Method and apparatus for implementing maximum transition run codes

• US 5,859,601 A
• Filed: 10/15/1996
• Issued: 01/12/1999
• Est. Priority Date: 04/05/1996
• Status: Expired due to Term

• Current Assignee: Regents of The University of Minnesota (University of Minnesota)
• Original Assignee: Regents of The University of Minnesota (University of Minnesota)
• Inventors: Moon, Jaekyun, Brickner, Barrett J.
• Primary Examiner(s): Gaffin, Jeffrey A.
• Assistant Examiner(s): KOST, JASON L
• Application Number:
  US08/730,716
• Time in Patent Office: 819 Days
• Field of Search: 341/58, 341/59, 341/61, 341/94
• US Class Current: 341/59
• CPC Class Codes: G11B 20/1426 conversion to or from block codes or representations thereof
• CPC Class Codes: H03M 13/31 combining coding for error detection or correction and efficient use of the spectrum without error detection or correction H03M5/14 , H03M5/145
• CPC Class Codes: H03M 5/145 Conversion to or from block codes or representations thereof

Abstract

Apparatus and method for coding to improve the minimum distance properties of sequence detectors operating at high densities in storage systems is presented. The coding scheme of the present invention is referred to as maximum transition run (MTR) code and eliminates data patterns producing long runs of consecutive transitions while imposing the usual k constraint necessary for timing recovery. The code has a distance gaining property similar to an existing (1,k) runlength-limited (RLL) code, but can be implemented with considerably higher code rates. When the MTR code is used with fixed delay tree search (FDTS) or high order partial response maximum likelihood (PRML) detectors, the bit error rate performance improves significantly over existing combinations of codes and detectors.

Claims

1. Apparatus for encoding m-bit binary datawords into n-bit binary codewords, in a recorded waveform, where m and n are preselected positive integers such that n is greater than m, comprising:

receiver means for receiving the dataword;

encoder means coupled to the receiver means, for producing sequences of fixed length codewords;

means for imposing a pair of constraints (j;

k) on the encoded waveform wherein the j constraint is defined as the maximum number of consecutive transitions allowed on consecutive clock periods in the encoded waveform to facilitate the reduction of a probability of a detection error in said receiver means;

said sequences generating no more than j consecutive transitions in the recorded waveform such that j is an integer equal to or greater than 2; and

said sequences generating no more than k consecutive sample periods without a transition in the recorded waveform.

2. Apparatus as in claim 1 wherein the j consecutive transition limit is defined by the relationship 2≦

j<

10.

3. Apparatus as in claim 2 wherein the encoder means produces a codeword, in response to each dataword sequentially, based on a predetermined word-by-word mapping of 2^m m-bit datawords to one of N n-bit codeword sets, wherein N may be written as N=2ⁱ and i is a positive integer and further that a selection of one of said N n-bit codeword sets is determined by a state of the encoder wherein said state is a predetermined function of a previous state and the encoder input and each set contains 2^m codewords wherein, a particular codeword may appear more than once in a given set and further a particular codeword may also appear in more than one set.

4. Apparatus as in claim 3 wherein the encoder means produces a codeword in response to each dataword sequentially, based on a predetermined word-by-word mapping of 2^m m-bit datawords to one of two n-bit codeword sets, where each particular codeword set contains 2^m different codewords, some of which may also be used in the other set and the set mapped to the encoder is chosen based on the last binary symbol of the previous codeword.

5. Apparatus as in claim 4 wherein, a first set (A) is chosen when a last binary symbol of a previous codeword (Z) is a 0 and a second set (B) is chosen when Z is a 1 and valid codewords for sets A and B are by the steps of:

removing binary words that contain more than one of j consecutive 1'"'"'s and more than k consecutive 0'"'"'s from each of two lists of 2ⁿ possible codewords for sets A and B, respectively;

removing words that end with two consecutive 1'"'"'s from both lists;

removing words from the list for set B that begin with two consecutive 1'"'"'s;

selecting k₁ +k₂ =k;

removing words from the list for set A that begin with one of k₁ +1 0'"'"'s and end with k₂ +1 0'"'"'s;

removing words from the list for set B that end with consecutive 0'"'"'s and k₂ +1; and

selecting the 2^m codewords used in each of set A and set B from the respective lists, each of which contains at least 2^m codewords.

6. Apparatus as in claim 2 wherein the sequences of codewords also satisfy a DC-free constraint.

7. Apparatus as in claim 6 wherein the encoder means produces a codeword in response to each dataword sequentially, based on a predetermined word-by-word mapping of 2^m m-bit datawords to 2^m n-bit codewords, where the codewords are preselected using a selection method comprising the steps of:

removing binary words that contain either "0101" or "1010" from a list of 2ⁿ possible n-bit binary words;

removing words that contain more than k+1 consecutive like symbols;

removing all words that begin with "01" or "10" and those that end with "101" or "010" having an equivalent effect of removing all words that begin with "101" or "010" and all words that end with "01" or "10";

removing one and combinations thereof of words that begin with k₁ +1 consecutive like symbols and words that end with k₂ +1 consecutive like symbols where k₁ +k₂ =k;

forming a set (A) of codewords with the number of 1'"'"'s not less than the than number of 0'"'"'s;

forming a set (B) of codewords with the number of 0'"'"'s not less than the than number of 1'"'"'s;

selecting codewords from set A if the number of 0'"'"'s in all the previous encoder outputs exceeds the number of 1'"'"'s; and

selecting codewords from set B if the number of 0'"'"'s in all the previous encoder outputs does not exceed the number of 1'"'"'s.

8. Apparatus as in claim 2 wherein the consecutive transition limit is defined by the relationship j=2.

9. Apparatus as in claim 2 wherein the binary sequences produced by combining codewords have no more than j consecutive 1'"'"'s and no more than k consecutive 0'"'"'s when used with a NRZI recording format.

10. Apparatus as in claim 2 wherein binary sequences produced by combining codewords have no more than one of j consecutive transitions from 0 to 1 and from 1 to 0 and no more than one of k+1 consecutive 0'"'"'s and k+1 consecutive 1'"'"'s when used in conjunction with a NRZ recording format.

11. Apparatus as in claim 2 wherein the encoder means produces a codeword in response to each dataword sequentially, based on a predetermined word-by-word mapping of 2^m m-bit datawords to 2^m n -bit codewords, wherein the codewords are preselected using a selection method comprising the steps of:

removing binary words that contain more than one of j consecutive 1'"'"'s and more than k consecutive 0'"'"'s from a list of 2ⁿ possible n-bit binary words;

removing one of binary words that begin and end with two consecutive 1'"'"'s;

removing one of binary words that begin with k₁ +1 consecutive 0'"'"'s and end with k₂ +1 consecutive 0'"'"'s where k₁ +k₂ =k; and

choosing 2^m codewords remaining in the list, which contains at least 2^m valid codewords.

12. Apparatus as in claim 2 wherein the receiver means incorporates means for removing certain code-violating patterns from the detection process wherein the detection process comprises at least one of the steps of:

removing states and state transitions corresponding to more than j consecutive transitions from a Viterbi trellis;

removing branches from a fixed delay tree search corresponding to more than j consecutive transitions;

removing branches from a fixed delay tree search corresponding to more than j consecutive transitions when the previous decision is considered part of the sequence;

forming boundaries for a signal space formulation such that points in the signal space constellation corresponding to sequences containing more than j consecutive transitions are not considered; and

selecting boundaries in a signal space formulation based on a constellation that does not include points corresponding to sequences containing more than j consecutive transitions when the previous decision is considered part of the sequence.

13. A method for encoding m-bit binary datawords into n-bit binary codewords in a recorded waveform, where m and n are preselected positive integers such that n is greater than m, comprising the steps of:

receiving binary datawords; and

producing sequences of n-bit codewords;

imposing a pair of constraints (j;

k) on the encoded waveform;

generating no more than j consecutive transitions of said sequence in the recorded waveform such that j≧

2; and

generating no more than k consecutive sample periods of said sequences without a transition in the recorded waveform.

14. The method as in claim 13 wherein the consecutive transition limit is defined by the equation 2≦

j<

10.

15. The method as in claim 14 wherein the consecutive transition limit is j=2.

16. The method as in claim 14 wherein the binary sequences produced by combining codewords have no more than j consecutive 1'"'"'s and no more than k consecutive 0'"'"'s when used with the NRZI recording format.

17. The method as in claim 14 wherein the binary sequences produced by combining codewords have no more than one of j consecutive transitions from 0 to 1 and from 1 to 0 and no more than one of k+1 consecutive 0'"'"'s and k+1 consecutive 1'"'"'s when used in conjunction with the NRZ recording format.

18. The method as in claim 14 wherein the encoder means produces a codeword in response to each dataword sequentially, based on a predetermined word-by-word mapping of 2^m m-bit datawords to 2^m n-bit codewords, where the codewords are preselected using a selection method comprising the steps of:

removing binary words that contain more than j consecutive 1'"'"'s and words that contain more than k consecutive 0'"'"'s from a list of 2ⁿ possible n-bit binary words;

removing one of binary words that begin and end with two consecutive 1'"'"'s;

removing one of words that begin with k₁ +1 consecutive 0'"'"'s and end with k₂ +1 consecutive 0'"'"'s where k₁ +k₂ =k;

choosing 2^m codewords from the remaining list, which contains at least 2^m valid codewords.

19. The method as in claim 14 wherein the encoder means produces a codeword in response to each dataword sequentially, based on a predetermined word-by-word mapping of 2^m m-bit datawords to one of N n-bit codeword sets, wherein N may be written as N=2ⁱ where i is a positive integer and the selection of one of N codeword sets is determined by the state of the encoder and said state is a predetermined function of the previous state and encoder input and that each set contains 2^m codewords and further that a particular codeword may appear more than once in a given sets and may also appear in more than one set.

20. The method as in claim 14 wherein the sequences of codewords also satisfy a DC-free constraint.

21. The method as in claim 13 wherein the method of receiving data incorporates the removal of certain code-violating patterns from the detection process wherein the detection process comprises at least one of the steps of:

removing states and state transitions corresponding to more than j consecutive transitions from a Viterbi trellis;

removing branches from a fixed delay tree search corresponding to more than j consecutive transitions;

removing branches from a fixed delay tree search corresponding to more than j consecutive transitions when the previous decision is considered part of the sequence;

forming boundaries for a signal space formulation such that points in the signal space constellation corresponding to sequences containing more than j consecutive transitions are not considered; and

selecting boundaries in a signal space formulation based on a constellation that does not include points corresponding to sequences containing more than j consecutive transitions when the previous decision is considered part of the sequence.