Method and controller for equalizing a received serial data stream
View Patent ↗A controller and methods for adaptively adjusting a dispersion compensation circuit in a receiver of a communication link are disclosed. The controller receives a signal from an error detector coupled to the receiver. The controller includes a logic engine that provides configuration information to the dispersion compensation circuit. The configuration information defines a digital filter in response to at least one tap constraint and at least one coefficient constraint. In an example embodiment, the logic engine generates a measure of residual inter-symbol interference and a measure of a noise enhancement penalty and iteratively provides a set of adjusted coefficient values that when applied in the digital filter will result in an equalized receiver output signal that minimizes a mathematical combination of the measure of the residual inter-symbol interference and the measure of the noise enhancement penalty. Alternatively, for bandwidth limited channels, the controller provides a predetermined set of coefficient values.
1. A controller for adaptively adjusting a dispersion compensation circuit in a receiver of a communication link, comprising:
a first interface for receiving a signal from an error detector responsive to an output of the dispersion compensation circuit; and
a logic engine coupled to the first interface and arranged to provide configuration information to the dispersion compensation circuit, the configuration information defining a digital filter in response to a tap constraint and a coefficient constraint, the logic engine further arranged to generate a measure of residual inter-symbol interference and a measure of noise enhancement penalty and iteratively generating a set of adjusted coefficient values that when applied in the digital filter will result in an equalized receiver output signal that minimizes a mathematical combination of the measure of the residual inter-symbol interference and the measure of the noise enhancement penalty, wherein the noise enhancement penalty is a function of an original noise prior to the dispersion compensation circuit and a square root of a sum of a set of ratios of the tap weights of non-reference taps divided by the tap weight of a reference tap where the set of ratios is squared.
2. The controller of claim 1 , wherein the tap constraint includes a minimum number of taps in the defined digital filter.
3. The controller of claim 1 , wherein the tap constraint includes a fractional unit interval space between adjacent taps in the defined digital filter.
4. The controller of claim 3 , wherein the tap constraint includes a range of fractional unit interval spaces between adjacent taps in the defined digital filter.
5. The controller of claim 4 , wherein the range of fractional unit interval spaces between adjacent taps is fixed with a value in the range of 0.3 to 0.7.
6. The controller of claim 1 , wherein the tap constraint includes at least two precursor taps in the defined digital filter.
7. The controller of claim 1 , wherein the tap constraint includes a location of a reference tap in the digital filter.
8. The controller of claim 7 , wherein the location of the reference tap in the digital filter is asymmetric with respect to remaining taps.
9. The controller of claim 1 , wherein the tap constraint is responsive to a comparison of a pulse spread of a received signal with a function of the number of taps and a fractional unit interval space between adjacent taps.
10. The controller of claim 9 , wherein the comparison confirms that the result of the function is at least as wide as the pulse spread of the received signal.
11. The controller of claim 1 , wherein the tap constraint includes a tap spread of at least 3 unit intervals.
12. The controller of claim 1 , wherein the coefficient constraint includes ensuring that a sign associated with a coefficient value for adjacent taps to a reference tap is positive.
13. The controller of claim 1 , wherein the coefficient constraint includes a range for a sum of coefficient values.
14. The controller of claim 13 , wherein the range for the sum of coefficient values is from 0.8 to 1.2.
15. A method for adaptively equalizing a received serial data stream processed by a dispersion compensation circuit, the method comprising:
receiving a signal from a detector responsive to an output of the dispersion compensation circuit;
using a logic engine responsive to the signal from the error detector, a tap constraint, and a coefficient constraint, to generate and provide configuration information to the dispersion compensation circuit;
using the logic engine to generate a measure of residue inter-symbol interference in a receive signal adjusted with the dispersion compensation circuit;
using the logic engine to generate a measure of noise enhancement penalty in the receive signal adjusted with the dispersion compensation circuit, wherein the noise enhancement penalty is a function of an original noise in a received serial data stream before adaptive equalization by the dispersion compensation circuit and a square root of a sum of a set of ratios of the tap weights of non-reference taps divided by the tap weight of a reference tap where the set of ratios is squared; and
iteratively generating a set of adjusted coefficient values that when applied in the dispersion compensation circuit will result in an equalized receiver output signal that minimizes a mathematical combination of the measure of the residue inter-symbol interference and the measure of the noise enhancement penalty.
16. The method of claim 15 , wherein the tap constraint includes a fractional unit interval space between adjacent taps in a digital filter.
17. The method of claim 15 , further comprising:
forwarding a tap delay adjustment to controllably manipulate a fractional unit interval space between adjacent taps in a digital filter.
18. The method of claim 17 , wherein the tap delay adjustment is applied to taps adjacent to a reference tap.
19. A method for equalizing a received serial data stream processed by a dispersion compensation circuit, the method comprising:
determining that a received signal is degraded due to a bandwidth limited communication channel; and
applying a set of coefficient values to the dispersion compensation circuit, the coefficient values generated as a function of a coefficient value for the adjacent taps to a reference tap, wherein a coefficient value associated with the reference tap is one, wherein a sign associated with both adjacent taps is positive.
20. The method of claim 19 , wherein the coefficient value for the adjacent taps to the reference tap is in a range of 0.3 to 0.5 and the coefficient value for taps further removed from the reference tap are determined from the product of −½ and the coefficient value for the adjacent taps.