Parallel maximum a posteriori detectors that generate soft decisions for a sampled data sequence

1. A soft decision value output detector comprising:

a plurality of maximum a posteriori (MAP) detectors configured to simultaneously generate state metrics for portions of a sampled data sequence, and to generate soft decision values based on the generated state metrics, wherein each of the MAP detectors comprises a first reverse Viterbi operator, a second reverse Viterbi operator, and a forward Viterbi operator, and wherein the plurality of MAP detectors comprise a number of MAP detectors that is at least one plus the ratio of a rate at which the sampled data sequence is input to the plurality of MAP detectors and a rate at which each MAP detector processes the sampled data sequence.

2. The soft decision value output detector of claim 1 , wherein the plurality of MAP detectors are configured to simultaneously generate the state metrics for partially overlapped multiple convergence length portions of the sampled data sequence.

3. The soft decision value output detector of claim 1 , wherein each of the plurality of MAP detectors is configured to generate soft decision values for a portion of the sampled data sequence that is adjacent to a portion of the sampled data sequence for which another one of the MAP detectors generates soft decision values.

4. The soft decision value output detector of claim 1 , wherein, within each of the plurality of MAP detectors, a first MAP unit is configured to generate soft decision values for a first portion of the sampled data sequence that is immediately adjacent to a second portion of the sampled data sequence for which a second MAP unit generates soft decision values.

5. The soft decision value output detector of claim 4 , wherein, within each of the plurality of MAP detectors, the first MAP unit is configured to generate the soft decision values about a defined number of P convergence lengths sample time before the second MAP unit generates the soft decision values.

6. The soft decision value output detector of claim 1 , wherein:

the plurality of MAP detectors are configured to generate soft decision values for sequential portions of the sampled data sequence.

7. The soft decision value output detector of claim 6 , wherein:

the plurality of MAP detectors are configured to generate soft decision values about one convergence length sample time after another one of the plurality of MAP detectors generates soft decision values.

8. The soft decision value output detector of claim 6 , wherein the soft decision values generated by the plurality of MAP detectors have less than a convergence length sample time therebetween.

9. The soft decision value output detector of claim 1 , wherein

the plurality of MAP detectors comprises four MAP detectors;

a first MAP detector is configured to generate soft decision values for a first portion of the sampled data sequence;

a second MAP detector is configured to generate soft decision values for a second portion of the sampled data sequence that is immediately adjacent to the first portion;

a third MAP detector is configured to generate soft decision values for a third portion of the sampled data sequence that is immediately adjacent to the second portion; and

a fourth MAP detector is configured to generate soft decision values for a fourth portion of the sampled data sequence that is immediately adjacent to the third portion.

10. The soft decision value output detector of claim 1 , wherein the plurality of MAP detectors are configured to generate soft decision values for portions of the sampled data sequence at least at a rate at which the portions of the sampled data sequence are input to the plurality of MAP detectors.

11. The soft decision value output detector of claim 1 , wherein the sampled data sequence is based on a read signal of data recorded on a disk in a disk drive.

12. A soft decision value output detector comprising:

a plurality of maximum a posteriori (MAP) detectors configured to simultaneously generate state metrics for portions of a sampled data sequence, and to generate soft decision values based on the generated state metrics,

wherein each of the MAP detectors comprises:

a first MAP unit that comprises a first forward Viterbi operator and a first reverse Viterbi operator that converge and output state metrics by respective forward iteration and reverse iteration through partially overlapped portions of the sampled data sequence; and

a second MAP unit that comprises a second forward Viterbi operator and a second reverse Viterbi operator that converge and output state metrics by respective forward iteration and reverse iteration through partially overlapped portions of the sampled data sequence, and the first and second MAP units generate the state metrics for partially overlapped portions of the sampled data sequence at a same time by alternating the first MAP unit and the second MAP unit between a learning phase starting from equiprobable states and a converged phase where reliable state metrics are produced.

13. The soft decision value output detector of claim 12 , wherein each of the MAP detectors further comprises a metric reversal stack that is configured to reverse order of the state metrics from the first and second reverse Viterbi operators.

14. A soft decision value output detector comprising:

a plurality of maximum a posteriori (MAP) detectors configured to simultaneously generate state metrics for portions of a sampled data sequence, and to generate soft decision values based on the generated state metrics,

wherein each of the MAP detectors comprises:

a first MAP unit that comprises a first forward Viterbi operator and a first reverse Viterbi operator that converge and output state metrics by respective forward iteration and reverse iteration through partially overlapped portions of the sampled data sequence; and

a second MAP unit that comprises a second forward Viterbi operator and a second reverse Viterbi operator that converge and output state metrics by respective forward iteration and reverse iteration through partially overlapped portions of the sampled data sequence, and wherein the plurality of MAP detectors comprise a number of MAP detectors that is at least one plus the ratio of a rate at which the sampled data sequence is input to the plurality of MAP detectors and a rate at which each MAP detector processes the sampled data sequence.

15. The soft decision value output detector of claim 14 , wherein each of the plurality of MAP detectors is configured to combine the state metrics that are output from the forward and reverse Viterbi operators in the first or second MAP unit to generate soft decision values.

16. A disk drive comprising:

a disk that is configured to store data;

a transducer that is configured to read data from the disk to generate a read signal;

an analog-to-digital converter that is configured to generate a sampled data sequence from the read signal; and

a plurality of maximum a posteriori (MAP) detectors that are configured to simultaneously generate state metrics for portions of the sampled data sequence, and to generate soft decision values based on the generated state metrics, wherein each of the MAP detectors comprises a first reverse Viterbi operator, a second reverse Viterbi operator, and a forward Viterbi operator, wherein the plurality of MAP detectors comprise a number of MAP detectors that is at least one plus the ratio of a rate at which the sampled data sequence is input to the plurality of MAP detectors and a rate at which each MAP detector processes the sampled data sequence.

17. The disk drive of claim 16 , wherein the plurality of MAP detectors are configured to simultaneously generate the state metrics for a plurality of adjacent convergence length portions of the sampled data sequence.

18. A method of generating soft decision values based on a sampled data sequence, the method comprising:

generating state metrics for partially overlapped portions of the sampled data sequence at a same time using a plurality of maximum a posteriori (MAP) detectors, wherein each of the MAP detectors operates a first reverse Viterbi operator and a second reverse Viterbi operator at a same time to reverse iterate through portions of the sampled data sequence to generate Viterbi based state metrics and operates a forward Viterbi operator to forward iterate through portions of the sampled data sequence to generate further Viterbi based state metrics; and

generating the soft decision values based on the generated Viterbi based state metrics, wherein each of the MAP detectors comprises a first MAP unit and a second MAP unit, and wherein generating the state metrics for partially overlapped portions of the sampled data sequence at the same time comprises:

alternating the first MAP unit and the second MAP unit between a learning phase starting from equiprobable states and a converged phase where reliable state metrics are produced.

19. The method of claim 18 , wherein generating the state metrics for partially overlapped portions of the sampled data sequence comprises generating the state metrics based on a plurality of partially overlapped convergence length portions of the sampled data sequence at the same time.

20. The method of claim 18 , wherein the first MAP unit comprises the first reverse Viterbi operator and the first forward Viterbi operator, the second MAP unit comprises the second reverse Viterbi operator and the second forward Viterbi operator, and wherein:

operating the first MAP unit comprises operating the first reverse and the first forward Viterbi operators at the same time to reverse iterate and forward iterate through portions of the sampled data sequence to generate Viterbi based state metrics through a learning mode and a converging mode; and

operating the second MAP unit comprises operating the second reverse and the second forward Viterbi operators at the same time to reverse iterate and forward iterate through portions of the sampled data sequence to generate Viterbi based state metrics through a learning mode and a converging mode, wherein the second MAP unit is operated at the same time as the first MAP unit.

21. The method of claim 18 , wherein generating the state metrics for partially overlapped portions of the sampled data sequence at the same time comprises generating the soft decision values for portions of the sampled data sequence at least at a rate at which the portions of the sampled data sequence are input to the MAP detectors.

22. A method of generating soft decision values based on a sampled data sequence, the method comprising:

operating a plurality of maximum a posteriori (MAP) detectors to simultaneously generate state metrics for portions of the sampled data sequence, and to generate the soft decision values based on the generated state metrics, wherein each of the MAP detectors comprises a first reverse Viterbi operator, a second reverse Viterbi operator, and a forward Viterbi operator, and wherein a number of the MAP detectors that are operated simultaneously to generate the state metrics for portions of the sampled data sequence is at least one plus the ratio of a rate at which the sampled data sequence is input to the plurality of MAP detectors and a rate at which each MAP detector processes the sampled data sequence.