Method and apparatus for compensation of linear or nonlinear intersymbol interference and noise correlation in magnetic recording channels

• US 5,889,823 A
• Filed: 12/13/1995
• Issued: 03/30/1999
• Est. Priority Date: 12/13/1995
• Status: Expired due to Term

• Current Assignee: Avago Technologies General IP PTE Limited (Broadcom, Inc.)
• Original Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Agazzi, Oscar E., Seshadri, Nambirajan
• Primary Examiner(s): Pham, Chi H.
• Assistant Examiner(s): Tran, Khai
• Application Number:
  US08/572,276
• Time in Patent Office: 1,203 Days
• Field of Search: 375/341, 375/340, 375/290, 375/261, 375/264, 375/298, 375/354, 371/43, 371/43.2, 371/43.4, 371/43.7, 371/43.8, 371/46
• US Class Current: 375/341
• CPC Class Codes: G11B 20/10009 Improvement or modification of read or write signals

Abstract

A method and apparatus for decoding a signal in the presence of linear or nonlinear interference, correlated noise or a combination thereof. The method involves a concatenated decoder comprising an inner decoder and a postprocessor. N most likely paths are generated by the inner decoder, or a combination of the inner decoder and an error event generator. These paths are then presented to the postprocessor which computes metrics for each path based on information about the signal, not considered by the inner decoder. These metrics can be based on linear or nonlinear model of the channel and optionally, a noise whitened signal for noise decorrelation. The path with the best metric is selected as the concatenated decoder output. The postprocessor can also be implemented in the presence of channel codes. It is also shown that the postprocessor is especially suited for a trellis implementation. When applied to magnetic recording channels, the method of the present invention allows for increased recording capacity. In addition, the present invention requires less computation and results in a reduction of the power dissipation of about two orders of magnitude compared to alternative approaches.

Claims

1. A method for decoding a received digital signal transmitted over a channel, said method comprising:

generating N most likely paths representing said received digital signal, each of said N most likely paths having a first type metric associated therewith, said first type metric based on a simplified model of said channel; and

postprocessing said N most likely paths for a final most likely path representative of said received digital signal, said most likely path indicated by a best second type metric, said best second type metric based on a comprehensive model of said channel which accounts for at least one source of channel information from a group consisting of intersymbol interference and nonlinearity.

2. A method according to claim 1 wherein said channel is a magnetic recording channel.

3. A method according to claim 1 wherein said generating step further comprises the steps of:

generating a single tentative most likely path representative of said received digital signal;

computing a finite number of most likely error events based on said simplified model of said channel; and

generating said N most likely paths by combining said tentative most likely path with said finite number of most likely error events using a trellis style implementation.

4. A method according to claim 3 wherein said most likely error events are based on said first type metric.

5. A method according to claim 1 wherein said simplified model of said channel does not consider a nonlinear component of said channel.

6. A method according to claim 1 wherein said simplified model of said channel does not consider a noise correlation component of said channel.

7. A method according to claim 5 wherein said nonlinearity of said comprehensive model is based on a Volterra series expansion.

8. A method according to claim 5 wherein said nonlinearity of said comprehensive model is based on a lookup table.

9. A method according to claim 1 wherein said generating step is implemented with an N-best Viterbi decoder.

10. A method according to claim 1 wherein said N most likely paths are constrained to paths having a first type metric value less than a predetermined threshold.

11. A method according to claim 1 wherein said comprehensive model of said channel further accounts for the effect of correlation of noise in said channel.

12. A method according to claim 1 wherein said received digital signal was transmitted with a channel code and said generating N most likely paths ignores said channel code.

13. A method according to claim 3 wherein said received digital signal was transmitted with a channel code, said generating a single tentative most likely path step ignores said channel code and said most likely error events are based on said channel code.

14. A method according to claim 1 wherein said received digital signal is divisible into one or more blocks of data, each of said blocks being constrained under a channel code, further comprising the steps of:

determining whether any of said blocks of said N most likely path, violates said channel code;

eliminating said paths of said N most likely paths having a channel code violation, said paths of said N most likely paths not eliminated, comprising a set of N'"'"' most likely paths; and

activating said postprocessing step if said set of N'"'"' most likely paths comprises more than one most likely path.

15. A method according to claim 11 wherein said postprocessing step includes the use of a noise-whitening filter.

16. A concatenated decoder for decoding a received digital signal transmitted over a channel comprising:

an inner N-best Viterbi decoder; and

a postprocessor comprising computer implemented means for computing a metric based on a comprehensive model of said channel, said comprehensive model accounting for at least one source of channel information from a group consisting of intersymbol interference and nonlinearity; and

computer implemented means for selecting a most likely path representative of said received digital signal, based on said metric, wherein said postprocessor further comprises means for adaptive channel estimation of characteristics of said channel.

17. A concatenated decoder according to claim 16 wherein said means for estimating channel characteristics of said channel includes means for estimating nonlinear components of said channel.

18. A concatenated decoder for decoding a received digital signal transmitted over a channel, comprising:

an inner decoder for generating N most likely paths representing said received digital signal, each of said N most likely paths having a first type metric associated therewith, said first type metric based on a simplified model of said channel; and

a postprocessor for determining a final most likely path from said N most likely paths representative of said received digital signal, said most likely path indicated by a best second type metric, said second type metric based on a comprehensive model of said channel which accounts for at least one source of channel information from the group consisting of intersymbol interference and nonlinearity.

19. A concatenated decoder according to claim 18 wherein said inner decoder further comprises:

a decoder having a single tentative most likely path output;

means for computing a finite number of most likely error events based on a simplified model of said channel; and

means for generating said N most likely paths by combining said single most likely path decoder output with said finite number of most likely error events using a trellis style implementation.

20. A concatenated decoder according to claim 18, wherein said received digital signal is divisible into one or more blocks of data, and each of said blocks being constrained under a channel code, further comprising:

means for determining whether any of said N most likely paths, violates said channel code;

means for eliminating said paths of said N most likely paths having a channel code violation, said paths of said N most likely paths not eliminated, comprising a set of N'"'"' most likely paths; and

means for activating said postprocessing step if said set of N'"'"' most likely paths comprises more than one most likely path.

21. A method for decoding a received digital signal transmitted over a channel, said method comprising:

generating a single tentative most likely path representative of said received digital signal based on a simplified model of said channel;

computing a finite number of most likely error events based on said simplified model of said channel using a trellis implementation;

generating N most likely paths by combining said single tentative most likely path with said finite number of most likely error events using said trellis style implementation;

computing a second type metric based on a comprehensive model of said channel; and

generating a single most likely path based on said second type metric using said trellis style implementation.

22. A method according to claim 21 wherein said trellis style implementation involves one or more states, each of said states representing a different length of error event.

23. A method according to claim 22 wherein said error event comprises one or more error event types and each of said error event types are represented by a different state.

24. A method according to claim 22 wherein said channel has a PR IV response and said error events are first order error events.

25. A method according to claim 23 wherein said channel has an EPR IV response and said error events are first order error events.

26. A method according to claim 21 wherein said comprehensive model accounts for at least one source of channel information from the group consisting of intersymbol interference, nonlinearity and noise correlation.