Method of detecting DC-free sequences

• US 5,910,969 A
• Filed: 11/05/1996
• Issued: 06/08/1999
• Est. Priority Date: 11/05/1996
• Status: Expired due to Term

• Current Assignee: Avago Technologies General IP PTE Limited (Broadcom, Inc.)
• Original Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Soljanin, Emina, Sayiner, Necip
• Primary Examiner(s): Pham, Chi H.
• Assistant Examiner(s): WEBSTER, BRYAN E
• Application Number:
  US08/744,255
• Time in Patent Office: 945 Days
• Field of Search: 341/58, 375/341, 375/340, 375/262, 375/263, 375/265, 371/43.1, 371/43.2, 371/43.4
• US Class Current: 375/341
• CPC Class Codes: G11B 20/1426 conversion to or from block codes or representations thereof
• CPC Class Codes: H03M 5/145 Conversion to or from block codes or representations thereof
• CPC Class Codes: H04L 1/0054 Maximum-likelihood or sequential decoding, e.g. Viterbi, Fano, ZJ algorithms

Abstract

A method of decoding a sequence of samples received over a communications channel, such as a partial response channel, in which the sequence represents a length k binary bit stream encoded as a sequence of bipolar symbols comprised of q groups of L symbols, each length L group of said symbols corresponding to one of 2^m subwords having a predetermined block digital sum. The decoding method makes use of the structure of the encoding method by finding, in one illustrative embodiment, for each group of L samples, the maximum likelihood path through a time varying trellis supporting the sets of subwords having predetermined block digital sums to identify said subwords; recording the state metrics and branch metrics for each path and identifying, from the metrics the maximum likelihood path, the order of concatenation of the subwords.

Claims

1. A method of decoding a sequence of samples received over a communications channel representing a length k binary bit stream encoded as a sequence of bipolar symbols comprised of q groups of L symbols, each length L group of said symbols corresponding to one of 2^m subwords having a predetermined block digital sum, said subwords having been encoded according to an inner rate m/L code and concatenated into blocks of complementary sign block digital sum according to an outer rate p/q code, where p is one part of said length k sequence and qm is the remaining part, comprising the steps of:

a. finding, for each group of L samples, the maximum likelihood paths through a trellis of said inner rate p/q code for said subwords having complementary sign block digital sums of +q and of -q;

b. recording the cumulative state metrics and cumulative branch metrics for each of said paths; and

c. identifying from said cumulative metrics for each said group of L samples a corresponding branch in a trellis for said outer rate m/L code to determine the order of concatenation of said subwords.

2. A method of decoding a sequence of samples according to claim 1 wherein said communcations channel is a partial response channel.

3. A method of decoding a sequence of samples according to claim 2 wherein said communcations channel is a partial response channel whose polynomial transfer function has a (1-D) factor.

4. A method of decoding a sequence of samples according to claim 3 wherein said trellis for said inner rate code is a time-varying trellis.

5. A method of detecting a sequence of samples according to claim 1 wherein the number m is determined from floor log₂ (L C n), where n is selected to provide a number of dc-free combinations having a BDS=(L-2n).

6. A method of decoding a sequence of samples received over a communications channel, said sequence representing a length k binary bit stream encoded as a sequence of bipolar symbols comprised of q groups of L symbols, each length L group of said symbols corresponding to one of 2^m subwords having a predetermined block digital sum, said subwords having been encoded according to an inner rate m/L code and concatenated into blocks of complementary sign block digital sum according to an outer rate p/q code, where p is one part of said length k sequence and qm is the remaining part, comprising the steps of:

a. finding, for each group of L samples, the maximum likelihood path through a trellis of said inner rate p/q code for each of a set of subwords having complementary block digital sums;

b. recording the cumulative state and branch metrics for each said path until the maximum likelihood paths for q groups of L samples have been found;

c. recording the difference between the branch metrics for each said path; and

d. determining the order of concatenation of said subwords according to the q/2 smallest ones of said metrics.

7. A method of detecting a sequence of samples according to claim 6 wherein the number m is determined from floor log₂ (L C n), where n is selected to provide a number of dc-free combinations having a BDS=(L-2n).

8. A method of detecting a sequence of samples received over a partial response channel in the presence of noise, said sequence representing a length k binary bit stream encoded as a sequence of bipolar symbols comprised of q groups of L symbols, each length L group of said symbols corresponding to one of 2^m subwords having a predetermined block digital sum, said subwords having been encoded according to an inner rate m/L code and concatenated into blocks of complementary sign block digital sum according to an outer rate p/q code, where p is one part of said length k sequence and qm is the remaining part, comprising the steps of:

a. distributing each successive group of L samples of signals received over said channel to a respective panel of a trellis supporting said complementary block digital sum subwords of said inner rate code;

b. finding the maximum likelihood path through each said panel for each of said complementary sign block digital sum subwords in said group of L samples;

c. tabulating the cumulative branch and state metrics through said panels respective to each said path; and

d. identifying from of said cumulative metrics a branch through a trellis supporting said outer rate m/L code to determine the order of concatenation of said subwords.

9. A method of detecting a sequence of samples according to claim 8 wherein the number m is determined from floor log₂ (L C n), where n is selected to provide a number of dc-free combinations having a BDS=(L-2n).

10. A method of detecting a sequence of samples received over a communications channel in the presence of noise, said sequence representing a length k binary bit stream encoded as a sequence of bipolar symbols comprised of q groups of L symbols, each length L group of said symbols corresponding to one of 2^m subwords having a predetermined block digital sum, said subwords having been encoded according to an inner rate m/L code and concatenated into blocks of complementary sign block digital sum according to an outer rate p/q code, where p is one part of said length k sequence and qm is the remaining part, comprising the steps of:

a. distributing each successive group of L samples of signals received over said channel to a respective panel of a time-varying trellis supporting said complementary block digital sum subwords;

b. finding the maximum likelihood paths for said complementary block digital sum subwords through each said panel for each said group of L samples;

c. recording the state metrics and branch metrics for each said of said paths;

d. recording the difference between the metrics for said paths; and

e. determining the order of concatenation of said subwords according to the q/2 smallest ones of said metrics.

11. A method of detecting a sequence of samples according to claim 10 wherein the number m is determined from floor log₂ (L C n), where n is selected to provide a number of dc-free combinations having a BDS=(L-2n).