EP0126301A2 — Adaptive equalizer for digital signals

Claim 1

Independent Claim 227 words

Method for the adaptive equalization of digital signals, according to which the signals coming from a transmission channel (1) to be equalized are sampled by an analog-to-digital converter (AD), the samples are filtered in a transversal filter (EQ) and the samples 5 outgoing from that transversal filter being sent to a decision device (CD), characterized in that a block of said samples is extracted at the output of the analog-to-digital converter (AD) and their convolution by an initial set of coefficients (c) is effected, the result being used to determine a first mean square error, hence for each 10 coefficient (c i ) a suitable increment and decrement are effected by a first (ß) and a second (α) quantity, respectively, and the new mean square errors of the so-modified coefficient sets are determined, obtaining in this way the optimal values ( c i ) of each coefficient (c i ), then the result is approximated to the nearest power of two 15 and is immediately substituted in the set of coefficients for determining the following coefficient, thus obtaining a new set consisting only of powers of two, then the mean square error of said new set of coefficients is determined and, if it results to be inferior than the above-mentioned first mean square error, the new set of coeffi-20 cients is transferred to the transversal filter (EQ).

Claim 2

Dependent Claim 227 words

Method for the adaptive equalization of digital signals, according to which the signals coming from a transmission channel (1) to be equalized are sampled by an analog-to-digital converter (AD), the samples are filtered in a transversal filter (EQ) and the samples 5 outgoing from that transversal filter being sent to a decision device (CD), characterized in that a block of said samples is extracted at the output of the analog-to-digital converter (AD) and their convolution by an initial set of coefficients (c) is effected, the result being used to determine a first mean square error, hence for each 10 coefficient (c i ) a suitable increment and decrement are effected by a first (ß) and a second (α) quantity, respectively, and the new mean square errors of the so-modified coefficient sets are determined, obtaining in this way the optimal values ( c i ) of each coefficient (c i ), then the result is approximated to the nearest power of two 15 and is immediately substituted in the set of coefficients for determining the following coefficient, thus obtaining a new set consisting only of powers of two, then the mean square error of said new set of coefficients is determined and, if it results to be inferior than the above-mentioned first mean square error, the new set of coeffi-20 cients is transferred to the transversal filter (EQ).

Claim 3

Dependent Claim 69 words

Method as in claim 1, characterized in that the optimal values ( c i ) of each coefficient (c.) are computed one by one according to the expression by means of three points (Pl, P2, P3) with ordinates equal to mean square errors (ε 1 ,ε 2 ,ε 3 ) and abscissae equal to the respective values of the coefficients (c i1 , c i 2' c i3 ).

Claim 4

Dependent Claim 187 words

Adaptive equalizer apt to implement the method of claim 1 or 2, characterized in that said block of samples is stored in a first register (RM), belonging to a simulator (SE) of a transversal filter to which a coefficient set is also transferred and is stored into a second register (RC), the contents of said first and second register being sent to a device (DC) apt to effect their convolution, which is subsequently stored in a third register (RS), whose output is connected to a device (ME) for the measurement of the mean square errors, which sends the digital signals resulting from the measurement to a computing unit (UC), apt to effect the determination of the optimal values for each of the coefficients (c i ) with predetermined values of said first (ß) and second (α) quantities and also apt to send said coefficient sets to the first register (RC) and to said transversal filter (EQ) and to control the development of the operations carried out by said simulator (SE), by said device (ME) for the measurement of mean square errors and by said transversal filter (EQ).

Claim 5

Dependent Claim 98 words

Adaptive equalizer as in claim 3, characterized in that said transversal filter (EQ) consists of a first plurality of shift registers (SRI, ..., SR(N-1)) apt to set up a delay line for the binary words representing the samples, of a second plurality of shift registers (CSRI, ..., CSRN) apt to effect the digital multiplications of the binary words by the power-of-two coefficients (c), effecting a number of shifts of each word equal to the exponent of the coefficients above, and of a binary-addition device (DS) whose inputs are connected to the outputs of said second plurality of registers.

Adaptive equalizer for digital signals

Abstract

Inventors

Bibliographic Details

Claims Analysis

References Cited (Backward Citations)

USPTO Annuity & Maintenance Windows

Assigned Categories & Tags

Assign Category Tag