US 6,199,188 B1Grant

System for finding roots of degree three and degree four error locator polynomials over GF(2M)

Issue Date:2001-03-06

•15 Claims

•4 Drawing Sheets

Abstract

A system determines the locations of four errors in a code word over GF(2^m), for any m, by transforming a degree-four error locator polynomial σ(x) ultimately into two quadratic equations, finding the solutions of these equations, and from these solutions determining the roots of the error locator polynomial. The system first manipulates the degree-four error locator polynomial into a polynomial θ(y) that has a coefficient of zero for the degree-three term. The system then factors this polynomial into two degree-two factors with four unknown variables. The system expands the factors and represents the coefficients of θ(y) as expressions that include the four unknown variables, and manipulates the expressions to produce a degree-three polynomial with only one of the unknown variables. The system next solves for that variable by finding a root of the degree-three polynomial in GF(2^m) if the field is an even-bit field or in an even-bit extension of GF(2^m) if the field is an odd-bit field. The system then substitutes the root into the expressions for the coefficients of θ(y) and produces a degree-two expression is with the remaining unknown variables. The system finds the roots of this expression, substitutes these values into the factors ofθ(y), and sets the factors equal to zero to produce two quadratic equations. The system then solves the equations to produce the roots ofθ(y), and from these solutions determines the roots of the degree-four error locator polynomial.

Metadata

Assignee

Quantum Corporation

Inventors

Ba-Zhong Shen
Lih-Jyh Weng

Application Information

Application Number:US 09/521,518

Filing Date:2000-03-08

Priority Date:1997-10-07

Art Unit:7

Classifications

IPC:

H03M 1300

Field of Search:

604 96708492714762714761714759714782714784

Patent Drawings (4 sheets)

Description

[0002] This application is a continuation of application Ser. No. 08/946,470, filed Oct. 7, 1997, now abandoned.

Field of the Invention

[0003] This invention relates generally to data processing systems and, more particularly, to a system for decoding and correcting errors in data using an error correction code.

Background of the Invention

[0004] Data stored on magnetic media, such as a magnetic disks, are typically stored in encoded form, so that errors in the stored data can possibly be corrected. The errors may occur, for example, because of inter-symbol interference, a defect in the disk, or noise. As the density of the data stored on the disk increases, more errors are likely, and the system is required to correct greater members of errors. The speed with which the system corrects the errors is important to the overall speed with which the system processes the data.

[0005] Prior to recording, multiple-bit data symbols are encoded using an error correction code (ECC). When the data symbols are retrieved from the disk and demodulated, the ECC is employed to, as the name implies, correct the erroneous data.

[0006] Specifically, before a sting of k data symbols is written to a disk, it is mathematically encoded using an (n, k) ECC to form n-k ECC symbols. The ECC symbols are then appended to the data string to form an n-symbol error correction code word, which is then written to, or stored, on the disk. When the data are read from the disk, the code words containing the data symbols and ECC symbols are retrieved and mathematically decoded. During decoding, 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,2nd Ed. MrT Press, 1972].

[0007] To correct multiple errors in strings of data symbols, the system typically uses an ECC that efficiently and effectively utilizes the various mathematical properties of sets of symbols known as Galois fields. Galois fields are represented “GF (P^m)”, where “P” is a prime number and “m” can be thought of as the number of digits, base “P”, in each element or symbol in the field. P usually has the value 2 in digital computer and disk drive applications and, therefore, m is the number of bits in each symbol. The ECC's commonly used with the Galois Fields are Reed Solomon codes or BCH codes.

[0008] There are essentially four major steps in decoding a corrupted code word of a Reed-Solomon code or a BCH code. The system first determines error syndromes that are based on the results of a manipulation of the ECC symbols. Next, using the error syndromes, the system determines an error locator polynomial which is a polynomial that has the same degree as the number of errors. The then finds the roots of the error locator polynomial and from each root determines the location of an associated error in the code word. Finally, the system finds error values for the error locations. In binary systems, such as digital computers, there is only one possible error value for an error location, and thus, the step of determining error values is trivial.

[0009] The steps of determining the syndromes and finding the error locations are the most time consuming in the error correction process. The invention described herein reduces the time it takes the error correction system to find the locations of either three or four errors. This involves finding the roots of third and fourth degree error locator polynomials.

[0010] In prior systems, the roots of degree-four polynomials are determined by trial and error, or by matrix manipulation or look-up table. The trial and error method is performed by substituting into the polynomial every possible value, i.e., every element of the applicable GF(2^2m) that is associated with a code word location, and for each value evaluating the polynomial. If the polynomial equals zero for a given value, the value is a root. The system continues the trial and error process by substituting a next possible value into the polynomial and determining if that value is a root; and so forth, until either all possible values have been tried or all four roots are determined. This trial and error process, which in an optimized form is commonly known as a Chien Search, is time consuming. Further, the time is unpredictable, since it varies with the locations of the errors in the code words.

[0011] The matrix manipulation and look-up table methods involve transforming a general degree-four polynomial of the form:

[0012] σ₄x⁴+σ₃x³+σ₂x²+σ₁x+σ₀

[0013] into a polynomial of the form:

[0014] x⁴+x²+ax+b.

[0015] If a look-up table is used, the table contains a set of solutions for every possible set of values of a and b. For the ECCs used today, typically over GF(2⁸), the table has 2¹⁶entries, each with 32 bits, i.e., four 8-bit elements. Thus, the table takes up a great deal of storage space and requires a relatively complex addressing scheme.

[0016] A system using the matrix manipulation method produces an 2^m-by-2^mmatrix, by determining f(α), f(α¹), f(α²) . . . fa(.α⁽²⁻¹⁾), where f(x)=x⁴+x²+ax+b. It next manipulates the matrix to determine the null space of the space spanned by the matrix. The roots of the polynomial can then be obtained from the vectors that span the null space. [For a more detailed description see, E. R Berlekamp, Algebraic Coding Theory,McGraw Hill Book Company, 1968]. This method requires bit-by-bit manipulation of an 2m-by-2m matrix and typically requires numerous manipulation operations to transform the matrix. Thus, it is time consuming and computation intensive.

[0017] The inventive system described herein uses the invention discussed in patent application Ser. No. 08/580,294 filed Dec. 28, 1995, now U.S. Pat. No. 5,761,102, which is assigned to a common assignee and incorporated herein by reference. The system determines the roots of the degree-four polynomial by solving two quadradic equations. The system makes use of a well known method of deter the roots of a degree-two polynomial, as descried by Berlekamp in Algebraic Coding Theory.Further, the system makes use of a method of determining the roots of a degree-three polynomial that is described by Van der Horst and Berger in Complete Decoding of Triple-Error-Correcting Binary BCH Codes, IEEE Transactions on Information Theory, Vol. IT-22, pp. 138-147, 1976. Van der Horst and Berger have established that the method works for even-bit fields GF(2^2m) for m=2 and 3. In the above-referenced application it was determined that the method works for at least even-bit fields GF(2^2m) for m=4, 5 and 6. We have determined that the method works for any GF(2^2m), as discussed below. We also provide herein a method to determine the roots of a degree-four polynomial in any odd-bit field GF (2^2m+1), as discussed in more detail below.

[0018] Finding the roots of the degree-three polynomial in accordance with the teachings of Van der Horst and Berger and in accordance with our methods discussed herein requires determining a cubic root of a Galois field element over GF(2^2m) or GF(2^2m+1). We have developed an inventive circuit for relatively quickly and easily finding the cubic roots in the even-bit field GF(2^m), and in the odd-bit field GF(2^2m+1).

Summary of the Invention

[0019] The invention is a system that determines the error locations of four errors in GF(2^m), for any m, by transforming a degree-four error locator polynomial ultimately into two quadratic equations, finding the solutions of these equations, and from these solutions determining the roots of the error locator polynomial.

[0020] More specifically, the system first manipulates the degree-four polynomial:

[0021] σ(x)=σ₄x⁴+σ₃x³+σ₂x²+σ₁x+σ₀ [1]

[0022] into the form:

[0023] θ(y)=y⁴+θ₂y²+θ₁y+θ₀ [2]

[0024] where the θ_i's are combinations of the coefficients of the terms of the error locator polynomial. Note that the coefficient of the y³term is zero. The system then factors θ(y) into:

[0025] θ(y)=(y²+t*y+u)*(y²+v*y+w) [3]

[0026] where “*” represents multiplication, and expands the expression to:

[0027] θ(y)=y⁴+(t+v)y³+(t*v+u+w)y²+(t*w+v*u)y+u*w, [4]

[0028] where t, u, v and w are unknown. Equating the coefficients of expressions 2 and 3, the system determines that:

[0029] t+v=0

[0030] t*v+u*w=t²+u*w=θ₂

[0031] t*w+u*v=t+w+t*u=t*(w+u)=θ₁

[0032] u*w=θ₀. [5]

[0033] The system then produces an equation with the variable t as the only unknown:

[0034] t³θ₂t+θ₁=0 [6]

[0035] This degree-three equation can be solved in an even-bit field GF(2^m), in part, based on Cardan's Rule for real fields, which we have proven can be similarly applied to even-bit Galois Fields. Further, we use the method described by Van der Horst and Berger, all as discussed in more detail below. The degree-three equation can be solved in an odd-bit field GF(2^2m+1), using a method that utilizes an even-bit extension field GF(2^4m+2) of the odd-bit field, as discussed in more detail below.

[0036] Once a root, t₀, of equation 6 is determined, the root is substituted into the expression for θ₁, and that expression and the expression for θ₀are used to from a quadratic equation with u and w as the roots:

[0037] p²+(θ₂+t₀²)p+θ₀=0 [7]

[0038] The system then finds the roots, p₀and p₁, of equation 7, using Berlekamp's method for solving degree-two polynomials. The roots p₀and p₁, which are u and w, respectively, are substituted into polynomial 3 and each term is set to zero, to produce two quadratic equations:

[0039] y²+t₀*y+p₀=0

[0040] y₂+t₀*y+p₁=0 [8]

[0041] The roots of equations 8, which are determined using the Berlekamp method, are also the roots of θ(y), and they can thus be used directly to produce the roots of the error locator polynomial.

[0042] This direct solution is faster than the prior trial and error or the matrix manipulation methods of finding the four roots. Further, it does not require the storage of and entry into the large look-up table that is required in other prior known systems.

[0043] We perform a necessary step of finding cubic roots in GF(2^2m) and GF(2^2m+1) using inventive methods that require either a look-up table that is limited in size or no look-up table at all, as discussed in more detail below.

Brief Description of the Drawings

[0044] For a fuller understanding of the nature and objects of the invention, reference should be had to the following detailed description taken in connection with the accompanying drawings, in which:

[0045] FIG. 1 is a functional block diagram of a decoder that is constructed in accordance with the invention;

[0046] FIG. 2 is a flow chart of the operations of the decoder of FIG. 1 in determining if roots of an equation are in GF(2^2m+1);

[0047] FIGS. 3A and 3B are a flow chart of the operations that the decoder of FIG. 1 performs in determining the locations of four errors;

[0048] FIG. 4 is a functional block diagram of a subsystem for determining the cubic roots of elements of GF(2^m+1);

[0049] FIG. 5 is a functional block diagram of a subsystem for determining the cubic roots of elements of GF(2^2m); and

[0050] FIG. 6 is a functional block diagram of an alternative system for determining the cubic roots of elements of GF(2^2m).

Detailed Description

[0051] The mathematical operations of addition, subtraction, multiplication and division discussed herein are Galois Field operations over the applicable Galois Fields.

[0052] Referring now to FIG. 1, a decoder 10 includes a syndrome generator 12 that operates in a conventional manner to produce an associated set of error syndromes. If the syndromes are all zero, the syndrome generator 12 determines that the code word is error-free. Otherwise, the syndrome generator sends the syndromes to an error locator polynomial generator 14, which produces in a conventional manner from the syndromes an error locator polynomial of degree “e”, where e is the number of errors in the code word. The error locator polynomial generator 14 then sends the error locator polynomial to an error locator processor 16, which selects and implements an appropriate process for determining the roots of the error locator polynomial, and thus, the error locations. FIG. 1 depicts these processes as various blocks or processors, namely, a processor 18 for determining one error location, a processor 20 for determining two error locations, a processor 22 for determining three error locations, a processor 24 for determining four error locations, and a processor 26 for determining more than four error locations.

[0053] The two-, three- and four-error processors 20, 22 and 24 are interconnected. As discussed below, the error locator processor 16 reduces the time it takes to determine the locations of three and four errors by transforming the corresponding error location polynomials ultimately into associated sets of quadratic equations. To do this, the four-error processor 24 makes use of both the three-error processor 22 and the two-error processor 20. Similarly, the three-error processor 22 makes use of the two-error processor 20. While these processes are depicted as occurring in separate, dedicated processors, they may be performed in one or more multi-purpose processors.

[0054] The error locator processor 16 determines the locations of one and two errors in a conventional manner. It determines the locations of three errors over GF(2^2m) in accordance with the method described by Van der Horst and Berger, and it uses a new method, which is based in part on the method described by Van der Horst and Berger, to determine the locations of three errors over GF(2^2m+1). With either method for finding three error locations, the processor 16 uses an inventive circuit to perform a necessary step of determining the cubic root of a Galois Field element. This circuit and the method of determining the three error locations are discussed below with reference to FIGS. 5-7.

[0055] The error locator processor 16 determines the locations of four errors using a process that is described in U.S. Pat. No. 5,761,102 filed Dec. 28, 1995 entitled SYSTEM AND METHOD FOR DETERMINING THE CUBE ROOT OF AN ELEMENT OF A GALOIS FIELD GF(2^2m) which is incorporated herein by reference. We have determined that this process can be used over any Galois Field GF(2^m), as discussed below. The operations of the processor in determining the four error locations is also described below with reference to FIG. 4.

[0056] A. Determining Two Error Locations

[0057] If the code word contains two errors, the error locator polynomial is of the form:

[0058] σ(x)=σ₂x²+σ₁x+σ₀x. [1]

[0059] To find the roots of the polynomial, the system sets the polynomial equal to zero:

[0060] σ(x)=σ₂x²+σ₁x+σ₀x=0 [2]

[0061] and solves the equation for the solutions, x₀and x₁.

[0062] To solve equation 2, the error locator processor 16 transforms the equation into the form:

[0063] y²+y+C=0 [3]

[0064] by changing the variable to:

\begin{matrix} x = \frac{σ_{1}}{σ_{2}} y & [4] \end{matrix}

[0065] such that equation 2 becomes

σ (\frac{σ_{1}}{σ_{2}} y) = \frac{σ_{1}^{2}}{σ_{2}} y^{2} + \frac{σ_{1}^{2}}{σ_{2}} y + σ_{0}

[0066] Then, dividing by

\frac{σ_{2}}{σ_{1}^{2}},

[0067] the system produces the equation:

\begin{matrix} y^{2} + y + \frac{σ_{0} σ_{2}}{σ_{1}^{2}} = 0 & [5] \end{matrix}

[0068] and the C of equation 3 is thus

\frac{σ_{0} σ_{2}}{σ_{1}^{2}} .

[0069] For a solution of equation 5 to exist in GF(2^m), the “traces” of both sides of the equation must be the same. A trace is defined as:

\begin{matrix} Trace (α^{i}) = \sum_{k = 0}^{m - 1} {(α^{i})}^{2^{k}} & [6] \end{matrix}

[0070] The Trace (0) equals 0, and thus, for solutions to exist within the Galois Field,

[0071] Trace [y²+y+C]=0 [7]

[0072] The trace of a sum of terms is equal to the sum of the traces of the individual terms. In GF(2^m), Trace (y²)=Trace (y) and their sum, using Galois Field addition, is zero. Accordingly, equation 5 has a solution in GF(2^m) if Trace (C)=0.

[0073] The general solution to equation 5 is a linear combination of m-1 “base solutions.” A base solution is a solution to equation 3 with C as one of the “base elements” C₁, C₂. . . C_2m−1of the Galois Field. The base elements are those elements of the field that can be linearly combined to produce all of the other elements of the field. These elements may, for example, be the set of 1m-1 elements that each include 1m-1 zeros and a single 1, i.e., αⁱ=000 . . . 1 to αⁱ=100 . . . 00. In such a field, C=00110 . . . 0 is a linear combination of base elements 00100 . . . 0 and 00010 . . . 0. The general solution to equation 5 is the linear combination of the base solutions that correspond to the base elements which are summed to produce

\frac{σ_{0} σ_{2}}{σ_{1}^{2}} .

[0074] In GF(2⁴) which is defined by the generator polynomial x⁴+x+1, each element, αⁱ, that has a trace of zero also has a zero as its most significant bit. Accordingly; the C's of interest in this field are combinations of the base elements α⁰=0001, α¹=0010 and α²=0100. The corresponding base solutions are:

\begin{matrix} {(α^{5})}^{2} + α^{5} = α^{0} & with y_{0} = α^{5} \\ {(α^{9})}^{2} + α^{9} = α^{1} & with y_{0} = α^{9} \\ {(α^{3})}^{2} + α^{3} = α^{2} & with y_{0} = α^{3} \end{matrix}

[0075] Another set of base solutions are:

\begin{matrix} {(α^{10})}^{2} + α^{10} = α^{0} & with y_{1} = α^{10} \\ {(α^{7})}^{2} + α^{7} = α^{1} & with y_{1} = α^{7} \\ {(α^{14})}^{2} + α^{14} = α^{2} & with y_{1} = α^{14} \end{matrix}

[0076] In this Galois Field, the solutions for equation 3 with C equal to α⁸=0101, for example, are determined by decomposing C into:

[0077] C=α⁰+α²=0001+0100

[0078] and summing a corresponding wet of base solutions, for example:

[0079] y₀=α⁵+α³=α¹¹

[0080] to produce a first root and determining a second root as:

[0081] y₁=y₀+α⁰

[0082] Next the system determines the roots of the degree-two error locator polynomial as:

\begin{matrix} x_{0} = y_{0} (\frac{σ_{1}}{σ_{2}}) \\ x_{1} = y_{1} (\frac{σ_{1}}{σ_{2}}) . \end{matrix}

[0083] B. Determining Three Error Locations

[0084] In general, for any degree-three error locator polynomial, σ(x), the system determines error locations by solving:

[0085] σ(x)=σ₃x³+σ₂x²+σ₁x+σ₀=0. [8]

[0086] Accordingly, the system changes variable to

x = y + \frac{σ_{2}}{σ_{3}}

[0087] and equation 1 becomes

\begin{matrix} σ (y + \frac{σ_{2}}{σ_{3}}) = {σ_{3} (y + \frac{σ_{2}}{σ_{3}})}^{3} + {σ_{2} (y + \frac{σ_{2}}{σ_{3}})}^{2} + σ_{1} (y + \frac{σ_{2}}{σ_{3}}) + σ_{0} = 0. & [9] \end{matrix}

[0088] which expands to:

σ_{3} (y^{3} + y^{2} \frac{σ_{2}}{σ_{3}} + {y (\frac{σ_{2}}{σ_{3}})}^{2} + {(\frac{σ_{2}}{σ_{3}})}^{3}) + σ_{2} (y^{2} + {(\frac{σ_{2}}{σ_{3}})}^{2}) + σ_{1} (y + \frac{σ_{2}}{σ_{3}}) + σ_{0} = 0

[0089] and after collecting like terms and summing identical terms to zero the equation becomes:

\begin{matrix} σ_{3} y^{3} + 0 y^{2} + (\frac{σ_{2}^{2}}{σ_{3}} + σ_{1}) y + 0 + \frac{σ_{1} σ_{2}}{σ_{3}} + σ_{0} = 0 & [10] \end{matrix}

[0090] The system then divides equation 10 by σ₃, and the equation becomes:

[0091] y³+αy+b=0 [11]

[0092] with

a = \frac{σ_{2}^{2}}{σ_{3}^{2}} + \frac{σ_{1}}{σ_{3}} and b = \frac{σ_{0}}{σ_{3}} + \frac{σ_{1} σ_{2}}{σ_{3}^{2}},

[0093] and y₀is a solution of equation 11 if and only if

y_{0} + \frac{σ_{2}}{σ_{3}}

[0094] is a solution of equation 8. Accordingly, we can determine the solutions of equation 8 by determining the solutions of equation 11.

[0095] To determine if all of the roots of equation 4 are in GF(2^2m) i.e., an even-bit field, we use Cardan's Rule as applied to even-bit Galois Fields. To determine if all of the roots of equation 4 are in GF(2^2m+1), which is an odd-bit field, we use the properties of an even-bit extension field GF(2^4m+2), as discussed below. If all the roots are in the appropriate Galois Field, the system then finds the solutions of equation 11 using for an even-bit field the methods discussed the co-pending patent application, or if the system is operating over an odd-bit field, using a modification of that method, as discussed in more detail in section B2 below.

[0096] One of the necessary steps in determining the roots of equation 11 is finding cube roots. To do this, the error locator processor 16 preferably uses the subsystem of FIGS. 5-7, as discussed below. The error locator processor 16 may instead determine the cube roots of an element of GF(2^2m) using the operations described in the co-pending patent application.

[0097] B1. Solutions Over An Even-Bit Field GF(2^2m)

[0098] According to Cardan's Rule as applied to even-bit Galois Fields, equation 11 has all three solutions y₁, y₂and y₃in GF(2^2m) if and only if each root y_iis of the form

y_{i} = x_{i} + \frac{a}{x_{i}}

[0099] where x_iε GF(2^2m). This means that y_iis a root of equation 11 if and only if

[0100] x_i²+y_ix_i+a=0 [12]

[0101] has all solutions in GF(2^2m). As discussed in the co-pending patent application, an equation of the form of equation 12 has solutions in GF(2^2m) if the “traces” of both sides of the equation are the same, where a trace is defined as Tr(α)=α+α₂+ . . . +α²⁻¹.

[0102] To determine the trace of the left-hand side of the equation 12, we first define in Step 202 for a polynomial Q(x)=(x−α₁) (x−α₂) (x−α₃) . . . (x−α_n), a function p(Q(x)):

p (Q (x)) = \sum_{1 \leq i \leq j \leq n} \frac{α_{i} α_{j}}{{(α_{i} + α_{j})}^{2}} .

[0103] We then determine if the trace of the functions, i.e., Tr(p), for the left-hand side of equation 12 is the same as the trace of the right-hand side of the equation, i.e., Tr(0), which is equal to zero.

[0104] The polynomial of interest, x_i²+y_ix_i+a=(x−α₁) (x−α₂), with α₁+α₂=y_iand α₁α₂=a, and

p (x_{i}^{2} + y_{i} x_{i} + a) = \frac{α_{1} α_{2}}{{(α_{1} + α_{2})}^{2}} = \frac{a}{y_{i}^{2}} .

[0105] Accordingly, equation 12 has solutions in GF(2^2m) if and only if

\begin{matrix} Tr (\frac{a}{y_{i}^{2}}) = 0. & [13] \end{matrix}

[0106] Further, we determine at equation 11 has all solutions in GF(2^2m) by determining if Tr(p(y³+ay+b))=0, which means that

Tr (\frac{a^{3}}{b^{2}} + 1) = 0.

[0107] Over GF(2^2m), in which Tr(1)=0,

Tr (\frac{a^{3}}{b^{2}}) + Tr (1) = Tr (\frac{a^{3}}{b^{2}}) + 0,

[0108] and thus if all solutions are GF(2^2m),

\begin{matrix} Tr (\frac{a^{3}}{b^{2}}) = 0. & [14] \end{matrix}

[0109] Since all solutions are in GF(2^2m), the system determines the roots of equation 11 and then the roots of equation 8. The system may use the method of determining the roots of degree-three polynomials described in the co-pending application. Specifically, using equation 11

[0110] y³+αy+b=0 [11]

[0111] and if α≠0, the processor 16 changes variable to:

\begin{matrix} y = z + \frac{a}{z} & [15] \end{matrix}

[0112] and equation 4 becomes

\begin{matrix} {(z + \frac{a}{z})}^{3} + a (z + \frac{a}{z}) + b = 0. & [16] \end{matrix}

[0113] It next expands equation 16 and collects like terms to produce:

\begin{matrix} z^{3} + \frac{a}{z} z^{2} + {(\frac{a}{z})}^{2} z + {(\frac{a}{z})}^{3} + az + \frac{a^{2}}{z} + b = z^{3} + az + \frac{a^{2}}{z} + \frac{a^{3}}{z^{3}} + az + \frac{a^{2}}{z} + b = z^{3} + \frac{a^{3}}{z^{3}} + b = 0. & [17] \end{matrix}

[0114] The system next multiplies equation 17 by z³to produce

[0115] (z³)²+bz³+a³=0, [18]

[0116] which is a quadratic equation with z³as the variable. Equation 18 can then be solved using the two-error method discussed in Section A.

[0117] Once one solution, z₀₃, is determined, the system finds the cubic root, z₀, which is a solution to equation 11. The operations of the system to determine the cubic root are described below with reference to FIGS. 5 and 6. The system determines a second solution, z_i:

z_{1} = z_{0} * α^{\frac{2^{2 m - 1}}{3}}

[0118] where the exponent

\frac{2^{2 m} - 1}{3}

[0119] is always an integer, since 2^2m−1 is divisible by 3, and a third solution:

[0120] z₂=z₁+z₀

[0121] The system then uses these solutions to determine the roots x₀, x₁, and x₂of the degree-three error locator polynomial of equation 8. The system thus substitutes the solutions z_iinto the expression for y, and individually adds the results of the substitution to

\frac{σ_{2}}{σ_{3}}

[0122] to produce the roots x₀, x₁and x₂.

[0123] If a=0, the solution to equation 18 is y=b^⅓. To find the cubic root of b the system again uses the circuit depicted in FIG. 5 or 6. Once the cubic root is known, the system determines y₁and y₂as:

y_{1} = y_{0} * α^{\frac{2^{2 m - 1}}{3}} and y_{2} = y_{0} + y_{1},

[0124] and from these values determines the roots x₀, x₁and x₂.

[0125] B2. Solutions Over An Odd-Bit Field GF(2^2m+1)

[0126] When the system is operating over GF(2^2m+1), and equation 11

[0127] y³+αy+b=0 [11]

[0128] has all solutions in GF(2^2m+1), we know that

Tr (\frac{a^{3}}{b^{2}} + 1) = 0

[0129] and thus

Tr (\frac{a^{3}}{b^{2}}) = 1

[0130] since Tr(1)=1 in the odd-bit field. This means that the equation 18

[0131] (z³)²+bz³+a³=0 [18]

[0132] has no roots in GF(2^2m+1). Thus, the error locator processor 16 cannot use the method described above to determine the solutions of equation 11 in the odd-bit field.

[0133] The error locator processor 16 instead determines the roots in an even-bit extension field of the odd-bit field, namely, GF(2^4m+2), which is defined as GF(2^4m+2)=GF(2^2m+1)[x]/(x²+x+1)=(αx+β|α, βεGF(2^2m+1)} with multiplication defined by:

[0134] (α₁x+β₁) (α₂x+β₂) modulo x²+x+1

[0135] such that

[0136] (α₁x+β₁) (α₂x+β₂)=(α₁α₂+α₁β₂+α₂β₁, α₁α₂+β₁β₂).

[0137] A simplified way to express GF(2^4m+1) is

[0138] GF(2^4m+2)={(α,↑)|α,βεGF(2^2m+1)}.

[0139] Further, the odd-bit field GF(2^2m+1)={0+β|(0,β) εGF(2^4m+2)}. This means that we can perform multiplication operations over the extension field by manipulating the 2^2m+1bit elements of GF(2^2m+1) in the pairs that represent elements of GF(2^4m+2).

[0140] Using the extension field, we can show that since the equation:

[0141] y³+ay+b=0 [11′]

[0142] has solutions in GF(2^2m+1), the equation

[0143] (z³)²+bz³+(a³+b²)=0 [18′]

[0144] also has all solutions in GF(2^2m+1). Using Cardan's Rule as applied to even-bit Galois Fields, we can show that every solution (b, z_i) is a cube of (s,t) ε GF(2^4m+2), and that

[0145] (b,z_i)=(s,t)³

[0146] such that s, t and s+t are all solutions of equation 11′. The value of z_iis determined using the two-error method discussed in Section A over GF(2^4m+1). The cubic root of z_iis determined using one of the subsystems described below with reference to FIGS. 4-6.

[0147] Referring now to FIG. 2, as an example, we determine the roots of

[0148] (y₁, y₂)³+(0, a)(y₁, y₂)+(0,b)=0 [11*]

[0149] over GF(2¹⁸), which is an extension field of GF(2⁹), with a=α⁴¹and b=α¹³. First, we substitute variables and produce in GF(2¹⁸) the polynomial

[0150] (z₁, z₂)²+(0, α¹³)(z₁, z₂)+(0, α⁴¹)³=(0,0) [18*]

[0151] which is equation 18′ over GF(2¹⁸). Next, in step 204 we multiply the terms out to produce

[0152] (z₁², z₁²+z₂²)+(z₁α¹³, α¹³z²)+(0,(α⁴¹)³)=(0,0)

[0153] and equate the terms of the various pairs:

[0154] z₁²+z₁α¹³=0

[0155] z₁²+z₂²+z₂α¹³+α¹³+(α⁴¹)³=0

[0156] and in step 206 solve for z₁:

[0157] z₁=α¹³

[0158] We then substitute z₁=α¹³to produce:

[0159] z²₂+(α¹³)²+z₂α¹³+(α⁴¹)³=z₂²+α¹³z₂+(α⁴¹)³+(α¹³)²=0 [18′]

[0160] which is equation 18′ with Z₂=Z³, and solve for the roots of this equation using the method of section A (step 208). The multiplication operations used to determine the roots involve the 2⁹-bit elements of equation 18′. The roots of equation 18′ are Z₂₀=α⁹⁵and Z₂₁=α¹¹⁷, and thus, the roots of equation 18* are z₀₁=(α¹³, α⁹⁵) and z⁰²=(α¹³, α¹¹⁷). In step 210, the system determines the cubic roots of z₀₁and z₀₂, which are equal, and are (α¹³², α¹³¹)=(s,t). The roots of equation 11* are of the form

(y_{1}, y_{2}) = (x_{1} x_{2}) + \frac{(0, a)}{(x_{1} x_{2})}

[0161] and we determine in step 212 that the roots of equation 11′ are s, t and s+t or α¹³²; α¹³¹; α¹³²+α¹³¹=α²⁶¹. The system thus finds roots over the extension field GF(2⁹), while making its computations over GF(2).

[0162] C. Finding the Roots Of Degree-Four Error Locator Polynomials

[0163] Referring to FIG. 3, and as discussed in the co-pending pent application, the error locator processor 16 can determine the roots of a degree-four polynomial,

[0164] h(x)=σ₄x⁴+σ₃x+σ₂x²+σ₁x+σ₀

[0165] in steps 300-306 by transforming the polynomial into

[0166] j(x)=y⁴+a₂y²+a₁y+a₀

[0167] and then in step 308 transforming the polynomial into two quadratic expressions, namely,

[0168] j(x)=y²+ty+u)(y²+uv+w)

[0169] The system in step 310 solves for the values of t, u, v and w, and determines that the roots of j(x) are

[0170] t=v, tv+u+w=a₂, tw+uv=a₁and uw=a₀.

[0171] The system in step 312 next determines an expression in which t is the only unknown, namely,

[0172] t³+a₂t=a₁=0

[0173] and in steps 314-316 solves for t using the methods discussed above over the appropriate odd- or even-bit field. The system then in step 318-320 determines the roots of h(x) based on the substitution of variables between h(x) and j(x).

[0174] We now discuss how the system determines cubic roots in even- and odd-bit fields.

[0175] D. Cubic Roots In An Odd-Bit Field

[0176] In GF(2^2m+1), every element has one and only one cubic root in the field. For each element β of GF(2^2m+1) there is an element β^u

u = \frac{2^{2 m + 2} - 1}{3}

[0177] in GF(2^2m+1). Further, there is also an element

[0178] (β^u)³=β²⁻¹ [20]

[0179] But 2^2m+2=2^2m+1+2^2m+1and equation 20 can be rewritten

[0180] (β^u)³=β²*β²⁻¹ [21]

[0181] In GF(2^2m+1), any element, α, raised to the 2^2mpower, i.e., α², equals α⁰=1 and any element raised to the 2^2m+1power is equal to the element, since α²=α²*α¹=α.

[0182] Accordingly, β²=β in expression 21 and

[0183] β²*β²⁻¹=β¹*(β¹*β⁻¹)=β(β⁰)=β,

[0184] which means that β^uis the cube root of β.

[0185] Referring now to FIG. 4, a subsystem 400 determines the cubic root of an element β in GF(2^2m+1) by raising that element to the power u in multipliers 402, which include of one or more conventional Galois Field multipliers. In, for example, GF(2³), the multipliers 402 raise the element β to the power:

\frac{2^{2 (1) + 2} - 1}{3} = \frac{2^{4} - 1}{3} = 5

[0186] The multipliers 402 thus produce

[0187] β⁵=(αⁱ)⁵=α⁵ⁱ

[0188] by, for example, raising B to B²and multiplying the result by B. If i=2, β=α², and the cube root is α¹⁰, which in GF(2³) is α³. To test if α³is the cube root of α², we multiply α³*α³*α³=α⁹and α⁹=α²*α⁷, where α⁷=α⁰=1 in GF(2³). Accordingly, the product (α³)³=α², and α³is thus the cubic root of α².

[0189] E. Cubic Roots In An Even-Bit Field

[0190] Some elements of GF(2^2m) do not have cubic roots in the field. Further, determining cubic roots in GF(2^2m) for those elements that do have roots is more complicated than determining the cubic roots of elements of an odd-bit field.

[0191] The system may use the method described in the co-pending application to determine the cubic roots of an element of GF(^2m). Using this method, the system raises an element β of GF(2^2m) to powers that correspond to factors of the identity 2^2m−1=(2^m−1)*(2^m+1), which for any value of m has one factor that is divisible by three and one factor that is not. The system then produces the values:

γ_{1} = β^{2^{m} + ε} and \dots γ_{2} = β^{\frac{2^{m} - s}{3}}

[0192] where ε=1 if m is even and ε=−1 if m is odd.

[0193] The system next enters a look-up table to determine the cubic root of γ₁, which has the exponent that is not divisible by three. The system then produces the product:

\begin{matrix} γ_{1}^{1 / 3} γ_{2} = {(β^{2^{m} + ϵ})}^{1 / 3} * β^{\frac{2^{m} - ϵ}{3}} = {(β^{1 / 3})}^{2^{m + 1}} & [22] \end{matrix}

[0194] Finally, the system raises this product to the 2^m−1power, to produce (β^⅓)²which in GF(2^2m) is (β^⅓)¹or the cube root β^⅓.

[0195] This method uses a look-up table of a particular size for a given value of m. In a system with limited storage space that requires a smaller table or in which no table can be used, the system may instead incorporate, respectively, the subsystem 500 or subsystem 600 discussed below with reference to FIGS. 5 and 6.

[0196] 1. A Limited Look-Up Table Method

[0197] We define a variable “N” as

[0198] N=2^2m−1=(2²−1)((2²)^m−1+(2²)^m−2+ . . . +1),

[0199] which is divisible by 3. There are integer values “e” and “N₀” such that N=3^eN₀and the greatest common divisor of 3 and N₀is 1. If L is a factor of N₀, L is also a factor of 2^2m−1 and the greatest common divisor of L and 3 is 1. If we define a variable “q” as an integer where

[0200] 2^q−1<L<2¹+1

[0201] there is then an integer “K” such that

[0202] 2^q=3K+1 or 3K=2

[0203] and we can define an integer pair

(Q, V) \equiv {\begin{matrix} (q, 0) \dots if . L = 2^{q} \\ (q, K - l) \dots if . L = 3 b + 1 \dots and \dots 2^{q} = 3 K + 1 \\ (q + 1, .2 K - l) \dots if . L = 3 b + 2 \dots and \dots 2^{q} = 3 K + 1 \\ (q + 1, .2 K - l + 1) \dots if . L = 3 b + 1 \dots and \dots 2^{q} = 3 K = 2 \\ (q, K - l + 1) \dots if . L = 3 b + 2 \dots and {…2}^{q} = 3 K = 2 \end{matrix}

[0204] where “b” is any integer. We then can say L+₃V=2^Q.

[0205] If α is the primitive element of GF(2^2m) there is a smallest positive integer “s” such that

[0206] (α^3L)^s=1

[0207] since 3L is a factor of 2^2m−1. This is true when 3Ls=2²⁻¹or an integer multiple thereof since α²¹=1 in GF(2^2m). We can then define a set T:

[0208] T={α^i3L|i=0,1, . . . , s−1}.

[0209] and a mapping function

M (γ) = γ^{1 / 3} = α^{iL} α^{l \frac{2^{2 m} - 1}{3}},

[0210] where γ=α^i3LεT and l is an element of {0,1,2}. This mapping function, which essentially implements the equation 22 of the prior method, maps α^i3L=β^L, to

[0211] (β^L)^⅓=α^iLA

[0212] where

A = \frac{l (2^{m} - 1)}{3}

[0213] for some l ε {0,1,2}, A is divisible by 3, and A³=1.

[0214] The mapping function M(γ) can be implemented in a look-up table, which has s elements. For larger values of m, this table contains fewer elements than the look-up table of the method described in the co-pending application. For smaller values of m and thus smaller fields GF(2^2m), however, the size of this table is the same size as the table used in the prior method.

[0215] For a given element β=α³ⁱε GF(2^2m) the mapping function M produces (β^L)^⅓=α^iLA. This product is then converted to the cubic root of β by first multiplying the product by β^V:

[0216] (β^L)^⅓β^V=α^iLAα^3iV [23]

[0217] substituting L+3V=2^Qinto equation 23:

[0218] (β^L)^⅓β^V=(αⁱ)²A [24]

[0219] Then, raising both sides of equation 24 to the 2^2m−Qpower:

{({(β^{L})}^{\frac{1}{3}} β^{v})}^{2^{2 m - Q}} = {(α^{{ⅈ2}^{Q}})}^{2^{2 m - Q}} A^{2^{2 m - Q}} = {(α^{i})}^{2^{2 m}} A^{2^{2 m - Q}} = α^{i} A^{2^{2 m - Q}}

[0220] Since A³=1 and αⁱ=β^⅓, we can show that

{{[{(β^{L})}^{\frac{1}{3}} β^{V})}^{2^{2 m - Q}}]}^{3} = {(β^{\frac{1}{3}} A^{2^{2 m - Q}})}^{3} = {(β (1))}^{2^{2 m - Q}} = β

[0221] and

{({(β^{L})}^{\frac{1}{3}} β^{V})}^{2^{2 m - Q}}

[0222] is thus a cubic root of β, and further:

[0223] β^⅓=(M(β^L)β^V))² [25 ]

[0224] The exponents L and V can be expressed as sums of powers of two:

L = \sum_{i = 0}^{n_{l}} l_{i} 2^{i} and V = \sum_{i = 0}^{n_{v}} v_{i} 2^{i}

[0225] and thus, raising the element β to the powers of L and V is readily implemented in hardware using conventional Galois Field multipliers.

[0226] Referring now to FIG. 5, a subsystem 500 determines the cubic root of an element β of GF(2^2m) by first raising the element to the power L in multipliers 502. The subsystem then uses the value β^Lto enter a look-up table 504 that contains the results of mapping β^Lto (β^L)^⅓. The value read from the table 504 is then multiplied in multiplier 508 by β^V, which is produced in multipliers 506. Next, the product (β^L)^⅓ β^Vis raised to the power 2^m−Qin multipliers 510, to produce a cubic root (β)^⅓.

[0227] The values of L, Q and V are given in Table I below for GF(2^2m) for 3<m≦15. The number of elements, s, in the look-up table is also included in the table. The value of L is chosen to minimize the number of elements in the look-up table. Also, if possible L is chosen such that there is a minimal number of non-zero values for l_iand v_brespectively.

[0228] 2.No Look-Up Table Method

[0229] Consider first that an even-bit field GF(2^4m+2) is an extension field of the odd-bit field GF(2^2m+1), and thus, the odd-bit field is a sub-field of the even-bit field. If α is the primitive element of GF(2^4m+2), every element β of GF(2^2m+1) can be expressed as

[0230] β=(αⁱ)²⁺¹

[0231] for some integer i. Since 2^2m+1+1 is divisible by 3, there is in GF(2^4m+2) a first cubic root:

\begin{matrix} (α^{i}) 2^{2 m + 1} + \frac{1}{3} & [26] \end{matrix}

[0232] which is an element of the odd-bit field GF(2^m), where m₀is an odd integer.

[0233] The root may not, however, be an element of GF(2²m+¹).

[0234] There are two additional roots that are related to the first root, namely

\begin{matrix} α^{i^{\frac{(2^{2 m + 1} + 1)}{3} + \frac{(2^{4 m + 2} - 1)}{3}}} = {(α^{\frac{1}{3}})}^{ⅈ (2 m + 1_{+ 1}) + \frac{2^{4 m + 2} - 1}{3}} & [27] \\ and \\ α^{i^{\frac{(2^{2 m + 1} + 1)}{3} + \frac{(2^{4 m + 2} - 1)}{3}}} = {(α^{\frac{1}{3}})}^{ⅈ (2 m + 1_{+ 1}) + 2 (2^{4 m + 2} - 1)} & [28] \end{matrix}

[0235] which are determined, respectively, by multiplying the first root by α²⁻¹=α⁰=1 and α²⁽²⁻¹⁾=α⁰+α⁰=α⁰=1. The cubic roots in GF(2^4m+2) are thus:

\begin{matrix} β^{\frac{1}{3}} = β^{u} = α^{i^{\frac{(2^{2 m + 1} + 1)}{3} + \frac{l (2^{4 m + 2} - 1)}{3}}} & [29] \end{matrix}

[0236] where the root is also an element of the odd-bit field GF(2^2m+1) for l equal to one of {0,1,2}. We now use this result to determine the cubic root of an element of any even-bit field GF(2^2m).

[0237] The even-bit field GF(2^2m) is an extension field of the odd-bit field GF(2^m), where m=2^km₀and m₀is an odd integer. For example, with GF(2⁴) m=2, and k=1 and m₀=1 and the field is an extension field of GF(2); with GF(2²⁶), m=13, and k=0 and m₀=13, and the field is an extension field of the odd-bit field GF(2¹³) and so forth.

[0238] Also, we can show that

[0239] 2^2m−1=(2^m−1)N

[0240] where N=(2^m+1)N₀and N₀is a positive integer. To do this we substitute 2^km₀for m and repeatedly use the identity x²−1=(x−1)(x+1), where x=2^m)²in a first iteration.

[0241] For every βε GF(2^2m), β^Nis an element of GF(2^m). Since 2^m+1 is divisible by three,

[0242] N=3^eL

[0243] where e>0 and L is not divisible by three. We use these results below to determine the cubic roots when e=1 and when e>1 in Sections 2a and 2b, respectively.

[0244] 2a. When e=1

[0245] To find the cubic root of an element β of GF(2^2m), with

[0246] β=α³ⁱ

[0247] where α is the primitive element of GF(2^2m), the system first raises the element β to the power {fraction (N/3+L )} to produce:

β^{\frac{N}{3}} = {(β^{\frac{1}{3}})}^{N} = {(α^{i})}^{N}

[0248] which is an element of the odd-bit field GF(2^m). As discussed above, the system then determines the cubic root over the odd-bit field by raising the element to the power

u = \frac{2^{m_{0} + 1} - 1}{3}

[0249] which based on equation 29 is:

\begin{matrix} {(β^{\frac{N}{3}})}^{u} = {(β^{\frac{N}{3}})}^{1 / 3} * α^{\frac{l (2^{m} - 1)}{3}} = {([{(α^{3 i})}^{\frac{3 L}{3}}])}^{1 / 3} * α^{l (\frac{2^{m} - 1}{3})} = α^{iL} * α^{l} 2^{m} - \frac{1}{3} & [30] \end{matrix}

[0250] where lε {0,1,2}, and for one of these values the cubic root is an element of GF(2^m). Letting

A = α^{l (\frac{2^{m} - 1}{3})}

[0251] the result of equation 30 is rewritten as:

\begin{matrix} {(β^{\frac{N}{3}})}^{u} = α^{iL} A . & [31] \end{matrix}

[0252] Multiplying both sides of equation 31 by β^V

\begin{matrix} {β^{V} (β^{\frac{N}{3}})}^{u} = (α^{iL}) (β^{V}) = A (α^{iL}) ({(α^{3 i})}^{V}) A = {(α^{i})}^{L + 3 V} A & [32] \end{matrix}

[0253] substituting L+3V=2^Qinto equation 32:

\begin{matrix} {β^{V} (β^{\frac{N}{3}})}^{u} = {(α^{i})}^{2^{Q}} A & [33] \end{matrix}

[0254] further substituting β^Lfor β^{{fraction (N/3)}} in equation 33:

{β^{V} [(β^{L})]}^{u} = {(α^{i})}^{2^{Q}} A

[0255] and raising both sides to the power 2^2m−Q, we produce:

{[{(β^{L})}^{u} β^{v}]}^{2^{2 m - Q}} = α^{i} A^{2^{2 m - Q}} = β^{\frac{1}{3}} A^{2^{2 m - Q}}

[0256] which, as discussed above is the cube root of β, sine A³=1.

[0257] Referring now to FIG. 6 a subsystem 600 determines the cubic root of an element β of GF(2^2m) by first raising the element to the L power in multipliers 602, and then raising the result to the u power in multipliers 604. The subsystem next multiplies in multipliers 608 the product (β^L)^uby β^V, which is produced by multipliers 606, and in multipliers 610 raising the result to the 22 power.

[0258] 2b. When e>1

[0259] As discussed above, we raise βε GF(2^2m) to the {fraction (N/3+L )} power to produce

[0260] β^{{fraction (N/3)}}=(αⁱ)^N

[0261] which is an element of GF(2^m). Then, we raise the result to the power

u = \frac{2^{m_{0} + 1} - 1}{3}

[0262] to produce

\begin{matrix} {(β^{\frac{N}{3}})}^{u} = {(α^{ⅈ N})}^{u^{e}} = α^{{ⅈ3}^{e - 1} L} l_{1} \frac{2^{2 m - 1}}{3} & [26] \end{matrix}

[0263] where l₁is one of {0, 1, 1}. The root, which is an element of GF(2^m), is then raised to the u power and so forth until we produce the product

\begin{matrix} {(β^{\frac{N}{3}})}^{u^{e}} = {(α^{i^{N}})}^{u^{e}} = α^{i^{L}} A_{e} & [27] \end{matrix}

[0264] where A_eis equal to α to the power

\sum_{j = 1}^{e} l_{j} \frac{2^{2 m} - 1}{3^{ⅇ - j + 1}}

[0265] Next, we multiply both sides of equation 27 by β^V=α^3iV, to produce

\begin{matrix} {(β^{\frac{N}{3}})}^{u^{e}} β^{V} = α^{iL} α^{i3V} & [28] \end{matrix}

[0266] and substitute L+3V=2^Qinto equation 28 to produce

{(β^{\frac{N}{3}})}^{u^{e}} β^{V} = {(α^{i})}^{2^{Q}} A .

[0267] We then raise both sides to the 22 power:

{[{(β^{\frac{N}{3}})}^{u^{e}} β^{V}]}^{2^{2 m - Q}} = α^{i} A^{2^{2 m - Q}}

[0268] However αⁱ=β^⅓, and we have

β^{\frac{1}{3}} A = {[{(β^{\frac{N}{3}})}^{u^{e}} β^{V}]}^{2^{2 m - Q}}

[0269] Since

{(A^{2^{2 m - Q}})}^{3} = {(α^{\sum_{i = 1}^{e - 1} l_{\frac{2^{2 m - 1}}{ⅈ^{ⅇ^{e} - j}}}})}^{2^{2 m - Q}}

[0270] it may not equal 1, which means that αⁱA²may not be a cubic root of β. However,

[0271] (A²)³=1

[0272] and thus (αⁱA²)³=(αⁱ³)^e=β^e

[0273] and

β = {(α^{i} A^{2^{2 m - Q}})}^{3} α^{k \frac{2^{2 m} - 1}{3^{e - 1}}}

[0274] for some k ε {0, 1, 2 . . . 3^e−1−1}, and thus,

\begin{matrix} β^{\frac{1}{3}} = (α^{i} A^{2^{2 m - Q}}) α^{k \frac{2^{2 m} - 1}{3^{e}}} & [28] \end{matrix}

[0275] The cubic roots can then be found by trying different values of k and testing the result by cubing it.

[0276] Alternatively, a look-up table with 3^e−1elements may be used to determine the cubic root. Since

[0277] (αⁱA²)³=β(A²)³

[0278] we can determine (A^2m−Q)³by dividing (αⁱA²)³by β. There are at most 3^e−1possible values of kA², namely,

{kA}^{2^{2 m - Q}} = α^{\sum_{i = 1}^{e - 1} l_{k} j \frac{2^{2 m - 1}}{3^{e - j - 1}}}

[0279] for l_kjε{0,1,2}

[0280] such that

A^{2^{2 m - Q}} = α^{\sum_{i = 1}^{e - 1} l_{k} j \frac{2^{2 m - 1}}{3^{e - j - 1}}}

[0281] The look-up table is entered with the value k³A²and produces the value kA². Then, the cubic root of β is:

\frac{α^{i} A^{2^{2^{2 m - Q}}}}{{kA}^{2^{2 m - Q}}} = α^{i} α^{le \frac{(2^{2 m} - 1)}{3}}

[0282] We list below in Table II the values of N, e, u, Q and V for GF(2^2m)3m≦15. When e=1, no look-up table is used. When e>1, the subsystem may use a look-up table of the size 3^e−1, which is listed in the last column of the table. An entry of 0 in the column indicates that no table is used. Alternatively, various values of k may instead be substituted into equation 28, to determine the cubic roots.

Claims

What is claimed is:

1. A method of operating a data processing system to determine the locations of erroneous symbols in a code word over GF(2^2m+1), the method including the steps of:

i. receiving a code word with erroneous data symbols;

ii. determining an associated error locator polynomial;

iii. transforming the error locator polynomial to a degree three polynomial;

iv. substituting variables in the degree three polynomial to produce in an extension field GF(2^4m+2) an equation (y₁, y₂)³+(0,b)(y₁, y₂)+(0, a)=0;

v. substituting variables to produce a quadratic equation (z₁, z₂)²+(0,b) (z₁, z₂)+(0,a)³=0

where (z₁, z₂) is an element of GF(2^4m+2) and solving the quadratic equation for (z₁, z₂);

vi. determining the cube root, (s,t), of (z₁, z₂);

vii. determining the roots, s, t, and s+t, of the equation of step iv based on the value of (s,t) of step vi; and

viii. determining that the code word symbols associated with the code word locations that correspond to the roots s, t and s+t are erroneous; and

ix. based on the location information

a. correcting errors in the data symbols of the code word,

b. labeling the code word as uncorrectable, or

c. requesting that the code word is again received and repeating steps i-ix as necessary.

2. The method of claim 1 wherein the step of determining the cube root of (z₁,z₂) includes raising β=(z₁,z₂) to a power

u = \frac{2^{2 m + 2} - 1}{3}

to produce the element that is the cubic root of β.

3. The method of claim 1 wherein the step of determining the cube root of (z₁,z₂) includes the steps of:

i. raising β=(z₁,z₂) to a power L, where L is a factor of 2^2m−1;

ii. raising β to a power V where L+3V=2^Q;

iii. determining a cubic root of β^L;

iv. multiplying (β^L)^⅓ by β^V; and

v. raising (β^L)^⅓ β^Vto a power 2^2m−Qto produce the element that is the cubic root of (z₁, z₂).

4. The method of claim 1 wherein the step of determining the cube root of (z₁,z₂) includes the steps:

where m₀=2^km

raising (β^L)^uβ^Vto a power 2^2m−Qto produce the element that is the cubic root of (z₁, z₂).

i. raising β=(z₁,z₂) to a power L, where L is a factor of 2^2m−1;

ii. raising β^Lto a power

u = \frac{2^{m_{0} - 1} - 1}{3}

iii. raising the element β to a power V, where L+3V=2^Q

iv. multiplying (β^L)^uby β^V; and

5. A method of operating a data processing system to identify erroneous symbols in a code word, the method including the steps of:

to produce the cubic root of β; and

A. receiving a code word with erroneous data symbols and producing an associated degree-three polynomial with a term β³that is an element of GF(2^2m+1);

B. raising the element β³to a power

u = \frac{2^{2 m + 2} - 1}{3}

C. using the cubic root of β, identifying which data symbols of the received code word are erroneous and

i. correcting the errors in the data symbols of the code word,

ii. labeling the code word as uncorrectable, or

iii. requesting receipt of the code word again and repeating steps A-C as necessary.

6. A method of operating a data processing system to identify erroneous data symbols in a code word, the method including the steps of:

i. receiving a code word with erroneous data symbols and producing an associated degree-three polynomial with a term β³that is an element of GF(2^2m);

ii. raising the element β³to a power L, where L is a factor of 2^2m−1;

iii. raising , to a power V where L+3V=2^Q;

iv. using β^Lto enter a stored lookup table to retrieve the cubic root of β^L;

v. multiplying (β^L)^⅓ by β^V;

vi raising (β^L)^⅓ β^Vto a power 2^2m−Qto produce the cubic root of; and

vii using the cubic root of β, identifying which data symbols of the received code word are erroneous and

a. correcting the errors in the data symbols of the code word,

b. labeling the code word as uncorrectable, or

c. requesting receipt of the code word again and repeating steps i-vii as necessary.

7. A method of operating a data processing system to identify erroneous data symbols in a received code word, the method including the steps of:

i. receiving a code word with erroneous data symbols and producing an associated degree-three polynomial with a term β³that is an element of GF(2^2m);

ii. raising the element β to a power L, where L is a factor of 2^2m−1;

iii. raising β^Lto a power

u = \frac{2^{m_{0} - 1} - 1}{3}

where m₀=2^km

iv. raising the element β to a power V, where L+3V=2^Q

v. multiplying (β^L)^uby β^V;

vi. raising (β^L)^uβ^Vto a power 2^2m−Qto produce the cubic root; and

vii. using the cubic root of β, identifying which data symbols of the received code word are erroneous and

a. correcting the errors in the data symbols of the code word,

b. labeling the code word as uncorrectable, or

c. requesting receipt of the code word again and repeating steps i-vii as necessary.

8. A data processing system for determining the roots of a degree-three polynomial that is associated with a data codeword over GF(2^2m+1), the system including:

where (z₁, z₂) is an element of GF(2^4m+2) and solving the quadratic equation for (z₁, z₂);

A. a processor for receiving a data code word containing errors and determining an associated degree-three polynomial;

B. manipulation means for substituting variables in the degree three polynomial to produce in an extension field GF(2^4m+2) an equation (y₁, y₂)³+(0,b)(y₁, y₂)+(0, a)=0;

C. manipulation means for substituting variables to produce a quadratic equation (z₁, z₂)²+(0,b) (z₁, z₂)+(0,a)³=0

D. calculating means for calculating the cube root, (s,t), of (z₁, z₂); and

E. means for transforming the roots (s, t) and s+t to the roots of the degree-three polynomial and associating the roots with the locations of erroneous data symbols in the data code word; and

F. an error correction processor for

i. correcting the erroneous data symbols in the code word,

ii. labeling the code word uncorrectable, or

iii. sending a repeat request for the code word.

9. The system of claim 8 wherein the calculating means includes a multiplier for raising β=(z₁, z₂) to a power

u = \frac{2^{2 m + 2} - 1}{3}

to produce the element that is the cubic root of β.

10. The system of claim 8 wherein the calculating means includes

i. one or more first multipliers for raising β=(z₁,z₂) to a power L, where L is a factor of 2^2m−1;

ii. one or more second multipliers for raising β to a power V where L+3V=2^Q;

iii. means for determining a cubic root of β^L;

iv. one or more third multipliers for multiplying the cubic root by β^V; and

v. one or more fourth multipliers for raising the product produced by the one or more third multipliers to a power 2^2m−Qto produce the cubic root.

11. The system of claim 10 wherein the means for determining a cubic root of β^Lincludes a lookup table.

12. The system of claim 10 wherein the means for determining a cubic root of β^Lincludes one or more fifth multipliers for raising β^Lto a power

u = \frac{2^{m_{0} - 1} - 1}{3},

, where m₀=2^km.

13. A data processing system for determining error locations in a data code word, the system including:

A. a first processor for receiving the data code word and determining an associated degree-three polynomial;

B. a cubic root processor for calculating the cubic root of a multiple bit symbol β that is associated with the degree-three error locator polynomial and is also an element of GF(2^2m), the cubic root processor including

i. one or more first multipliers for raising β to a power L, where L is a factor of 2^2m−1;

ii. one or more second multipliers for raising β to a power V where L+3V=2^Q;

iii. means for determining a cubic root of β^L;

iv. one or more third multipliers for multiplying the cubic root by β^V; and

v. one or more fourth multipliers for raising the product produced by the one or more third multipliers to a power 2^2m−Qto produce the multiple bit symbol that is cubic root of β; and

C. a second processor for determining which symbols are in error in the data code word based on the multiple bit symbol that is the cubic root of β, the error location processor providing error location information to an error correction processor.

14. The system of claim 13 wherein the means for determining a cubic root of β^Lincludes a lookup table.

15. The system of claim 13 wherein the means for determining a cubic root of β^Lincludes one or more fifth multipliers for raising β^Lto a power

u = \frac{2^{m_{0} - 1} - 1}{3},

where m₀=2^km.

Patent Citations (8)

Patent	Date	Inventor	Cited By
US4099160	1978-07-01	Flagg	Examiner
US4567594	1986-01-01	Deodhar	Examiner
US4839896	1989-06-01	Glover et al.	Examiner
US4866716	1989-09-01	Weng	Examiner
US5001715	1991-03-01	Weng	Examiner
US5710782	1998-01-01	Weng	Examiner
US5761102	1998-06-01	Weng	Examiner
US5822336	1998-10-01	Weng et al.	Examiner

Abstract

Description

[0002] This application is a continuation of application Ser. No. 08/946,470, filed Oct. 7, 1997, now abandoned.

Field of the Invention

[0003] This invention relates generally to data processing systems and, more particularly, to a system for decoding and correcting errors in data using an error correction code.

Background of the Invention

[0011] The matrix manipulation and look-up table methods involve transforming a general degree-four polynomial of the form:

[0012] σ₄x⁴+σ₃x³+σ₂x²+σ₁x+σ₀

[0013] into a polynomial of the form:

[0014] x⁴+x²+ax+b.

Summary of the Invention

[0020] More specifically, the system first manipulates the degree-four polynomial:

[0021] σ(x)=σ₄x⁴+σ₃x³+σ₂x²+σ₁x+σ₀ [1]

[0022] into the form:

[0023] θ(y)=y⁴+θ₂y²+θ₁y+θ₀ [2]

[0024] where the θ_i's are combinations of the coefficients of the terms of the error locator polynomial. Note that the coefficient of the y³term is zero. The system then factors θ(y) into:

[0025] θ(y)=(y²+t*y+u)*(y²+v*y+w) [3]

[0026] where “*” represents multiplication, and expands the expression to:

[0027] θ(y)=y⁴+(t+v)y³+(t*v+u+w)y²+(t*w+v*u)y+u*w, [4]

[0028] where t, u, v and w are unknown. Equating the coefficients of expressions 2 and 3, the system determines that:

[0029] t+v=0

[0030] t*v+u*w=t²+u*w=θ₂

[0031] t*w+u*v=t+w+t*u=t*(w+u)=θ₁

[0032] u*w=θ₀. [5]

[0033] The system then produces an equation with the variable t as the only unknown:

[0034] t³θ₂t+θ₁=0 [6]

[0037] p²+(θ₂+t₀²)p+θ₀=0 [7]

[0039] y²+t₀*y+p₀=0

[0040] y₂+t₀*y+p₁=0 [8]

[0041] The roots of equations 8, which are determined using the Berlekamp method, are also the roots of θ(y), and they can thus be used directly to produce the roots of the error locator polynomial.

Brief Description of the Drawings

[0045] FIG. 1 is a functional block diagram of a decoder that is constructed in accordance with the invention;

[0046] FIG. 2 is a flow chart of the operations of the decoder of FIG. 1 in determining if roots of an equation are in GF(2^2m+1);

[0047] FIGS. 3A and 3B are a flow chart of the operations that the decoder of FIG. 1 performs in determining the locations of four errors;

[0048] FIG. 4 is a functional block diagram of a subsystem for determining the cubic roots of elements of GF(2^m+1);

[0049] FIG. 5 is a functional block diagram of a subsystem for determining the cubic roots of elements of GF(2^2m); and

[0050] FIG. 6 is a functional block diagram of an alternative system for determining the cubic roots of elements of GF(2^2m).

Detailed Description

[0051] The mathematical operations of addition, subtraction, multiplication and division discussed herein are Galois Field operations over the applicable Galois Fields.

[0056] A. Determining Two Error Locations

[0057] If the code word contains two errors, the error locator polynomial is of the form:

[0058] σ(x)=σ₂x²+σ₁x+σ₀x. [1]

[0059] To find the roots of the polynomial, the system sets the polynomial equal to zero:

[0060] σ(x)=σ₂x²+σ₁x+σ₀x=0 [2]

[0061] and solves the equation for the solutions, x₀and x₁.

[0062] To solve equation 2, the error locator processor 16 transforms the equation into the form:

[0063] y²+y+C=0 [3]

[0064] by changing the variable to:

\begin{matrix} x = \frac{σ_{1}}{σ_{2}} y & [4] \end{matrix}

[0065] such that equation 2 becomes

σ (\frac{σ_{1}}{σ_{2}} y) = \frac{σ_{1}^{2}}{σ_{2}} y^{2} + \frac{σ_{1}^{2}}{σ_{2}} y + σ_{0}

[0066] Then, dividing by

\frac{σ_{2}}{σ_{1}^{2}},

[0067] the system produces the equation:

\begin{matrix} y^{2} + y + \frac{σ_{0} σ_{2}}{σ_{1}^{2}} = 0 & [5] \end{matrix}

[0068] and the C of equation 3 is thus

\frac{σ_{0} σ_{2}}{σ_{1}^{2}} .

[0069] For a solution of equation 5 to exist in GF(2^m), the “traces” of both sides of the equation must be the same. A trace is defined as:

\begin{matrix} Trace (α^{i}) = \sum_{k = 0}^{m - 1} {(α^{i})}^{2^{k}} & [6] \end{matrix}

[0070] The Trace (0) equals 0, and thus, for solutions to exist within the Galois Field,

[0071] Trace [y²+y+C]=0 [7]

\frac{σ_{0} σ_{2}}{σ_{1}^{2}} .

\begin{matrix} {(α^{5})}^{2} + α^{5} = α^{0} & with y_{0} = α^{5} \\ {(α^{9})}^{2} + α^{9} = α^{1} & with y_{0} = α^{9} \\ {(α^{3})}^{2} + α^{3} = α^{2} & with y_{0} = α^{3} \end{matrix}

[0075] Another set of base solutions are:

\begin{matrix} {(α^{10})}^{2} + α^{10} = α^{0} & with y_{1} = α^{10} \\ {(α^{7})}^{2} + α^{7} = α^{1} & with y_{1} = α^{7} \\ {(α^{14})}^{2} + α^{14} = α^{2} & with y_{1} = α^{14} \end{matrix}

[0076] In this Galois Field, the solutions for equation 3 with C equal to α⁸=0101, for example, are determined by decomposing C into:

[0077] C=α⁰+α²=0001+0100

[0078] and summing a corresponding wet of base solutions, for example:

[0079] y₀=α⁵+α³=α¹¹

[0080] to produce a first root and determining a second root as:

[0081] y₁=y₀+α⁰

[0082] Next the system determines the roots of the degree-two error locator polynomial as:

\begin{matrix} x_{0} = y_{0} (\frac{σ_{1}}{σ_{2}}) \\ x_{1} = y_{1} (\frac{σ_{1}}{σ_{2}}) . \end{matrix}

[0083] B. Determining Three Error Locations

[0084] In general, for any degree-three error locator polynomial, σ(x), the system determines error locations by solving:

[0085] σ(x)=σ₃x³+σ₂x²+σ₁x+σ₀=0. [8]

[0086] Accordingly, the system changes variable to

x = y + \frac{σ_{2}}{σ_{3}}

[0087] and equation 1 becomes

\begin{matrix} σ (y + \frac{σ_{2}}{σ_{3}}) = {σ_{3} (y + \frac{σ_{2}}{σ_{3}})}^{3} + {σ_{2} (y + \frac{σ_{2}}{σ_{3}})}^{2} + σ_{1} (y + \frac{σ_{2}}{σ_{3}}) + σ_{0} = 0. & [9] \end{matrix}

[0088] which expands to:

σ_{3} (y^{3} + y^{2} \frac{σ_{2}}{σ_{3}} + {y (\frac{σ_{2}}{σ_{3}})}^{2} + {(\frac{σ_{2}}{σ_{3}})}^{3}) + σ_{2} (y^{2} + {(\frac{σ_{2}}{σ_{3}})}^{2}) + σ_{1} (y + \frac{σ_{2}}{σ_{3}}) + σ_{0} = 0

[0089] and after collecting like terms and summing identical terms to zero the equation becomes:

\begin{matrix} σ_{3} y^{3} + 0 y^{2} + (\frac{σ_{2}^{2}}{σ_{3}} + σ_{1}) y + 0 + \frac{σ_{1} σ_{2}}{σ_{3}} + σ_{0} = 0 & [10] \end{matrix}

[0090] The system then divides equation 10 by σ₃, and the equation becomes:

[0091] y³+αy+b=0 [11]

[0092] with

a = \frac{σ_{2}^{2}}{σ_{3}^{2}} + \frac{σ_{1}}{σ_{3}} and b = \frac{σ_{0}}{σ_{3}} + \frac{σ_{1} σ_{2}}{σ_{3}^{2}},

[0093] and y₀is a solution of equation 11 if and only if

y_{0} + \frac{σ_{2}}{σ_{3}}

[0094] is a solution of equation 8. Accordingly, we can determine the solutions of equation 8 by determining the solutions of equation 11.

[0097] B1. Solutions Over An Even-Bit Field GF(2^2m)

[0098] According to Cardan's Rule as applied to even-bit Galois Fields, equation 11 has all three solutions y₁, y₂and y₃in GF(2^2m) if and only if each root y_iis of the form

y_{i} = x_{i} + \frac{a}{x_{i}}

[0099] where x_iε GF(2^2m). This means that y_iis a root of equation 11 if and only if

[0100] x_i²+y_ix_i+a=0 [12]

[0102] To determine the trace of the left-hand side of the equation 12, we first define in Step 202 for a polynomial Q(x)=(x−α₁) (x−α₂) (x−α₃) . . . (x−α_n), a function p(Q(x)):

p (Q (x)) = \sum_{1 \leq i \leq j \leq n} \frac{α_{i} α_{j}}{{(α_{i} + α_{j})}^{2}} .

[0104] The polynomial of interest, x_i²+y_ix_i+a=(x−α₁) (x−α₂), with α₁+α₂=y_iand α₁α₂=a, and

p (x_{i}^{2} + y_{i} x_{i} + a) = \frac{α_{1} α_{2}}{{(α_{1} + α_{2})}^{2}} = \frac{a}{y_{i}^{2}} .

[0105] Accordingly, equation 12 has solutions in GF(2^2m) if and only if

\begin{matrix} Tr (\frac{a}{y_{i}^{2}}) = 0. & [13] \end{matrix}

[0106] Further, we determine at equation 11 has all solutions in GF(2^2m) by determining if Tr(p(y³+ay+b))=0, which means that

Tr (\frac{a^{3}}{b^{2}} + 1) = 0.

[0107] Over GF(2^2m), in which Tr(1)=0,

Tr (\frac{a^{3}}{b^{2}}) + Tr (1) = Tr (\frac{a^{3}}{b^{2}}) + 0,

[0108] and thus if all solutions are GF(2^2m),

\begin{matrix} Tr (\frac{a^{3}}{b^{2}}) = 0. & [14] \end{matrix}

[0110] y³+αy+b=0 [11]

[0111] and if α≠0, the processor 16 changes variable to:

\begin{matrix} y = z + \frac{a}{z} & [15] \end{matrix}

[0112] and equation 4 becomes

\begin{matrix} {(z + \frac{a}{z})}^{3} + a (z + \frac{a}{z}) + b = 0. & [16] \end{matrix}

[0113] It next expands equation 16 and collects like terms to produce:

\begin{matrix} z^{3} + \frac{a}{z} z^{2} + {(\frac{a}{z})}^{2} z + {(\frac{a}{z})}^{3} + az + \frac{a^{2}}{z} + b = z^{3} + az + \frac{a^{2}}{z} + \frac{a^{3}}{z^{3}} + az + \frac{a^{2}}{z} + b = z^{3} + \frac{a^{3}}{z^{3}} + b = 0. & [17] \end{matrix}

[0114] The system next multiplies equation 17 by z³to produce

[0115] (z³)²+bz³+a³=0, [18]

[0116] which is a quadratic equation with z³as the variable. Equation 18 can then be solved using the two-error method discussed in Section A.

z_{1} = z_{0} * α^{\frac{2^{2 m - 1}}{3}}

[0118] where the exponent

\frac{2^{2 m} - 1}{3}

[0119] is always an integer, since 2^2m−1 is divisible by 3, and a third solution:

[0120] z₂=z₁+z₀

\frac{σ_{2}}{σ_{3}}

[0122] to produce the roots x₀, x₁and x₂.

y_{1} = y_{0} * α^{\frac{2^{2 m - 1}}{3}} and y_{2} = y_{0} + y_{1},

[0124] and from these values determines the roots x₀, x₁and x₂.

[0125] B2. Solutions Over An Odd-Bit Field GF(2^2m+1)

[0126] When the system is operating over GF(2^2m+1), and equation 11

[0127] y³+αy+b=0 [11]

[0128] has all solutions in GF(2^2m+1), we know that

Tr (\frac{a^{3}}{b^{2}} + 1) = 0

[0129] and thus

Tr (\frac{a^{3}}{b^{2}}) = 1

[0130] since Tr(1)=1 in the odd-bit field. This means that the equation 18

[0131] (z³)²+bz³+a³=0 [18]

[0132] has no roots in GF(2^2m+1). Thus, the error locator processor 16 cannot use the method described above to determine the solutions of equation 11 in the odd-bit field.

[0134] (α₁x+β₁) (α₂x+β₂) modulo x²+x+1

[0135] such that

[0136] (α₁x+β₁) (α₂x+β₂)=(α₁α₂+α₁β₂+α₂β₁, α₁α₂+β₁β₂).

[0137] A simplified way to express GF(2^4m+1) is

[0138] GF(2^4m+2)={(α,↑)|α,βεGF(2^2m+1)}.

[0140] Using the extension field, we can show that since the equation:

[0141] y³+ay+b=0 [11′]

[0142] has solutions in GF(2^2m+1), the equation

[0143] (z³)²+bz³+(a³+b²)=0 [18′]

[0144] also has all solutions in GF(2^2m+1). Using Cardan's Rule as applied to even-bit Galois Fields, we can show that every solution (b, z_i) is a cube of (s,t) ε GF(2^4m+2), and that

[0145] (b,z_i)=(s,t)³

[0147] Referring now to FIG. 2, as an example, we determine the roots of

[0148] (y₁, y₂)³+(0, a)(y₁, y₂)+(0,b)=0 [11*]

[0149] over GF(2¹⁸), which is an extension field of GF(2⁹), with a=α⁴¹and b=α¹³. First, we substitute variables and produce in GF(2¹⁸) the polynomial

[0150] (z₁, z₂)²+(0, α¹³)(z₁, z₂)+(0, α⁴¹)³=(0,0) [18*]

[0151] which is equation 18′ over GF(2¹⁸). Next, in step 204 we multiply the terms out to produce

[0152] (z₁², z₁²+z₂²)+(z₁α¹³, α¹³z²)+(0,(α⁴¹)³)=(0,0)

[0153] and equate the terms of the various pairs:

[0154] z₁²+z₁α¹³=0

[0155] z₁²+z₂²+z₂α¹³+α¹³+(α⁴¹)³=0

[0156] and in step 206 solve for z₁:

[0157] z₁=α¹³

[0158] We then substitute z₁=α¹³to produce:

[0159] z²₂+(α¹³)²+z₂α¹³+(α⁴¹)³=z₂²+α¹³z₂+(α⁴¹)³+(α¹³)²=0 [18′]

(y_{1}, y_{2}) = (x_{1} x_{2}) + \frac{(0, a)}{(x_{1} x_{2})}

[0162] C. Finding the Roots Of Degree-Four Error Locator Polynomials

[0163] Referring to FIG. 3, and as discussed in the co-pending pent application, the error locator processor 16 can determine the roots of a degree-four polynomial,

[0164] h(x)=σ₄x⁴+σ₃x+σ₂x²+σ₁x+σ₀

[0165] in steps 300-306 by transforming the polynomial into

[0166] j(x)=y⁴+a₂y²+a₁y+a₀

[0167] and then in step 308 transforming the polynomial into two quadratic expressions, namely,

[0168] j(x)=y²+ty+u)(y²+uv+w)

[0169] The system in step 310 solves for the values of t, u, v and w, and determines that the roots of j(x) are

[0170] t=v, tv+u+w=a₂, tw+uv=a₁and uw=a₀.

[0171] The system in step 312 next determines an expression in which t is the only unknown, namely,

[0172] t³+a₂t=a₁=0

[0174] We now discuss how the system determines cubic roots in even- and odd-bit fields.

[0175] D. Cubic Roots In An Odd-Bit Field

[0176] In GF(2^2m+1), every element has one and only one cubic root in the field. For each element β of GF(2^2m+1) there is an element β^u

u = \frac{2^{2 m + 2} - 1}{3}

[0177] in GF(2^2m+1). Further, there is also an element

[0178] (β^u)³=β²⁻¹ [20]

[0179] But 2^2m+2=2^2m+1+2^2m+1and equation 20 can be rewritten

[0180] (β^u)³=β²*β²⁻¹ [21]

[0181] In GF(2^2m+1), any element, α, raised to the 2^2mpower, i.e., α², equals α⁰=1 and any element raised to the 2^2m+1power is equal to the element, since α²=α²*α¹=α.

[0182] Accordingly, β²=β in expression 21 and

[0183] β²*β²⁻¹=β¹*(β¹*β⁻¹)=β(β⁰)=β,

[0184] which means that β^uis the cube root of β.

\frac{2^{2 (1) + 2} - 1}{3} = \frac{2^{4} - 1}{3} = 5

[0186] The multipliers 402 thus produce

[0187] β⁵=(αⁱ)⁵=α⁵ⁱ

[0189] E. Cubic Roots In An Even-Bit Field

γ_{1} = β^{2^{m} + ε} and \dots γ_{2} = β^{\frac{2^{m} - s}{3}}

[0192] where ε=1 if m is even and ε=−1 if m is odd.

[0193] The system next enters a look-up table to determine the cubic root of γ₁, which has the exponent that is not divisible by three. The system then produces the product:

\begin{matrix} γ_{1}^{1 / 3} γ_{2} = {(β^{2^{m} + ϵ})}^{1 / 3} * β^{\frac{2^{m} - ϵ}{3}} = {(β^{1 / 3})}^{2^{m + 1}} & [22] \end{matrix}

[0194] Finally, the system raises this product to the 2^m−1power, to produce (β^⅓)²which in GF(2^2m) is (β^⅓)¹or the cube root β^⅓.

[0196] 1. A Limited Look-Up Table Method

[0197] We define a variable “N” as

[0198] N=2^2m−1=(2²−1)((2²)^m−1+(2²)^m−2+ . . . +1),

[0200] 2^q−1<L<2¹+1

[0201] there is then an integer “K” such that

[0202] 2^q=3K+1 or 3K=2

[0203] and we can define an integer pair

(Q, V) \equiv {\begin{matrix} (q, 0) \dots if . L = 2^{q} \\ (q, K - l) \dots if . L = 3 b + 1 \dots and \dots 2^{q} = 3 K + 1 \\ (q + 1, .2 K - l) \dots if . L = 3 b + 2 \dots and \dots 2^{q} = 3 K + 1 \\ (q + 1, .2 K - l + 1) \dots if . L = 3 b + 1 \dots and \dots 2^{q} = 3 K = 2 \\ (q, K - l + 1) \dots if . L = 3 b + 2 \dots and {…2}^{q} = 3 K = 2 \end{matrix}

[0204] where “b” is any integer. We then can say L+₃V=2^Q.

[0205] If α is the primitive element of GF(2^2m) there is a smallest positive integer “s” such that

[0206] (α^3L)^s=1

[0207] since 3L is a factor of 2^2m−1. This is true when 3Ls=2²⁻¹or an integer multiple thereof since α²¹=1 in GF(2^2m). We can then define a set T:

[0208] T={α^i3L|i=0,1, . . . , s−1}.

[0209] and a mapping function

M (γ) = γ^{1 / 3} = α^{iL} α^{l \frac{2^{2 m} - 1}{3}},

[0210] where γ=α^i3LεT and l is an element of {0,1,2}. This mapping function, which essentially implements the equation 22 of the prior method, maps α^i3L=β^L, to

[0211] (β^L)^⅓=α^iLA

[0212] where

A = \frac{l (2^{m} - 1)}{3}

[0213] for some l ε {0,1,2}, A is divisible by 3, and A³=1.

[0215] For a given element β=α³ⁱε GF(2^2m) the mapping function M produces (β^L)^⅓=α^iLA. This product is then converted to the cubic root of β by first multiplying the product by β^V:

[0216] (β^L)^⅓β^V=α^iLAα^3iV [23]

[0217] substituting L+3V=2^Qinto equation 23:

[0218] (β^L)^⅓β^V=(αⁱ)²A [24]

[0219] Then, raising both sides of equation 24 to the 2^2m−Qpower:

{({(β^{L})}^{\frac{1}{3}} β^{v})}^{2^{2 m - Q}} = {(α^{{ⅈ2}^{Q}})}^{2^{2 m - Q}} A^{2^{2 m - Q}} = {(α^{i})}^{2^{2 m}} A^{2^{2 m - Q}} = α^{i} A^{2^{2 m - Q}}

[0220] Since A³=1 and αⁱ=β^⅓, we can show that

{{[{(β^{L})}^{\frac{1}{3}} β^{V})}^{2^{2 m - Q}}]}^{3} = {(β^{\frac{1}{3}} A^{2^{2 m - Q}})}^{3} = {(β (1))}^{2^{2 m - Q}} = β

[0221] and

{({(β^{L})}^{\frac{1}{3}} β^{V})}^{2^{2 m - Q}}

[0222] is thus a cubic root of β, and further:

[0223] β^⅓=(M(β^L)β^V))² [25 ]

[0224] The exponents L and V can be expressed as sums of powers of two:

L = \sum_{i = 0}^{n_{l}} l_{i} 2^{i} and V = \sum_{i = 0}^{n_{v}} v_{i} 2^{i}

[0225] and thus, raising the element β to the powers of L and V is readily implemented in hardware using conventional Galois Field multipliers.

[0228] 2.No Look-Up Table Method

[0230] β=(αⁱ)²⁺¹

[0231] for some integer i. Since 2^2m+1+1 is divisible by 3, there is in GF(2^4m+2) a first cubic root:

\begin{matrix} (α^{i}) 2^{2 m + 1} + \frac{1}{3} & [26] \end{matrix}

[0232] which is an element of the odd-bit field GF(2^m), where m₀is an odd integer.

[0233] The root may not, however, be an element of GF(2²m+¹).

[0234] There are two additional roots that are related to the first root, namely

\begin{matrix} α^{i^{\frac{(2^{2 m + 1} + 1)}{3} + \frac{(2^{4 m + 2} - 1)}{3}}} = {(α^{\frac{1}{3}})}^{ⅈ (2 m + 1_{+ 1}) + \frac{2^{4 m + 2} - 1}{3}} & [27] \\ and \\ α^{i^{\frac{(2^{2 m + 1} + 1)}{3} + \frac{(2^{4 m + 2} - 1)}{3}}} = {(α^{\frac{1}{3}})}^{ⅈ (2 m + 1_{+ 1}) + 2 (2^{4 m + 2} - 1)} & [28] \end{matrix}

[0235] which are determined, respectively, by multiplying the first root by α²⁻¹=α⁰=1 and α²⁽²⁻¹⁾=α⁰+α⁰=α⁰=1. The cubic roots in GF(2^4m+2) are thus:

\begin{matrix} β^{\frac{1}{3}} = β^{u} = α^{i^{\frac{(2^{2 m + 1} + 1)}{3} + \frac{l (2^{4 m + 2} - 1)}{3}}} & [29] \end{matrix}

[0238] Also, we can show that

[0239] 2^2m−1=(2^m−1)N

[0240] where N=(2^m+1)N₀and N₀is a positive integer. To do this we substitute 2^km₀for m and repeatedly use the identity x²−1=(x−1)(x+1), where x=2^m)²in a first iteration.

[0241] For every βε GF(2^2m), β^Nis an element of GF(2^m). Since 2^m+1 is divisible by three,

[0242] N=3^eL

[0243] where e>0 and L is not divisible by three. We use these results below to determine the cubic roots when e=1 and when e>1 in Sections 2a and 2b, respectively.

[0244] 2a. When e=1

[0245] To find the cubic root of an element β of GF(2^2m), with

[0246] β=α³ⁱ

[0247] where α is the primitive element of GF(2^2m), the system first raises the element β to the power {fraction (N/3+L )} to produce:

β^{\frac{N}{3}} = {(β^{\frac{1}{3}})}^{N} = {(α^{i})}^{N}

[0248] which is an element of the odd-bit field GF(2^m). As discussed above, the system then determines the cubic root over the odd-bit field by raising the element to the power

u = \frac{2^{m_{0} + 1} - 1}{3}

[0249] which based on equation 29 is:

\begin{matrix} {(β^{\frac{N}{3}})}^{u} = {(β^{\frac{N}{3}})}^{1 / 3} * α^{\frac{l (2^{m} - 1)}{3}} = {([{(α^{3 i})}^{\frac{3 L}{3}}])}^{1 / 3} * α^{l (\frac{2^{m} - 1}{3})} = α^{iL} * α^{l} 2^{m} - \frac{1}{3} & [30] \end{matrix}

[0250] where lε {0,1,2}, and for one of these values the cubic root is an element of GF(2^m). Letting

A = α^{l (\frac{2^{m} - 1}{3})}

[0251] the result of equation 30 is rewritten as:

\begin{matrix} {(β^{\frac{N}{3}})}^{u} = α^{iL} A . & [31] \end{matrix}

[0252] Multiplying both sides of equation 31 by β^V

\begin{matrix} {β^{V} (β^{\frac{N}{3}})}^{u} = (α^{iL}) (β^{V}) = A (α^{iL}) ({(α^{3 i})}^{V}) A = {(α^{i})}^{L + 3 V} A & [32] \end{matrix}

[0253] substituting L+3V=2^Qinto equation 32:

\begin{matrix} {β^{V} (β^{\frac{N}{3}})}^{u} = {(α^{i})}^{2^{Q}} A & [33] \end{matrix}

[0254] further substituting β^Lfor β^{{fraction (N/3)}} in equation 33:

{β^{V} [(β^{L})]}^{u} = {(α^{i})}^{2^{Q}} A

[0255] and raising both sides to the power 2^2m−Q, we produce:

{[{(β^{L})}^{u} β^{v}]}^{2^{2 m - Q}} = α^{i} A^{2^{2 m - Q}} = β^{\frac{1}{3}} A^{2^{2 m - Q}}

[0256] which, as discussed above is the cube root of β, sine A³=1.

[0258] 2b. When e>1

[0259] As discussed above, we raise βε GF(2^2m) to the {fraction (N/3+L )} power to produce

[0260] β^{{fraction (N/3)}}=(αⁱ)^N

[0261] which is an element of GF(2^m). Then, we raise the result to the power

u = \frac{2^{m_{0} + 1} - 1}{3}

[0262] to produce

\begin{matrix} {(β^{\frac{N}{3}})}^{u} = {(α^{ⅈ N})}^{u^{e}} = α^{{ⅈ3}^{e - 1} L} l_{1} \frac{2^{2 m - 1}}{3} & [26] \end{matrix}

[0263] where l₁is one of {0, 1, 1}. The root, which is an element of GF(2^m), is then raised to the u power and so forth until we produce the product

\begin{matrix} {(β^{\frac{N}{3}})}^{u^{e}} = {(α^{i^{N}})}^{u^{e}} = α^{i^{L}} A_{e} & [27] \end{matrix}

[0264] where A_eis equal to α to the power

\sum_{j = 1}^{e} l_{j} \frac{2^{2 m} - 1}{3^{ⅇ - j + 1}}

[0265] Next, we multiply both sides of equation 27 by β^V=α^3iV, to produce

\begin{matrix} {(β^{\frac{N}{3}})}^{u^{e}} β^{V} = α^{iL} α^{i3V} & [28] \end{matrix}

[0266] and substitute L+3V=2^Qinto equation 28 to produce

{(β^{\frac{N}{3}})}^{u^{e}} β^{V} = {(α^{i})}^{2^{Q}} A .

[0267] We then raise both sides to the 22 power:

{[{(β^{\frac{N}{3}})}^{u^{e}} β^{V}]}^{2^{2 m - Q}} = α^{i} A^{2^{2 m - Q}}

[0268] However αⁱ=β^⅓, and we have

β^{\frac{1}{3}} A = {[{(β^{\frac{N}{3}})}^{u^{e}} β^{V}]}^{2^{2 m - Q}}

[0269] Since

{(A^{2^{2 m - Q}})}^{3} = {(α^{\sum_{i = 1}^{e - 1} l_{\frac{2^{2 m - 1}}{ⅈ^{ⅇ^{e} - j}}}})}^{2^{2 m - Q}}

[0270] it may not equal 1, which means that αⁱA²may not be a cubic root of β. However,

[0271] (A²)³=1

[0272] and thus (αⁱA²)³=(αⁱ³)^e=β^e

[0273] and

β = {(α^{i} A^{2^{2 m - Q}})}^{3} α^{k \frac{2^{2 m} - 1}{3^{e - 1}}}

[0274] for some k ε {0, 1, 2 . . . 3^e−1−1}, and thus,

\begin{matrix} β^{\frac{1}{3}} = (α^{i} A^{2^{2 m - Q}}) α^{k \frac{2^{2 m} - 1}{3^{e}}} & [28] \end{matrix}

[0275] The cubic roots can then be found by trying different values of k and testing the result by cubing it.

[0276] Alternatively, a look-up table with 3^e−1elements may be used to determine the cubic root. Since

[0277] (αⁱA²)³=β(A²)³

[0278] we can determine (A^2m−Q)³by dividing (αⁱA²)³by β. There are at most 3^e−1possible values of kA², namely,

{kA}^{2^{2 m - Q}} = α^{\sum_{i = 1}^{e - 1} l_{k} j \frac{2^{2 m - 1}}{3^{e - j - 1}}}

[0279] for l_kjε{0,1,2}

[0280] such that

A^{2^{2 m - Q}} = α^{\sum_{i = 1}^{e - 1} l_{k} j \frac{2^{2 m - 1}}{3^{e - j - 1}}}

[0281] The look-up table is entered with the value k³A²and produces the value kA². Then, the cubic root of β is:

\frac{α^{i} A^{2^{2^{2 m - Q}}}}{{kA}^{2^{2 m - Q}}} = α^{i} α^{le \frac{(2^{2 m} - 1)}{3}}

Claims

What is claimed is:

1. A method of operating a data processing system to determine the locations of erroneous symbols in a code word over GF(2^2m+1), the method including the steps of:

i. receiving a code word with erroneous data symbols;

ii. determining an associated error locator polynomial;

iii. transforming the error locator polynomial to a degree three polynomial;

iv. substituting variables in the degree three polynomial to produce in an extension field GF(2^4m+2) an equation (y₁, y₂)³+(0,b)(y₁, y₂)+(0, a)=0;

v. substituting variables to produce a quadratic equation (z₁, z₂)²+(0,b) (z₁, z₂)+(0,a)³=0

where (z₁, z₂) is an element of GF(2^4m+2) and solving the quadratic equation for (z₁, z₂);

vi. determining the cube root, (s,t), of (z₁, z₂);

vii. determining the roots, s, t, and s+t, of the equation of step iv based on the value of (s,t) of step vi; and

viii. determining that the code word symbols associated with the code word locations that correspond to the roots s, t and s+t are erroneous; and

ix. based on the location information

a. correcting errors in the data symbols of the code word,

b. labeling the code word as uncorrectable, or

c. requesting that the code word is again received and repeating steps i-ix as necessary.

2. The method of claim 1 wherein the step of determining the cube root of (z₁,z₂) includes raising β=(z₁,z₂) to a power

u = \frac{2^{2 m + 2} - 1}{3}

to produce the element that is the cubic root of β.

3. The method of claim 1 wherein the step of determining the cube root of (z₁,z₂) includes the steps of:

i. raising β=(z₁,z₂) to a power L, where L is a factor of 2^2m−1;

ii. raising β to a power V where L+3V=2^Q;

iii. determining a cubic root of β^L;

iv. multiplying (β^L)^⅓ by β^V; and

v. raising (β^L)^⅓ β^Vto a power 2^2m−Qto produce the element that is the cubic root of (z₁, z₂).

4. The method of claim 1 wherein the step of determining the cube root of (z₁,z₂) includes the steps:

where m₀=2^km

raising (β^L)^uβ^Vto a power 2^2m−Qto produce the element that is the cubic root of (z₁, z₂).

i. raising β=(z₁,z₂) to a power L, where L is a factor of 2^2m−1;

ii. raising β^Lto a power

u = \frac{2^{m_{0} - 1} - 1}{3}

iii. raising the element β to a power V, where L+3V=2^Q

iv. multiplying (β^L)^uby β^V; and

5. A method of operating a data processing system to identify erroneous symbols in a code word, the method including the steps of:

to produce the cubic root of β; and

A. receiving a code word with erroneous data symbols and producing an associated degree-three polynomial with a term β³that is an element of GF(2^2m+1);

B. raising the element β³to a power

u = \frac{2^{2 m + 2} - 1}{3}

C. using the cubic root of β, identifying which data symbols of the received code word are erroneous and

i. correcting the errors in the data symbols of the code word,

ii. labeling the code word as uncorrectable, or

iii. requesting receipt of the code word again and repeating steps A-C as necessary.

6. A method of operating a data processing system to identify erroneous data symbols in a code word, the method including the steps of:

i. receiving a code word with erroneous data symbols and producing an associated degree-three polynomial with a term β³that is an element of GF(2^2m);

ii. raising the element β³to a power L, where L is a factor of 2^2m−1;

iii. raising , to a power V where L+3V=2^Q;

iv. using β^Lto enter a stored lookup table to retrieve the cubic root of β^L;

v. multiplying (β^L)^⅓ by β^V;

vi raising (β^L)^⅓ β^Vto a power 2^2m−Qto produce the cubic root of; and

vii using the cubic root of β, identifying which data symbols of the received code word are erroneous and

a. correcting the errors in the data symbols of the code word,

b. labeling the code word as uncorrectable, or

c. requesting receipt of the code word again and repeating steps i-vii as necessary.

7. A method of operating a data processing system to identify erroneous data symbols in a received code word, the method including the steps of:

i. receiving a code word with erroneous data symbols and producing an associated degree-three polynomial with a term β³that is an element of GF(2^2m);

ii. raising the element β to a power L, where L is a factor of 2^2m−1;

iii. raising β^Lto a power

u = \frac{2^{m_{0} - 1} - 1}{3}