Modified viterbi detector which accounts for correlated noise

• US 6,282,251 B1
• Filed: 03/21/1995
• Issued: 08/28/2001
• Est. Priority Date: 03/21/1995
• Status: Expired due to Fees

• Current Assignee: Seagate Technology LLC (Seagate Technology Holdings Plc), United Microelectronics Corporation
• Original Assignee: Seagate Technology LLC (Seagate Technology Holdings Plc)
• Inventors: Worstell, Glen Douglas
• Primary Examiner(s): Chin, Stephen
• Application Number:
  US08/407,230
• Time in Patent Office: 2,352 Days
• Field of Search: 375/341, 371/43, 360/29, 714/794
• US Class Current: 375/341
• CPC Class Codes: H03M 13/3961 Arrangements of methods for branch or transition metric calculation
• CPC Class Codes: H03M 13/41 using the Viterbi algorithm or Viterbi processors
• CPC Class Codes: H03M 13/6343 Error control coding in combination with techniques for partial response channels, e.g. recording
• CPC Class Codes: H04L 25/03184 Details concerning the metric
• CPC Class Codes: H04L 25/03203 Trellis search techniques

Abstract

A system detects an input signal received over a plurality of time periods. The input signal corresponds to one of a plurality of states during each time period, the states being connected by branches. The input signal has a value that is changeable from one of the plurality of time periods to the next. A first merged state is detected in which the branches during a first merge time period lead to only one of the plurality of states. A metric is determined for each of the plurality of states for each time period. The metric is based on the value of the input signal during a present time period, the value of the input signal during a previous time period, and on the plurality of branches connected to the states for which the metric is being determined. The likely branch leading to each of the plurality of states is identified based on the metric determined for that state. A second merged state is detected in which the branches during a second merge time period lead to only one of the plurality of states. The value of the input signal is determined for each time period between the first and second merged states based on the likely branches leading between the first and second merged states.

Claims

1. A method of detecting an input signal received over a plurality of time periods and corresponding to one of a plurality of states during each time period, the states being connected by branches, and the input signal having a value that is changeable from one of the plurality of time periods to a next of the plurality of time periods, the method comprising:

detecting a first merged state in which the branches during a first merge time period lead to only one of the plurality of states;

determining a metric for each of the plurality of states for each time period, the metric being based on an estimated value of noise in the input signal during a present time period and an estimated value of noise in the input signal during a previous time period and on the plurality of branches connected to the state for which the metric is being determined;

identifying a likely branch leading to each of the plurality of states based on the metric determined;

detecting a second merged state in which the branches during a second merge time period lead to only one of the plurality of states; and

determining the value of the input signal for each time period between the first and second merged states based on the likely branches leading between the first and second merged states.

2. The method of claim 1 wherein determining a metric for each of the plurality of states comprises:

determining a branch metric for each branch connected to each of the plurality of states; and

adding the branch metric to a previous state metric corresponding to a previous state during a previous time period to which the branch is connected to obtain a plurality of intermediate metrics for each of the plurality of states.

3. The method of claim 2 wherein identifying a likely branch comprises:

determining, for each of the plurality of states, which of the corresponding intermediate metrics has a smallest value; and

storing the intermediate metric having the smallest value as the metric for the corresponding state, the metric being indicative of the likely branch.

4. The method of claim 1 wherein determining a metric further comprises:

determining a current observed noise value in the input signal for the present time period; and

subtracting from the current observed noise value a prior noise value including at least one prior observed noise value corresponding to noise observed in the input signal during a prior time period, to obtain a metric noise value, wherein the prior observed noise value is weighted by a weighting factor.

5. The method of claim 4 wherein determining a metric further comprises:

determining an absolute value of the metric noise value.

6. The method of claim 4 wherein the prior noise value comprises a sum of a plurality of prior observed noise values weighted by weighting factors.

7. The method of claim 4 wherein the weighting factor comprises a unity weighting factor.

8. The method of claim 1 wherein the noise includes substantially uncorrelated noise and correlated noise.

9. The method of claim 8 wherein the correlated noise comprises:

error introduced by filtering.

10. The method of claim 1 wherein determining a metric further comprises:

squaring a current observed noise value for a first time period to obtain a present squared noise value; and

subtracting from the present squared noise value a prior noise value including at least one prior observed noise value corresponding to noise in the input signal observed during a prior time period, wherein the prior observed noise value is weighted by a weighting factor.

11. The method of claim 10 wherein the prior noise value includes a sum of a plurality of prior observed noise values multiplied by weighting factors.

12. The method of claim 10 wherein the weighting factor comprises a unity weighting factor.

13. The method of claim 11 wherein a preselected number of prior observed noise values are multiplied by the current observed noise value.

14. A Viterbi detector detecting an input signal in a disc drive, the input signal having a value corresponding to one of a plurality of states during each of a plurality of time periods, the input signal reaching the one of the states from one of a plurality of previous states during a previous time period, the Viterbi detector comprising:

a metric calculator for determining a metric for each of the plurality of states associated with each time period, the metric calculator determining the metric based on an estimated value of noise in the input signal during a previous time period and an estimated value of noise in the input signal during a current time period;

a comparator, coupled to the metric calculator, comparing the metrics determined for each state; and

a branch identifier, coupled to the comparator, identifying a previous state which likely preceded each of the plurality of states based on the comparison of the metrics.

15. The Viterbi detector of claim 14 wherein the previous states are connected to the plurality of states by branches and wherein the branch identifier identifies the previous state by identifying which of the metrics calculated for a given state is smallest.

16. The Viterbi detector of claim 15 wherein the metric calculator includes means for determining a current observed noise value in the input signal and subtracting from the current observed noise value a previous observed noise value corresponding to noise observed in the input signal during a previous time period, the previous observed noise value being weighted by a weighting factor.

17. A Viterbi detector detecting an input signal in a disc drive, the input signal having a value corresponding to one of a plurality of states during each of a plurality of time periods, the input signal reaching the one of the states from one of a plurality of previous states during a previous time period, the Viterbi detector comprising:

a metric calculator for determining a metric for each of the plurality of states associated with each time period, the metric calculator determining the metric based on an estimated value of noise in the input signal during a previous time period to account for correlated noise in the disc drive.

18. The Viterbi detector of claim 17 and further comprising:

a comparator, coupled to the metric calculator, comparing the metrics determined for each state; and

a branch identifier, coupled to the comparator, identifying a plurality of branches, one of which likely led to each of the plurality of states, based on the comparison of the metrics.

19. The Viterbi detector of claim 18 wherein the metric calculator calculates a potential metric associated with each branch and wherein the branch identifier identifies the branch having a lowest associated potential metric, based on an output from the comparator.