US 7,010,739 B1Grant

Error evaluator for inversionless Berlekamp-Massey algorithm in Reed-Solomon decoders

Issue Date:2006-03-07

•12 Claims

•7 Drawing Sheets

Abstract

An error correcting Reed-Solomon decoder includes a syndrome calculator that calculates syndrome values. An error locator polynomial generator communicates with the syndrome calculator and generates an error locator polynomial. An error location finder communicates with at least one of the syndrome calculator and the error locator polynomial generator and generates error locations. An error values finder communicates with at least one of the syndrome calculator, the error location finder and the error locator polynomial generator and generates error values using an error value relationship that is not based on the traditional error evaluator polynomial. The error locator polynomial generator is an inversionless Berlekamp-Massey algorithm (iBMA), which calculates an error locator polynomial and a scratch polynomial. The error value relationship is based on the error locator polynomial and the scratch polynomial.

Metadata

Assignee

Marvell International Ltd.

Inventors

Weishi Feng
Zhan Yu

Application Information

Application Number:US 10/304,511

Filing Date:2002-11-26

Priority Date:2002-04-11

Art Unit:2133

Classifications

IPC:

H03M13/03

Patent Drawings (7 sheets)

Description

Cross-Reference to Related Applications

[0001] This application claims the benefit of U.S. Provisional Application No. 60/371,898, filed Apr. 11, 2002, which is hereby incorporated by reference.

Field of the Invention

[0002] The present invention relates to forward error correction (FEC), and more particularly to FEC using a Reed-Solomon decoder.

Background of the Invention

[0003] Many communications systems perform forward error correction (FEC) to improve data transmission accuracy and to ensure data integrity. FEC helps reduce bit error rates (BER) in applications such as data storage, digital video broadcasts, and wireless communications. Reed-Solomon (RS) error-correcting codes are commonly used for FEC.

[0004] Referring now to FIG. 1, a first device 10-1 communicates with a second device 10-2 over a communications channel 12. The communications channel can be hardwired or wireless. For example, the communications channel 12 can be an Ethernet network, a wireless local area network, a bus for a hard drive, etc. The first device 10-1 includes components 14-1 that output signals to a RS encoder 16-1 and that receive signals from a RS decoder 18-1. Likewise, the device 10-2 includes components 14-2 that output signals to a RS encoder 16-2 and that receive signals from a RS decoder 18-2. The components 14-1 of the first device 10-1 may be similar to or different than the components 14-2 of the second device 10-2. The encoders 16 encode the data before the data is output onto the communications channel 12. The encoders 16 insert redundant bits into the data stream. The decoders 18 use the redundant bits to detect and, when possible, to correct errors in the received data.

[0005] Referring now to FIG. 2, steps that are performed by a RS decoder are shown generally at 20. In step 22, the RS decoder computes syndrome values. In step 24, the RS decoder computes an error locator polynomial. The error locator polynomial can be calculated using a Berlekamp-Massey algorithm (BMA), inversionless BMA (iBMA), Euclidean algorithm, or other suitable algorithms. In step 26, the Reed-Solomon decoder calculates an error evaluator polynomial, which is typically based on the syndrome values and the error locator polynomial.

[0006] In step 28, the RS decoder finds error locations. For example, Chien's search algorithm, which will be described below, can be used. In step 30, error values are found. For example, Forney's algorithm, which will be described below, is often used to find the error values. Steps 24 and 28 may be performed in parallel in hardware implementations.

[0007] Referring now to FIG. 3, a RS decoder 32 typically includes a syndrome calculator 34 and an error locator polynomial generator 36. The Reed-Solomon decoder 32 also includes an error evaluator polynomial generator 38, an error location finder 40 and a error value finder 42. Control devices 44 and storage devices 46 may also be used to control decoding and to store data values for use by the RS decoder 32. The RS decoder 32 can be implemented using register-based VLSI, software and a processor, an application specific integrated circuit (ASIC) or in any other suitable manner.

[0008] In “High-speed Decoding of BCH Codes Using a New Error-Evaluation Algorithm,” T. Horiguchi, Electronics and Comm. in Japan, vol. 72, no. 12, (1989), an error evaluator for RS codes is computed during the BMA iterations. In “On the Determination of Error Values For Codes From a Class of Maximal Curves”, R. Koetter, Proc. 35^thAnnual Allerton Conference on Communications, Control and Computing, Univ. of Illinois at Urbana-Champaign (1997), an error evaluator for algebraic geometry codes is disclosed. The error evaluator in Koetter reduces to the error evaluator in, Horiguchi for RS codes.

[0009] The error evaluators in Horiguchi and Koetter require the BMA to be formulated in a manner that cannot be implemented in hardware easily. Berlekamp's formulation of the BMA has a more regular structure and can be readily implemented in hardware. A variation of the error evaluator for Berlekamp's formulation of the BMA is also disclosed in “On Decoding Reed-Solomon Codes up to and beyond the Packing Radii”, W. Feng, Ph. D. dissertation, Univ. of Illinois at Urbana-Champaign (1999).

[0010] A typical BMA employs a Galois field inverter, which computes a multiplicative inverse of Galois field elements. An inversionless (or division-free) BMA (iBMA) eliminates the Galois field inverter. One advantage of iBMA is the reduced delay of a critical path of a VLSI implementation.

[0011] RS codes operate on finite fields (GF(2^m)). GF(q) is a Galois field with q elements. c=(c_O, c₁, . . . , c_n−1) is a vector of length n over GF(q) where n=q−1. The Fourier transform of the vector c is C=(C₀, C₁, . . . , C_n−1), where

C_{j} = \sum_{i = 0}^{n - 1} c_{i} α^{ij}, j = 0, 1, \dots, n - 1

and α is a primitive element of GF(q). A t-error-correcting RS code is the collection of all vectors c with a Fourier transform satisfying C_m₀=C_m₀₊₁= . . . =C_m₀_2t−1=0 for some integer m₀. The code has minimum Hamming distance d_min=2t+1. Note that a RS code can be shortened to a length n<q−1 if needed.

[0012] A vector c=(c₀, c₁, . . . , c_n−1) can also be represented as a polynomial

c (x) = \sum_{i = 0}^{n - 1} c_{i} x^{i} .

The Fourier transform is a polynomial evaluation at x=α⁰,α¹, . . . ,αⁿ⁻¹. In other words, C_j=c(α^j). A t-error-correcting RS code is the collection of all vectors c such that c(α^m⁰)=c(α^m⁰⁺¹)= . . . =c(α^m⁰^+2t−1)=0. Therefore, every codeword is a multiple of
g(x)=(x−α^m⁰)(x−α^m⁰⁺¹) . . . (x−α^m⁰^+2t−1).
The polynomial g(x) is a generator polynomial of the Reed-Solomon code.

[0013] v=c+e is the received vector where e=(e₀, e₁, . . . , e_n−1) is the error vector and e_i≠0 when there is an error at the ith position. The codeword c and the error vector e are not known. The received vector v and the Fourier transform of c satisfy C_m₀= . . . =C_m₀_+2t−1=0. V and E are the Fourier transforms of v and e, respectively. Because the Fourier transform is a linear transformation,
V_i=C_i+E_ifor i=0, 1, . . . , n−1.
Because C_i=0 for i=m₀, m₀+1, . . . , m₀+2t−1, it follows that:
E_i=V_ifor i=m₀, m₀₊₁, . . . , m₀+2t−1.
Therefore, the decoder can compute E_ifrom the received vector for i=m₀, . . . , m₀+2t−1. E_iare denoted as S_i=E_l+m₀for i=0, . . . , 2t−1 and are called syndromes.

[0014] The RS decoder computes e from the 2t syndromes. The task can be divided into two parts. First, the error locations are found. In other words, the decoder finds all i such that e_i≠0. Second, the error values e_iare found for all of the error locations. Then, c can be recovered by subtracting e from v.

[0015] The RS decoder finds the error locator polynomial, which is defined as a polynomial Λ(x) satisfying:
Λ(0)=1; and Λ(α⁻¹)=0 if and only if e_i≠0.
The error locations are obtained by finding the zeros of Λ(x). To determine the error values, a decoder usually computes the error evaluator polynomial, which is given by Γ(x)=Λ(x)S(x) mod x^2twhere

S (x) = \sum_{i = 0}^{2 t - 1} S_{i} x^{i}

is syndrome polynomial.

[0016] If there are at most t errors, the error values can be determined (as set forth in G. D. Forney, Jr., “On Decoding BCH Codes”, IEEE Trans. Inform. Theory, vol. 11, pp. 547–557 (1965), which is hereby incorporated by reference):

e_{i} = {\begin{matrix} 0 & if Λ (α^{- i}) \neq 0 \\ \frac{x^{m_{0}} Γ (x)}{x Λ^{'} (x)} ❘_{x = α^{- i}} & if Λ (α^{- i}) = 0 \end{matrix}

where Λ′(x) is the formal derivative of Λ(x). This formula is known as the Forney algorithm.

[0017] In a first decoding step, assuming the received vector is fed into the decoder symbol by symbol, Homer's rule can be used to reduce complexity. In other words, E_jis evaluated as
E_j=( . . . ((v_n−1α^jv_n−2)αj)+ . . . +v₁)α^j+v₀.
Note that αⁿ=1 and that only S_j=E_m₀_+jfor j=0, 1, . . . , 2t−1 are needed. In a second decoding step, the error locator polynomial is calculated from the syndromes using any suitable algorithm such as Peterson-Gorenstein-Zierler algorithm, the BMA, and the Euclidean algorithm.

[0018] In a third decoding step, two approaches are generally used for computing the error evaluator polynomial if the BMA is used. A first approach computes the error evaluator polynomial simultaneously with the error locator polynomial using an iteration algorithm similar to the BMA. For example, see E. R. Berlekamp, Algebraic Coding Theory, New York, McGraw-Hill (1968). The cost of implementing this approach is high. A second approach computes the error evaluator polynomial after computing the error locator polynomial. The second approach is less complex than the first approach. However, decoding latency of the second approach is higher than the first approach.

[0019] In a fourth decoding step, Λ(x) and Γ(x) are known. The error values are determined using Forney's algorithm. Chien's search, which is used to find the zeros of the error locator polynomial Λ(x), is typically performed in parallel with the error evaluation in hardware implementations.

[0020] When using the BMA, a linear-feedback shift register (Λ, L) with length L and coefficients Λ₀=1, Λ₁, . . . , Λ_Lproduces the sequence S₀, S₁, S₃, . . . if:

S_{j} = - \sum_{i = 1}^{L} Λ_{i} S_{j - i} for j = L, L + 1, \dots .

If (Λ, L) produces S₀, . . . , S_r−1but does not produce S₀, . . . , S_r−1, S_r, then (Λ,L) has a discrepancy Δ_rat S_r, where

Δ_{r} = \sum_{i = 0}^{L} Λ_{i} S_{r - i} .

Λ (x) = \sum_{i = 0}^{L} Λ_{i} x^{i}

is the feedback polynomial. Note that the degree of Λ(x) is less than or equal to L because Λ_Lmight be zero.

[0021] The BMA finds the shortest linear feedback shift register that produces a given sequence. If the number of errors is less than or equal to t, the error locator polynomial is the polynomial of lowest degree that produces the syndrome sequence S₀, S₁, . . . , S_2t−1. If the number of errors is more than t, the BMA still finds the polynomial of lowest degree that generates the syndrome sequence. However, this polynomial is usually not a locator polynomial and the decoding algorithm fails.

[0022] Referring now to FIG. 4, one BMA implementation that is disclosed in Horiguchi and in J. L. Massey, “Shift-Register Synthesis and BCH Decoding”, IEEE Trans. Inform. Theory, vol. 15, no. 1 (1969), which is hereby incorporated by reference is shown. In step 50, variables are initialized (Λ(x)←1, B(x)←1, p←1, a←0, r←0, L←0, Δ_B←1). In step 54, control determines whether r=2t. If true, control ends in step 58. Otherwise, the discrepancy

Δ ⟵ \sum_{i = 0}^{L} Λ_{i} S_{r - i}

is computed in step 60.

[0023] If Δ=0 in step 62, then p←p+1 in step 64 and control continues with step 68. Otherwise, if Δ≠0 and 2L>r as determined in step 70, then Λ(x)←Λ(x)−ΔΔ_B⁻¹x^pB(x) in step 72 and control continues with step 64. If Δ≠0 and 2L≦r as determined in step 76, then T(x)←Λ(x), Λ(x)←Λ(x)−ΔΔ_B⁻¹x^pB(x), B(x)←T(x), a←r−2L, Δ_B←Δ, L←r+1−L, and p←1 in step 80. Control continues from steps 64, 76 and 80 with step 68 where r←r+1. Control continues from step 68 to step 54 until control ends in step 58. The parameter a in this algorithm is not disclosed in Massey. Parameter a was introduced by Horiguchi and is used for error evaluation.

[0024] u≦t is the number of errors. X_l=αⁱ^land Y^lare the error positions and the error values, respectively (that is Λ(α⁻ⁱ^l)=0, i=1, . . . , u). At the end of this algorithm:
{circumflex over (B)}(x)=Δ_B⁻¹x^αB(x).
The error values are given by Horiguchi as follows:

Y_{l} = \frac{X_{l}^{- 2 u - m_{0} + 3}}{\hat{B} (X_{l}^{- 1}) Λ^{'} (X_{l}^{- 1})} .

The structure of this formulation is not easily implemented in register-based VLSI in which the coefficients of Λ(x) and B(x) are stored in registers. When updating the polynomial Λ(x) in steps 76 and 80, the relationship between the coefficients of Λ(x) and B(x) depends on the variable p. In other words, Λ_i=Λ_i−ΔΔ_B⁻¹B_i−p, where p could be 1, or 2, or 3, . . . x. This means that the circuit connections are not fixed. While multiplexers can be used to avoid this problem, this approach is not sufficiently cost effective.

[0025] Referring now to FIG. 5, Berlekamp's formulation with a minor modification has a more regular structure and is desirable for register-based VLSI implementations. The algorithm is disclosed in R. E. Blahut, Theory and Practice of Error Control Codes, Reading, Mass., Addison-Wesley Publishing Company (1983), which is hereby incorporated by reference. The algorithm is similar to Berlekamp's formulation.

[0026] The syndrome sequence S₀, S₁, . . . , S_2t−1and t are used as inputs. In step 100, initialization of variables is performed (Λ(x)←1, B(x)←1, r←0, L←0). In step 104, if r=2t, control ends in step 108. Otherwise, the discrepancy

Δ ⟵ \sum_{i = 0}^{L} Λ_{i} S_{r - i}

is computed in step 110. If Δ≠0 and 2L≦r in step 112, then δ←1, L←r+1−L in step 114. Otherwise, δ←0 in step 116. In step 120, the polynomials are updated as follows:

(\begin{matrix} Λ (x) \\ B (x) \end{matrix}) ⟵ (\begin{matrix} 1 - Δ x \\ Δ^{- 1} δ (1 - δ) x \end{matrix}) (\begin{matrix} Λ (x) \\ B (x) \end{matrix})

In step 124, r←r+1 and control continues with step 104.

[0027] At each clock cycle, the register bank for B(x) either shifts right B_i←B_i−1corresponding to B(x)←xB(x), or parallel loads B_i←Δ⁻¹Λ_icorresponding to B(x)←Δ⁻¹Λ(x). The register bank for Λ(x) is updated by Λ_i←Λ_i−ΔB_i−1. Note that this algorithm does not provide an error evaluator. In W. Feng, “On Decoding Reed-Solomon Codes Up to and Beyond the Packing Radii”, Reading, Ph.D. dissentation, Univ. of IL at Urbana-Champaign (1999), which is hereby incorporated by reference, an error evaluator for this BMA formulation was derived as follows:

e_{i} = {\begin{matrix} 0 & if Λ (α^{- i}) \neq 0, \\ \frac{α^{- i (m_{0} + 2 t - 2)}}{B^{(2 t)} (α^{- i}) Λ^{'} (α^{- i})} & if Λ (α^{- i}) = 0, \end{matrix}

or equivalently

Y_{l} = \frac{X_{l}^{- (m_{0} + 2 t - 2)}}{B^{(2 t)} (X_{l}^{- 1}) Λ^{'} (X_{l}^{- 1})} .

where B^(2t)(x) is the scratch polynomial at the end of 2t iterations of the algorithm shown in FIG. 5. This evaluator is functionally equivalent to the Horiguchi-Koetter evaluator.

[0028] Referring now to FIG. 6, the iBMA is a modification of the algorithm in FIG. 5. The syndrome sequence S₀, S₁, . . . , S_2t−3and t are inputs. In step 130, Λ(x)←1, B(x)←1, r←0, L←0, Δ_B←1. In step 134, if r=2t, control ends in step 136. Otherwise in step 138, the discrepancy

Δ ⟵ \sum_{i = 0}^{L} Λ_{i} S_{r - i}

is computed. If Δ≠0 as determined in step 142, Λ(x)←Δ_BΛ(x)+ΔxB(x) in step 144. Otherwise, Λ(x)←Λ(x) in step 146. If Δ≠0 and 2 L≦r as determined in step 146, B(x)←Λ(x), L←r+1−L and Δ_B←Δ in step 148. Otherwise, B(x)←xB(x) in step 150. In step 154, r←r+1. Control continues from step 154 to step 134.

[0029] Note that the Λ(x) produced by the algorithm in FIG. 6 is a constant multiple of the Λ(x) produced by the algorithm of FIG. 5 and no longer satisfies the condition of Λ(0)=1. Λ(x) is still referred to as the error-locator polynomial because it still “locates” the errors. In other words, the roots of this polynomial point to the error locations if the number of errors in the received vector is less than or equal to the correction power t.

[0030] As can be appreciated from the forgoing, the steps performed by RS decoders can be complex and can involve a large number of calculations. Reducing the number of calculations and increasing the speed of RS decoding would be desirable.

Summary of the Invention

[0031] An error correcting Reed-Solomon decoder according to the present invention includes a syndrome calculator that calculates syndrome values. An error locator polynomial generator communicates with the syndrome calculator and generates an error locator polynomial. An error location finder communicates with at least one of the syndrome calculator and the error locator polynomial generator and generates error locations. An error values finder communicates with at least one of the syndrome calculator, the error location finder and the error locator polynomial generator and generates error values without calculating an error evaluator polynomial.

[0032] In other features, the error locator polynomial generator is an inversionless Berlekamp-Massey algorithm (iBMA). The error location finder finds zeros in the error locator polynomial. The iBMA calculates an error locator polynomial and a scratch polynomial. The error values are calculated using an error value relationship that is based on the error locator polynomial and the scratch polynomial.

[0033] An error correcting Reed-Solomon decoder according to the present invention includes an error locator polynomial generator that generates an error locator polynomial and a scratch polynomial based on an inversionless Berlekamp-Massey algorithm (iBMA). An error location finder communicates with the error locator polynomial generator and generates error locations. An error values finder communicates with at least one of the error location finder and the error locator polynomial generator and generates error values using an error value relationship that is based on the error locator polynomial and the scratch polynomial.

[0034] Further areas of applicability of the present invention will become apparent from the detailed description provided hereinafter. It should be understood that the detailed description and specific examples, while indicating the preferred embodiment of the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention.

Brief Description of the Drawings

[0035] The present invention will become more fully understood from the detailed description and the accompanying drawings, wherein:

[0036] FIG. 1 is a functional block diagram of first and second devices that include RS encoders/decoders according to the prior art;

[0037] FIG. 2 illustrates steps for RS decoding according to the prior art;

[0038] FIG. 3 is a functional block diagram of a RS decoder according to the prior art;

[0039] FIG. 4 illustrates steps of a BMA disclosed by Massey and Horiguchi for calculating an error locator polynomial according to the prior art;

[0040] FIG. 5 illustrates steps of Berlekamp's formulation of the BMA for calculating an error locator polynomial according to the prior art;

[0041] FIG. 6 illustrates steps of an inversionless BMA (iBMA) for calculating an error locator polynomial according to the prior art;

[0042] FIG. 7 illustrates steps performed by a RS decoder according to the present invention; and

[0043] FIG. 8 is a functional block diagram of a RS decoder according to the present invention.

Detailed Description of the Preferred Embodiments

[0044] The following description of the preferred embodiment(s) is merely exemplary in nature and is in no way intended to limit the invention, its application, or uses. For purposes of clarity, the same reference numbers will be used in the drawings to identify similar elements.

[0045] The calculation of the traditional error evaluator polynomial increases the amount of time and/or more hardware that is required to perform RS decoding. The present invention discloses a RS decoder that eliminates the need for the error evaluator polynomial. As a result, the error evaluator need not be calculated and the amount of time that is required to perform RS decoding is significantly reduced.

[0046] The RS decoder according to the present invention calculates error values based on the error-locator polynomial and the scratch polynomial that are generated by the inversionless Berlekamp-Massey (iBMA) algorithm. In addition to reduced decoding time, the hardware that is required to implement the RS decoder according to the present invention is significantly reduced.

[0047] Referring now to FIG. 7, steps that are performed by a RS decoder according to the present invention are shown generally at 200. In step 210, the RS decoder computes syndrome values as previously described above. In step 214, the RS decoder computes an error locator polynomial using an inversionless Berlekamp-Massey algorithm (iBMA), such as the iBMA algorithm shown in FIG. 6.

[0048] In step 218, the RS decoder finds error locations. For example, Chien's search algorithm can be used. If the degree of ^(x) is small, the roots of ^(x) can be found directly without doing a search. In step 222, error values are found using an error value relationship according to the present invention that will be described further below.

[0049] At the end of iBMA algorithm in FIG. 5, the error values can be calculated by the error values finder 248 using the error value relationship set forth below:

e_{i} = {\begin{matrix} 0 & if Λ (α^{- i}) \neq 0, \\ \frac{Λ_{0} Δ_{B} α^{- i (m_{0} + 2 t - 1)}}{α^{- i} B^{(2 t)} (α^{- i}) Λ^{'} (α^{- i})} & if Λ (α^{- i}) = 0, \end{matrix}

or equivalently

Y_{l} = \frac{Λ_{0} Δ_{B} X_{l}^{- (m_{0} + 2 t - 1)}}{X_{l} B^{(2 t)} (X_{l}^{- 1}) Λ^{'} (X_{l}^{- 1})} .

where Λ₀is the coefficient of the 0-th degree term of Λ(x), m₀and t are integers defining the Reed-Solomon code, Δ_Bis a discrepancy value, α is a primitive element of GF(q), B^(2t)(x) is the scratch polynomial at the end of iBMA, and Λ′(x) is the formal derivative of the error locator polynomial Λ(x).

[0050] Referring now to FIG. 8, a t-error correcting RS decoder 230 according to the present invention includes a syndrome calculator 234 that generates syndrome values S₀, S₁, . . . , S_2t-1. An error locator polynomial generator 236 generates the error locator polynomial using iBMA. An error locations finder 244, such as Chien's search or other error locations finding algorithm, locates error locations.

[0051] An error values finder 248 locates errors using the error value relationship according to the present invention. The error values finder 248 is not based on the error evaluator polynomial and therefore does not require the error evaluator polynomial to be calculated. Control 250 coordinates and controls decoding and storage 252 (such as registers, latches, memory or other electronic storage) stores data values for use by the RS decoder 230.

[0052] The RS decoder 230 can be a register-based VLSI circuit, software and a processor, an application specific integrated circuit (ASIC) and/or combinations thereof. One suitable register-based implementation of the error value relationship disclosed above and the iBMA is disclosed in “Efficient High-Speed Reed-Solomon Decoder”, U.S. patent application Ser. No. 10/305,091, filed Nov. 26, 2002, which is hereby incorporated by reference. As can be appreciated, some of the components of the RS decoder 230 can share multipliers and/or other elements to reduce cost.

[0053] Those skilled in the art can now appreciate from the foregoing description that the broad teachings of the present invention can be implemented in a variety of forms. Therefore, while this invention has been described in connection with particular examples thereof, the true scope of the invention should not be so limited since other modifications will become apparent to the skilled practitioner upon a study of the drawings, the specification and the following claims.

Claims

Certificate of Correction present

The claims shown below may not include correction changes. Use the corrected PDF for authoritative text.

What is claimed is:

1. An error correcting Reed-Solomon decoder, comprising:

a syndrome calculator that calculates syndrome values;

an error locator polynomial generator that communicates with said syndrome calculator and that generates an error locator polynomial;

an error location finder that communicates with at least one of said syndrome calculator and said error locator polynomial generator and that generates error locations; and

an error values finder that communicates with at least one of said syndrome calculator, said error location finder and said error locator polynomial generator and that generates error values without calculating an error evaluator polynomial.

2. The error correcting Reed-Solomon decoder of claim 1 wherein said error locator polynomial generator is an inversionless Berlekamp-Massey algorithm (iBMA).

3. The error correcting Reed-Solomon decoder of claim 2 wherein said iBMA generates an error locator polynomial and a scratch polynomial and wherein said error values finder calculates said error values using an error value relationship that is based on said error locator polynomial and said scratch polynomial.

4. The error correcting Reed-Solomon decoder of claim 1 wherein said error location finder finds zeros in said error locator polynomial.

5. An error correcting Reed-Solomon decoder, comprising:

syndrome calculating means for calculating syndrome values;

error locator polynomial generating means that communicates with said syndrome calculating means for generating an error locator polynomial;

error location finding means that communicates with at least one of said syndrome calculating means and said error locator polynomial generating means for generating error locations; and

error values finding means that communicates with at least one of said syndrome calculating means, said error location finding means and said error locator polynomial generating means for generating error values without calculating an error evaluator polynomial.

6. The error correcting Reed-Solomon decoder of claim 5 wherein said error locator polynomial generating means implements an inversionless Berlekamp-Massey algorithm (iBMA).

7. The error correcting Reed-Solomon decoder of claim 6 wherein said iBMA generates an error locator polynomial and a scratch polynomial and wherein said error values finding means calculates said error values using an error value relationship that is based on said error locator polynomial and said scratch polynomial.

8. The error correcting Reed-Solomon decoder of claim 5 wherein said error location finding means finds zeros in said error locator polynomial.

9. A method for performing error correction in a Reed-Solomon decoder, comprising:

calculating syndrome values;

generating an error locator polynomial based on said syndrome values;

identifying error locations using said error locator polynomial; and

identifying error values without calculating an error evaluator polynomial.

10. The method of claim 9 wherein said error locator polynomial is generated using an inversionless Berlekamp-Massey algorithm (iBMA).

11. The method of claim 10 wherein said step of generating said error locator polynomial includes:

generating an error locator polynomial; and

generating a scratch polynomial, and wherein said error values are identified using an error value relationship that is based on said error locator polynomial and said scratch polynomial.

12. The method of claim 9 wherein said step of identifying error locations includes finding zeros in said error locator polynomial.

Patent Citations (12)

Patent	Date	Inventor	Cited By
US4839896(A)	1989-06-01	Glover et al.	Examiner
US6092233(A)	2000-07-01	Yang	Applicant
US6119262(A)	2000-09-01	Chang et al.	Examiner
US6122766(A)	2000-09-01	Fukuoka et al.	Applicant
US6209115(B1)	2001-03-01	Truong et al.	Applicant
US6256763(B1)	2001-07-01	Oh et al.	Applicant
US6286123(B1)	2001-09-01	Kim	Applicant
US6317858(B1)	2001-11-01	Cameron	Examiner
US6347389(B1)	2002-02-01	Boyer	Applicant
US6374384(B1)	2002-04-01	Ohta et al.	Applicant
US6378104(B1)	2002-04-01	Okita	Applicant
WO99/09694	1999-02-01		Applicant

Non-Patent Literature (11)

Sarwate et al., High speed architecture for Reed Solomon Decoders, Oct. 2001, IEEE Trans. on VLSI sys. vol. 9, No. 5, p. 641 655.Examiner
Bose-Chaudhuri-Hocquenghme Codes, 7.5 Fast Decoding of BCH Codes, pp. 186-187.Applicant
Feng, On Decoding Reed-Solomon Codes Up To And Beyond The Packing Radii, Chapter 2, University of Illinois at Urbana-Champaign, 1999, pp. 7-32.Applicant
Forney, Jr., “On Decoding BCH Codes”, IEEE Transactions on Information Theory, vol. 11, No. 4, Oct. 1965, pp. 549-557.Applicant
Horiguchi, “High-Speed Decoding of BCH Codes Using a New Error-Evaluation Algorithm”, Electronics and Communications in Japan, Part 3, vol. 72, No. 12, 1989, pp. 63-71.Applicant
Massey, “Shift-Register Synthesis and BCH Decoding”, IEEE Transactions on Information Theory, vol. 15, No. 1, Jan. 1969, pp. 122-127.Applicant
Chien, “Cycle Decoding Procedures for Bose-Chaudhuri-Hocquenghem Codes”, IEEE Transactions on Information Theory, Oct. 1964, pp. 357-363.Applicant
Elwyn R. Berlekamp, “Algebraic Coding Theory, Binary BHC Codes for Correcting Multiple Errors”, 1968, pp. 176-196.Applicant
Ralf Koetter, “On the determination of error values for codes from a class of maximal curves”, University of Illinois Coordinated Science Laboratory, Urbana, Illinois, pp. 44-53.Applicant
Truong/Jeng/Hung, “Inversionless Decoding of Both Errors and Erasures of Reed-Solomon Code”, IEEE Transactions on Communications, vol. 46, No. 8, Aug. 1998, pp. 973-976.Applicant
U.S. Appl. No. 10/305,091, filed Nov. 26, 2002, Zhan Yu et al.Applicant

Abstract

Description

Cross-Reference to Related Applications

[0001] This application claims the benefit of U.S. Provisional Application No. 60/371,898, filed Apr. 11, 2002, which is hereby incorporated by reference.

Field of the Invention

[0002] The present invention relates to forward error correction (FEC), and more particularly to FEC using a Reed-Solomon decoder.

Background of the Invention

C_{j} = \sum_{i = 0}^{n - 1} c_{i} α^{ij}, j = 0, 1, \dots, n - 1

[0012] A vector c=(c₀, c₁, . . . , c_n−1) can also be represented as a polynomial

c (x) = \sum_{i = 0}^{n - 1} c_{i} x^{i} .

S (x) = \sum_{i = 0}^{2 t - 1} S_{i} x^{i}

is syndrome polynomial.

e_{i} = {\begin{matrix} 0 & if Λ (α^{- i}) \neq 0 \\ \frac{x^{m_{0}} Γ (x)}{x Λ^{'} (x)} ❘_{x = α^{- i}} & if Λ (α^{- i}) = 0 \end{matrix}

where Λ′(x) is the formal derivative of Λ(x). This formula is known as the Forney algorithm.

[0020] When using the BMA, a linear-feedback shift register (Λ, L) with length L and coefficients Λ₀=1, Λ₁, . . . , Λ_Lproduces the sequence S₀, S₁, S₃, . . . if:

S_{j} = - \sum_{i = 1}^{L} Λ_{i} S_{j - i} for j = L, L + 1, \dots .

If (Λ, L) produces S₀, . . . , S_r−1but does not produce S₀, . . . , S_r−1, S_r, then (Λ,L) has a discrepancy Δ_rat S_r, where

Δ_{r} = \sum_{i = 0}^{L} Λ_{i} S_{r - i} .

Λ (x) = \sum_{i = 0}^{L} Λ_{i} x^{i}

is the feedback polynomial. Note that the degree of Λ(x) is less than or equal to L because Λ_Lmight be zero.

Δ ⟵ \sum_{i = 0}^{L} Λ_{i} S_{r - i}

is computed in step 60.

Y_{l} = \frac{X_{l}^{- 2 u - m_{0} + 3}}{\hat{B} (X_{l}^{- 1}) Λ^{'} (X_{l}^{- 1})} .

Δ ⟵ \sum_{i = 0}^{L} Λ_{i} S_{r - i}

(\begin{matrix} Λ (x) \\ B (x) \end{matrix}) ⟵ (\begin{matrix} 1 - Δ x \\ Δ^{- 1} δ (1 - δ) x \end{matrix}) (\begin{matrix} Λ (x) \\ B (x) \end{matrix})

In step 124, r←r+1 and control continues with step 104.

e_{i} = {\begin{matrix} 0 & if Λ (α^{- i}) \neq 0, \\ \frac{α^{- i (m_{0} + 2 t - 2)}}{B^{(2 t)} (α^{- i}) Λ^{'} (α^{- i})} & if Λ (α^{- i}) = 0, \end{matrix}

or equivalently

Y_{l} = \frac{X_{l}^{- (m_{0} + 2 t - 2)}}{B^{(2 t)} (X_{l}^{- 1}) Λ^{'} (X_{l}^{- 1})} .

where B^(2t)(x) is the scratch polynomial at the end of 2t iterations of the algorithm shown in FIG. 5. This evaluator is functionally equivalent to the Horiguchi-Koetter evaluator.

Δ ⟵ \sum_{i = 0}^{L} Λ_{i} S_{r - i}