Maximum likelihood sequence estimating device and method therefor

Claim 2

A maximum likelihood sequence estimator comprising sampling means for sampling reception signals entered from transmission lines at a symbol interval T with respect to respective sampling phases of simple non-increasing or simple non-decreasing sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), obtaining sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals y(t), and outputting said sample value sequences obtained to correlation value calculating means, correlation value calculating means for obtaining complex cross-correlation functions C i of training sequences on the basis of said sample value sequences i and training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i according to the following equation, and (where, i = 1, 2, ..., M; * means that complex conjugation is used.) outputting these functions to the maximum correlation index selecting means, maximum correlation index selecting means for comparing the absolute values of said M number of complex cross-correlation functions C i , obtaining a sampling phase τ (I k ) where said absolute values are maximized and outputting the index I k of said sampling phase τ (I k ) to a sampling phase smoothing means, sampling phase smoothing means for obtaining a sectional weighted mean value I avg (k), a sectional maximum value I max (k) and a sectional minimum value I min (k) of said index I k according to the following equations at the time of the kth sampling phase selection, (where, w a (j) (j = 1, 2, ..., Ls) is a weighted coefficient and L s denotes the maximum length of calculation) a selective sampling phase index I s according to the following equation, and Is = [I avg (k) + w s (I max (k) - I min (k))] (where, [ ] indicates the Gaussian notation which returns the largest integer which does not exceed a value in the notation and w s denotes a shift coefficient) a sampling phase τ s corresponding to said index I s according to the following equation and outputting said sampling phases τ s to a sampling processing part, a delay processing part for delaying said reception signals entered from said transmission lines as long as specified and outputting the signals to a sampling processing part, a sampling processing part for obtaining a sampling value sequence = {yn} (n = 1, 2, ..., N) by sampling said reception signals from said delay processing part with respect to a symbol interval T and said sampling phase τ s and outputting said sample value sequence to a transmission estimating part and a Viterbi algorithm processing part, a transmission line estimating part for obtaining estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences and estimated values E from estimated values E = {Exn} (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences and outputting said estimated values E , and a Viterbi algorithm processing part for obtaining said estimated values E from said sample value sequences on the basis of said estimated values E according to the Viterbi algorithm and outputting said estimated values.

Claim 3

A maximum likelihood sequence estimating method for obtaining and outputting sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals by sampling reception signals entered from transmission lines at a symbol interval T with respect to respective sampling phases of simple non-increasing or simple non-decreasing sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), complex cross-correlation functions {C i } (i = 1, 2, ..., M) of training sequences by said sample value sequences i and said training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i , a sampling phase τ (I k ) where the absolute values of said M number of complex cross-correlation functions C i are maximized by comparing said absolute values, an index I k of said sampling phase τ (Ik), a sectional weighted mean value, a sectional maximum value and a sectional minimum value of said index I k , an index I s obtained by adding a value, which is obtained by subtracting said sectional minimum vale from said maximum value and multiplying a resultant value by a constant, to said sectional weighted mean value, said sampling phase τ s corresponding to said index I s , sample value sequences = {y n } (n = 1, 2, ..., N) by sampling said reception signals entered at specified timings with respect to a symbol interval T and said sampling phase τ s , estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences , estimated values E from estimated values E = {Ex n } (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences , and said estimated values E from said sample value sequences according to the Viterbi algorithm in accordance with said estimated values E .

Claim 4

A maximum likelihood sequence estimator comprising sampling means for sampling reception signals entered from transmission lines at a symbol interval T with respect to respective sampling phases of simple non-increasing or simple non-decreasing sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), obtaining sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals, and outputting said sample value sequences obtained, correlation value calculating means for obtaining complex cross-correlation functions {C i } (i = 1, 2, ..., M) of training sequences by said sample value sequences i and training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i , maximum correlation index selecting means for comparing the absolute values of said M number of complex cross-correlation functions C i to obtain a sampling phase τ (I k ) where said absolute values are maximized and outputting an index Ik of said sampling phase τ (I k ), sampling phase smoothing means for obtaining a weighted mean value, a maximum value and a minimum value of said index I k , an index I s obtained by adding a value, which is obtained by subtracting said minimum value from said maximum value and multiplying a resultant value by a constant, to said weighted mean value and said sampling phase τ s corresponding to said index I s , and outputting said sampling phase τ s , a sampling processing part for sampling said reception signals entered at specified timings with respect to a symbol interval T and said sampling phase τ s to obtain sample value sequences = {y n } (n = 1, 2, ..., N) and outputting said sample value sequences, a transmission line estimating part for obtaining estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences and estimated values E from estimated values E = {Ex n } (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences and outputting said estimated values E , and a Viterbi algorithm processing part for obtaining said estimated values E from said sample value sequences on the basis of said estimated values E according to the Viterbi algorithm and outputting said estimated values.

Claim 5

A maximum likelihood sequence estimator comprising sampling means for sampling reception signals entered from transmission lines at a symbol interval T with respect to respective sampling phases of simple non-increasing or simple non-decreasing sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), obtaining sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals y(t), and outputting said sample value sequences obtained to correlation value calculating means, correlation value calculating means for obtaining complex cross-correlation functions C i of training sequences on the basis of said sample value sequences i and training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i according to the following equation (where, i = 1, 2, ..., M; * means that complex conjugation is used.) and outputting these functions to the maximum correlation index selecting means, maximum correlation index selecting means for comparing the absolute values of said M number of complex cross-correlation functions C i , obtaining a sampling phase τ (I k ) where said absolute values are maximized and outputting the index I k of said sampling phase τ (I k ) to a sampling phase smoothing means, sampling phase smoothing means for obtaining a weighted mean value I AVG (k), a maximum value I MAX (k) and a minimum value I MIN (k) of said index I k according to the following equations at the time of the kth sampling phase selection, I MAX (k) = max I j 1 ≦ j ≦ k I MIN (k) = min I j 1 ≦ j ≦ k I AVG (k) = (1 - α) I AVG (k - 1) + α Ik (α:constant) I AVG (o) = 0 selective sampling phase index I s according to the following equation, and I s = [I AVG (k) + W s (I MAX (k) - I MIN (k))] (where, [ ] indicates the Gaussian notation which returns back the largest integer which does not exceed a value in the notation and w s denotes a shift coefficient) a sampling phase τ s corresponding to said index I s according to the following equation and outputting said sampling phases τ s to a sampling processing part, a delay processing part for delaying said reception signals entered from said transmission lines as long as specified and outputting the signals to a sampling processing part, a sampling processing part for obtaining a sampling value sequence = {yn} (n = 1, 2, ..., N) by sampling said reception signals from said delay processing part with respect to a symbol interval T and said sampling phase τ s and outputting said sample value sequence to a transmission estimating part and a Viterbi algorithm processing part, a transmission line estimating part for obtaining estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences and estimated values E from estimated values E = {Exn} (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences and outputting said estimated values E , and a Viterbi algorithm processing part for obtaining said estimated values E from said sample value sequences on the basis of said estimated values E according to the Viterbi algorithm and outputting said estimated values.

Claim 6

A maximum likelihood sequence estimating method for obtaining and outputting sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals by sampling reception signals entered from transmission lines at a symbol interval T with respect to respective sampling phases of simple non-increasing or simple non-decreasing sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), complex cross-correlation functions {C i } (i = 1, 2, ..., M) of training sequences by said sample value sequences i and said training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i , a sampling phase τ (I k ) where the absolute values of said M number of complex cross-correlation functions C i are maximized by comparing said absolute values, an index I k of said sampling phase τ (Ik), a weighted mean value, a maximum value and a minimum value of said index I k , an index I s obtained by adding a value, which is obtained by subtracting said minimum vale from said maximum value and multiplying a resultant value by a constant, to said weighted mean value, said sampling phase τ s corresponding to said index I s , sample value sequences = {y n } (n = 1, 2, ..., N) by sampling said reception signals entered at specified timings with respect to a symbol interval T and said sampling phase τ s , estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences , estimated values Eh from estimated values E = {Ex n } (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences , and said estimated values E from said sample value sequences according to the Viterbi algorithm in accordance with said estimated values E .

Claim 7

A maximum likelihood sequence estimator comprising sampling means for sampling reception signals entered from transmission lines at a symbol interval T with respect to respective sampling phases of sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), starting from a specified estimated leading time determined by a frame synchronization mechanism, obtaining sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals, and outputting said sample value sequences obtained, correlation value calculating means for obtaining and outputting complex cross-correlation functions {C i } (i = 1, 2, ..., M) of training sequences by said sample value sequences i and training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i , maximum correlation sampling phase selecting means for obtaining a sampling phase τ (I k ) where the absolute values of said M number of complex cross-correlation functions C i are maximized by comparing said absolute values and outputting said sampling phase τ (I k ), sampling phase smoothing means for obtaining and outputting a coordinate transformation term in accordance with said estimated leading time, a sectional weighted mean value, a sectional maximum value and a sectional minimum value of said sampling phase τ (I k ) after adding said coordinate transformation term to said sampling phase τ (I k ), and a sampling phase τ s obtained by adding a value, which is obtained by subtracting the absolute value of said coordinate transformation term and said sectional minimum value and multiplying a resultant value by a constant, to said sectional weighted mean value and subtracting said coordinate transformation term from said resultant value, a sampling processing part for sampling said reception signals entered at specified timings with respect to a symbol interval T and said sampling phase τ s , starting from said estimated leading time to obtain and output the sample value sequences = {y n } (n = 1, 2, ..., N), a transmission line estimating part for obtaining estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences and estimated values E from estimated values E = {Ex n } (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences and outputting said estimated values E , and a Viterbi algorithm processing part for obtaining said estimated values E from said sample value sequences on the basis of said estimated values E according to the Viterbi algorithm and outputting said estimated values.

Claim 8

A maximum likelihood sequence estimator comprising sampling means for sampling reception signals entered from transmission lines at a symbol interval T with respect to respective sampling phases of sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), starting from a estimated leading time Et slot (k) determined by a frame synchronization mechanism when in processing the kth assigned time slot, obtaining sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of said reception signals y(t), and outputting said sample value sequences obtained to correlation value calculating means, correlation value calculating means for obtaining complex cross-correlation functions C i of training sequences on the basis of said sample value sequences i and training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i according to the following equation and (where, i = 1, 2, ..., M; * means that complex conjugation is used.) outputting these functions to the maximum correlation index selecting means, maximum correlation sampling phase selecting means for comparing the absolute values of said M number of complex cross-correlation functions C i , obtaining a sampling phase τ (I k ) where said absolute values are maximized and outputting said sampling phase τ (I k ) to a sampling phase smoothing means, sampling phase smoothing means for obtaining a coordinate transformation term tc (k) according to the following equations, Δ tc (k) = Et slot (k) - Et slot (k-1)   k > 1 a sectional weighted mean value τ avg (k), a sectional maximum value τ max (k) and a sectional minimum value τ min (k) of said sampling phase τ (I k ) according to the following equations, (where, w a (j) (j = 1, 2, ..., L s ) is a weighted coefficient and L s denotes the maximum calculation length) a selective sampling phase τ s according to the following equation, and τ s = τ avg (k)-tc (k)+w s (τ max (k)-τ min (k) | tc (k) |) (where, w s indicates the shift coefficient, w a (j) (j = 1, 2, ..., L s ) is a weighted coefficient and L s denotes the maximum calculation length) outputting said sampling phase τ s to a sampling processing part, a delay processing part for delaying said reception signals entered from said transmission lines as long as specified and outputting the signals to a sampling processing part, a sampling processing part for obtaining sampling value sequences = {y n } (n = 1, 2, ..., N) by sampling said reception signals from said delay processing part with respect to a symbol interval T and said sampling phase τ s and outputting said sample value sequences to a transmission estimating part and a Viterbi algorithm processing part, a transmission line estimating part for obtaining estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences and estimated values E from estimated values E = {Exn} (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences and outputting said estimated values E , and a Viterbi algorithm processing part for obtaining said estimated values E from said sample value sequences on the basis of said estimated values E according to the Viterbi algorithm and outputting said estimated values.

Claim 9

A maximum likelihood estimating method for obtaining and outputting sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals by sampling reception signals entered from transmission lines at a symbol interval T, starting from the estimated leading time determined by the frame synchronization mechanism with respect to individual sampling phases of sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), complex cross-correlation functions {C i } (i = 1, 2, ..., M) of training sequences by said sample value sequences i and said training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i , a sampling phase τ (I k ) where the absolute values of said M number of complex cross-correlation functions C i are maximized by comparing said absolute values, a coordinate transformation term based on said estimated leading time, a sectional weighted mean value, a sectional maximum value and a sectional minimum value of said sampling phase τ (I k ) after adding said coordinate transformation term to said sampling phase τ (I k ), a sampling phase τ s obtained by adding a value, which is obtained by subtracting said sectional minimum value from said sectional maximum value and multiplying a resultant value by a constant, to said sectional weighted mean value and subtracting said coordinate transformation term from said resultant value, sample value sequences = {y n } (n = 1, 2, ..., N) by sampling said reception signals y(t) entered at specified timings with respect to a symbol interval T and said sampling phase τ s , starting from said estimated leading time, estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences , estimated values E from estimated values E = {Ex n } (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences , and said estimated values E from said sample value sequences according to the Viterbi algorithm in accordance with said estimated values E .

Claim 10

A maximum likelihood sequence estimator comprising sampling means for sampling reception signals entered from transmission lines at a symbol interval T, starting from a estimated leading time determined by a frame synchronization mechanism, with respect to respective sampling phases of sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), obtaining sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals, and outputting said sample value sequences obtained, correlation value calculating means for obtaining complex cross-correlation functions {C i } (i = 1, 2, ..., M) of training sequences by said sample value sequences i and training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i , maximum correlation sampling phase selecting means for obtaining a sampling phase τ (I k ) where the absolute values of said M number of complex cross-correlation functions C i are maximized by comparing said absolute values and outputting said sampling phase τ (I k ), sampling phase smoothing means for obtaining first and second coordinate transformation terms in accordance with said estimated leading time, obtaining a sectional weighted mean value, a sectional maximum value and a sectional minimum value of said sampling phase τ after adding said second coordinate transformation term to said sampling phase τ (I k ), obtaining a sampling phase τ s calculated by adding a value, which is obtained by subtracting the absolute value of said first coordinate transformation term and said sectional minimum value from said sectional maximum value and multiplying a resultant value by a constant, to said sectional weighted mean value, and outputting said sampling phase τ s , a sampling processing part for obtaining sample value sequences = {y n } (n = 1, 2, ..., N) by sampling said reception signals entered at specified timings with respect to a symbol interval T and said sampling phase τ s , starting from said estimated leading time and outputting said sample value sequences, a transmission line estimating part for obtaining estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences and estimated values E from estimated values E = {Ex n } (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences and outputting said estimated values E , and a Viterbi algorithm processing part for obtaining said estimated values E from said sample value sequences on the basis of said estimated values E according to the Viterbi algorithm and outputting said estimated values.

Claim 11

A maximum likelihood sequence estimator comprising sampling means for sampling reception signals entered from transmission lines at a symbol interval T with respect to respective sampling phases of sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), starting from a estimated leading time Et slot (k) determined by a frame synchronization mechanism when in processing the kth assigned time slot, obtaining sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of said reception signals y(t), and outputting said sample value sequences obtained to correlation value calculating means, correlation value calculating means for obtaining complex cross-correlation functions C i of training sequences on the basis of said sample value sequences i and training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i according to the following equation and (where, i = 1, 2, ..., M; * means that complex conjugation is used.) outputting these functions to the maximum correlation index selecting means, maximum correlation sampling phase selecting means for obtaining a sampling phase τ (I k ) where the absolute values of said M number of complex cross-correlation functions C i are maximized by comparing said absolute values, and outputting said sampling phase τ (I k ) to a sampling phase smoothing means, sampling phase smoothing means for obtaining a first coordinate transformation term t c1 (k) and a second coordinate transformation term t c2 (j, k) according to the equations given below, a sectional weighted mean value τ AVG (k), a sectional maximum value τ MAX (k) and a sectional minimum value τ MIN (k) of said sampling phase τ (I k ) according to the following equations, (where, w a (j) (j = 1, 2, ..., L s ) is a weighted coefficient and L s denotes the maximum calculation length.) a selective sampling phase τ s according to the following equation, and τ s = τ AVG (k)-tc (k)+w s (τ MAX (k)-τ MIN (k) | tc (k) |) (where, w s indicates the shift coefficient, w a (j) (j = 1, 2, ..., L s ) is a weighted coefficient and L s denotes the maximum calculation length.) outputting said sampling phase τ s to a sampling processing part, a delay processing part for delaying said reception signals entered from said transmission lines as long as specified and outputting the signals to a sampling processing part, a sampling processing part for obtaining sampling value sequences = {y n } (n = 1, 2, ..., N) by sampling said reception signals from said delay processing part with respect to a symbol interval T and said sampling phase τ s and outputting said sample value sequences to a transmission estimating part and a Viterbi algorithm processing part, a transmission line estimating part for obtaining estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences and estimated values E from estimated values E = {Exn} (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences and outputting said estimated values E , and a Viterbi algorithm processing part for obtaining said estimated values E from said sample value sequences on the basis of said estimated values E according to the Viterbi algorithm and outputting said estimated values.

Claim 12

A maximum likelihood estimating method for obtaining and outputting sample value sequences i = {y in } (i = 1, 2, ..., M; n = 1, 2, ..., Nt) of the reception signals by sampling said reception signals entered from transmission lines at a symbol interval T starting from the estimated leading time determined by the frame synchronization mechanism with respect to individual sampling phases of sampling phase sequences τ = {τ (i)} (i = 1, 2, ..., M), complex cross-correlation functions = {C i }(i = 1, 2, ..., M) of training sequences by said sample value sequences i and said training sequences = {φ n } (n = 1, 2, ..., Nt) of said sample value sequences i , a sampling phase τ (I k ) where the absolute values of said M number of complex cross-correlation functions C i are maximized by comparing said absolute values, first and second coordinate transformation terms in accordance with said estimated leading time, a sectional weighted mean value, a sectional maximum value and a sectional minimum value of said sampling phase τ (I k ) after adding said second coordinate transformation term to said sampling phase τ (I k ), a sampling phase τ s obtained by adding a value, which is obtained by subtracting the absolute value of said first coordinate transformation value and said sectional minimum vale from said sectional maximum value and multiplying a resultant value by a constant, to said sectional weighted mean value from said resultant value, sample value sequences = {y n } (n = 1, 2, ..., N) by sampling said reception signals y(t) entered at specified timings with respect to a symbol interval T and said sampling phase τs, starting from said estimated leading time, estimated values E = {Eh j } (j = 0, 1, ..., L) of impulse responses of transmission lines from said sample value sequences and the training sequences corresponding to said sample value sequences according to an adaptive equalization algorithm in processing of a part including said training sequences in said sample value sequences , estimated values E from estimated values E = {Ex n } (n = 1, 2, ..., N) of said sample value sequences and transmission symbol sequences according to the adaptive equalization algorithm in processing of a part other than said training sequences in said sample value sequences , and said estimated values E from said sample value sequences according to the Viterbi algorithm in accordance with said estimated values E .