US 5,710,793 AGranted

Error signal quantization method and hardware for mixed blind and decision directed equalization

AssigneeNational Semiconductor Corporation

InventorCraig Bradley Greenberg

Life of this patent

Estimating the term…

Loading patent…

Abstract

According to a first aspect, a novel method of performing an eight point quantization is developed. The eight point quantization consists of selecting one eight specific points depending upon the phase of the measured complex error signal. According to a second aspect, a novel method of performing eight-point quantization of an error signal by rotating a measured complex sample point by positive pi/8 and by negative pi/8. Two sign comparisons results in two control signals. These control signals used in conjunction with the signs of the real and imaginary parts of the variable being quantized results in all the information necessary to select one of eight equal and radially symmetric decision regions. According to a third aspect, an efficient hardware circuit implements the complex coefficient update algorithm using the quantization method according to the first aspect on the error signal. The coefficient update circuit on a per tap basis requires four adders, eight multiplexors, two negation units, and two barrel shifters (if a programmable adaptation constant is supported). The equalizer filter requires a four multiplier control circuit for supply the multiplexor select signals. According to a fourth aspect, a raw error threshold is developed which indicates that for sample points received within a threshold distance T from a constellation point, the error signal for the equalizer coefficient update circuit is the difference vector in the measured error signal and the constellation point.

Claims41

Dependent claims

A method of quantizing a measured complex error signal to supply a quantized complex error signal, the measured complex error signal including a measured real error component and a measured imaginary error component, the measured complex error signal being defined by a measured error magnitude and a measured error phase angle in a two-dimensional complex plane, the quantized complex error signal including a quantized real error component and a quantized imaginary error component, comprising the steps of:

assigning a third predetermined correction value to the quantized real error component and assigning a first predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a first area of the complex plane;
assigning a second predetermined correction value to the quantized real error component and assigning the second predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a second area of the complex plane;
assigning the first predetermined correction value to the quantized real error component and assigning the third predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a third area of the complex plane;
assigning a negative of the second predetermined correction value to the quantized real error component and assigning the second predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a fourth area of the complex plane;
assigning a negative of the third predetermined correction value to the quantized real error component and assigning a first predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a fifth area of the complex plane;
assigning the negative of the second predetermined correction value to the quantized real error component and assigning the negative of the second predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a sixth area of the complex plane;
assigning the first predetermined correction value to the quantized real error component and assigning the negative of the third predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a seventh area of the complex plane;
and
assigning the second predetermined correction value to the quantized real error component and assigning the negative of the second predetermined correction value to the quantized imaginary error component when the measured complex error signal is in an eighth area of the complex plane.

6 more elements

A method of performing an eight-point quantization of a measured complex error signal to supply a quantized complex error signal being defined by one of eight possible complex correction constellations, the measured complex error signal including a measured real error component and a measured imaginary error component, the measured complex error signal being defined by a measured error magnitude and a measured error phase angle in a two-dimensional complex plane, the quantized complex error signal including a quantized real error component and a quantized imaginary error component, comprising the steps of:

multiplying the measured complex error signal by e.sup.j.pi./8 to provide a first complex decision value having a first decision value real component and a first decision value imaginary component;
multiplying the measured complex error signal by e.sup.-j.pi./8 to provide a second complex decision value having a second decision value real component and a second decision value imaginary component;
performing a sign operation on the first and second complex decision values to produce a first real sign, a first imaginary sign, a second real sign, and a second imaginary sign;
assigning a first constellation to the quantized complex error signal, when the first real sign is positive, the second real sign is positive, and the first imaginary sign is opposite of the second imaginary sign;
assigning a second constellation to the quantized complex error signal, when the first real sign is positive, the second real sign is positive, the first imaginary sign is positive, and the second imaginary sign is positive;
assigning a third constellation to the quantized complex error signal, when the first real sign and second real sign are opposite, the first imaginary sign is positive, and the second imaginary sign is positive;
assigning a fourth constellation to the quantized complex error signal, when the first real sign is negative, the second real sign is negative, and the first imaginary sign is positive, and the second imaginary sign is positive;
assigning a fifth constellation to the quantized complex error signal, when the first real sign is negative, the second real sign is negative, and the first imaginary sign is opposite of the second imaginary sign;
assigning a sixth constellation to the quantized complex error signal, when the first real sign is negative, the second real sign is negative, the first imaginary sign is negative, and the second imaginary sign is negative;
assigning a seventh constellation to the quantized complex error signal, when the first real sign is opposite from the second real sign, the first imaginary sign is negative, and the second imaginary sign is negative;
and
assigning an eighth constellation to the quantized complex error signal, when the first real sign is positive, the second real sign is positive, the first imaginary sign is negative, and the second imaginary sign is negative.

9 more elements

A method of deriving a quantized complex error signal from a complex output signal and a complex decision constellation, the quantized complex error signal including a quantized real error component and a quantized imaginary error component, comprising the steps of:

computing a raw error scaler monotonically related to a magnitude difference between the complex output signal and the complex decision constellation;
comparing the raw error scaler to a raw error threshold;
subtracting the complex decision constellation from the complex output signal to provide a measured complex error signal when the raw error scaler is less than the raw error threshold, assigning the complex output signal to the measured complex error signal when the raw error scaler is greater than the raw error threshold, and wherein the measured complex error signal consists of a measured real error component and a measured imaginary error component, the measured complex error signal being defined by a measured error magnitude and a measured error phase angle in a two-dimensional complex plane;
assigning a third predetermined correction value to the quantized real error component and assigning a first predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a first area of the complex plane;
assigning a second predetermined correction value to the quantized real error component and assigning the second predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a second area of the complex plane;
assigning the first predetermined correction value to the quantized real error component and assigning the third predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a third area of the complex plane;
assigning a negative of the second predetermined correction value to the quantized real error component and assigning the second predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a fourth area of the complex plane;
assigning a negative of the third predetermined correction value to the quantized real error component and assigning a first predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a fifth area of the complex plane;
assigning the negative of the second predetermined correction value to the quantized real error component and assigning the negative of the second predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a sixth area of the complex plane;
assigning the first predetermined correction value to the quantized real error component and assigning the negative of the third predetermined correction value to the quantized imaginary error component when the measured complex error signal is in a seventh area of the complex plane;
assigning the second predetermined correction value to the quantized real error component and assigning the negative of the second predetermined correction value to the quantized imaginary error component when the measured complex error signal is in an eighth area of the complex plane;
and
if the raw error scaler is greater than the raw error threshold,
computing an output magnitude scaler monotonically related to a magnitude of the complex output signal,
comparing the output magnitude scaler to an output magnitude threshold, and
reversing a sign of the quantized complex error signal when the output magnitude scaler is greater than the output magnitude threshold.

13 more elements

A complex coefficient update circuit having a complex sample as input including a real sample component and an imaginary sample component, having a complex error signal as input including a real error component and an imaginary error component, the complex error signal having an associated error signal phase angle and an associated error signal magnitude, the circuit comprising:

a selection signal generator having as inputs the real error component and the imaginary error component and having as outputs an imaginary error sign selection signal, a real error sign selection signal, a rotated real error sign equality selection signal, and a rotated imaginary error sign equality selection signal;
and
an arithmetic logical module having as inputs the real sample component, the imaginary sample component, the imaginary error sign selection signal, the real error sign selection signal, the rotated real error sign equality selection signal, and the rotated imaginary error sign equality selection signal, having as outputs a real coefficient correction value and an imaginary coefficient correction value;
wherein the selection signal generator asserts the real error sign selection signal if the real error component is negative and deasserts the real error sign selection signal if the real error component is positive;
wherein the arithmetic logical module outputs the real coefficient correction value equal to one and one-half times the real sample component, one and one-half times the imaginary sample component, negative one and one-half times the real sample component, negative one and one-half times the imaginary sample component, the real sample component plus the imaginary sample component, negative sum of the real sample component plus the imaginary sample component, the real sample component minus the imaginary sample component, or the imaginary sample component minus the real sample component, as selected by the imaginary error sign selection signal, the real error sign selection signal, the rotated imaginary error sign equality selection signal, and the rotated imaginary error sign equality selection signal;
and
wherein the arithmetic logical module outputs the real coefficient correction value equal to one and one-half times the real sample component, one and one-half times the imaginary sample component, negative one and one-half times the real sample component, negative one and one-half times the imaginary sample component, the real sample component plus the imaginary sample component, negative sum of the real sample component plus the imaginary sample component, the real sample component minus the imaginary sample component, or the imaginary sample component minus the real sample component, as selected by the imaginary error sign selection signal, the real error sign selection signal, the rotated real error sign equality selection signal, and the rotated imaginary error sign equality selection signal.

4 more elements

Dependencies are inferred from the claim text.