, wherein step (b) includes calculating location according to the following: α@+α@ # 1 1 # i # 2m-1

5. The error decoding method as claimed in claim 4, wherein step (c) includes obtaining l, and 12 to determine the locations of the error words by evaluating the following: X1 = α2l1 X2 = α2l2

6. The error decoding method as claimed in claim 5, wherein step (c) further includes evaluating the following: i if not shift α # i/2 if shift and i is even i + (2m - 1) i - 1 = 2m-1 + if shift and i is odd.

2 - 2 ii s and i is odd.

7. The error decoding method as claimed in claim 6, wherein step (d) includes obtaining the values of the two error words Y1 and Y2 by evaluating the following: S2 + X1 .S0 Y2 = X1 + X2 S0 # S0 + Y1 Y1 = S0

8. An error decoding method for decoding syndrome signals using Reed-Solomon codes substantially as hereinbefore described with reference to andlor as illustrated in any one of or any combination of the accompanying drawings.