Method of equalization in digital mobile communication with improved tracking ability

• US 5,274,670 A
• Filed: 02/28/1992
• Issued: 12/28/1993
• Est. Priority Date: 03/01/1991
• Status: Expired due to Term

• Current Assignee: Fujitsu Toshiba Mobile Communications Limited (Fujitsu Limited)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Namekata, Minoru, Serizawa, Mutsumu
• Primary Examiner(s): Kuntz, Curtis
• Assistant Examiner(s): Bocure, Tesfaldet
• Application Number:
  US07/843,714
• Time in Patent Office: 669 Days
• Field of Search: 375/13, 375/14, 333/18, 333/28 R, 364/724.19, 364/724.20
• US Class Current: 375/231
• CPC Class Codes: H04L 25/03038 with a non-recursive structure H04L25/03031 takes precedence

Abstract

A method of equalization in a digital mobile communication capable of realizing an improved tracking of the variations of the transmission path characteristics, such that it becomes possible to achieve the good error rate characteristic. In the method, the forward and backward equalizations are carried out, while the forward and backward evaluation functions indicating reliabilites of the forward and backward equalizations are calculated by using the signal sequence, and a final equalization output is obtained by selecting for each bit of the signal sequence one of the forward and backward equalization outputs for which the reliability indicated by corresponding one of the forward and backward evaluation functions is higher as final equalization output for each bit. The detection of an abnormal state such as a sudden drop of a signal strength of the signal sequence or a limit cycle state of an equalizer may be used instead of the evaluation functions.

Claims

1. A method of equalization, comprising the steps of:

obtaining a signal sequence from received signals by applying frequency transformation and sampling;

carrying out a forward equalization of the signal sequence to obtain a forward equalization output in which the signal sequence is equalized in an order in which the received signals are received;

calculating a forward evaluation function indicating a reliability of the forward equalization by using the signal sequence;

carrying out a backward equalization of the signal sequence to obtain a backward equalization output in which the signal sequence is equalized in an order opposite to that in which the received signals are received;

calculating a backward evaluation function indicating a reliability of the backward equalization by using the signal sequence; and

obtaining a final equalization output by selecting for each bit of the signal sequence one of the forward equalization output and the backward equalization output for which the reliability indicated by a corresponding one of the forward evaluation function and the backward evaluation function is higher as the final equalization output for each bit, such that the final equalization output is given as a composition of those parts of the forward equalization output for which the reliability indicated by the forward evaluation function is higher than the reliability of the backward equalization output indicated by the backward evaluation function and those parts of the backward equalization output for which the reliability indicated by the backward evaluation function is higher than the reliability of the forward equalization output indicated by the forward evaluation function.

2. The method of claim 1, wherein the forward evaluation function and the backward evaluation function are correlation values of the signal sequence with respect to the forward equalization output and the backward equalization output, respectively.

3. The method of claim 2, wherein at the step of obtaining the final equalization output, the final equalization output is obtained by differentially demodulating the forward equalization output and the backward equalization, and wherein at the calculating steps the forward evaluation function and the backward evaluation function are calculated from the forward equalization output and the backward equalization output which are not differentially demodulated.

4. The method of claim 1, wherein the signal sequence comprises data to be equalized, and first and second training sequences which sandwich the data to be equalized, and wherein the forward equalization is carried out for the first training sequence and the data to be equalized from a top of the first training sequence to an end of the data to be equalized, while the backward equalization is carried out for the second training sequence and the data to be equalized from an end of the second training sequence to a top of the data to be equalized.

5. The method of claim 4, wherein the first and second training sequences are unique words assigned to different users, and at the steps of carrying out the forward and backward equalizations the first and second training sequences are identified from a number of predetermined training sequences.

6. The method of claim 1, wherein the forward evaluation function and the backward evaluation function are obtained as an absolute value of a low pass filtered linear sum of sequentially delayed signal sequences multiplied by a conjugate of the forward equalization output and the backward equalization output, respectively.

7. The method of claim 6, wherein the low pass filtered linear sum is obtained by integrating sequentially delayed linear sums with each sequentially delayed linear sum multiplied by a window function.

8. The method of claim 7, wherein the signal sequence comprises data to be equalized, and first and second training sequences which sandwich the data to be equalized, and wherein the forward equalization is carried out for the first training sequence and the data to be equalized from a top of the first training sequence to an end of the data to be equalized, while the backward equalization is carried out for the second training sequence and the data to be equalized from an end of the second training sequence to a top of the data to be equalized, and wherein the window function for the forward evaluation function is in a form of an exponential function which is decreasing toward the first training sequence, while the window function for the backward evaluation function is in a form of an exponential function which is decreasing toward the second training sequence.

9. The method of claim 6, wherein the forward evaluation function is further adjusted by using a multiplicative correction function for increasing values of the data to be equalized in a vicinity of a first training sequence, while the backward evaluation function is further adjusted by using a multiplicative correction function for increasing values of the data to be equalized in a vicinity of a second training sequence.

10. The method of claim 6, wherein the forward evaluation function is further adjusted by using an additive correction function for increasing values of the data to be equalized in a vicinity of a first training sequence, while the backward evaluation function is further adjusted by using an additive correction function for increasing values of the data to be evaluated in a vicinity of a second training sequence.

11. The method of claim 1, wherein the forward equalization and the backward equalization are carried out by using a decision feedback equalizer.

12. The method of claim 1, further comprising the step of temporarily storing the signal sequence, the forward equalization output, and the forward evaluation function in memory means, and wherein at the step of carrying out the backward equalization, the backward equalization is carried out by using the temporarily stored signal sequence at equalizer means used for carrying out the forward equalization, while at the step of calculating the backward evaluation function, the backward evaluation function is calculated by using the temporarily stored signal sequence at calculation means used for calculating the forward evaluation function, and at the step of obtaining the final equalization output, the temporarily stored forward evaluation function is compared with the backward evaluation function obtained by the calculation means, in response to a result of which the selecting is carried out from the temporarily stored forward equalization output and the backward equalization output obtained by the equalizer means.

13. A method of equalization, comprising the steps of:

obtaining a signal sequence from received signals by applying frequency transformation and sampling;

carrying out a forward equalization of the signal sequence to obtain a forward equalization output in which the signal sequence is equalized in an order in which the received signals are received;

carrying out a backward equalization of the signal sequence to obtain a backward equalization output in which the signal sequence is equalized in an order opposite to that in which the received signals are received;

detecting an abnormal state in one of the forward equalization and the backward equalization; and

obtaining a final equalization output by making a selection of one of the forward equalization output and the backward equalization output as the final equalization output until the abnormal state of said one of the forward equalization and the backward equalization is detected, and switching the selection for the final equalization output to another one of the forward equalization output and the backward equalization output after the abnormal state of said one of the forward equalization and the backward equalization is detected.

14. The method of claim 13, wherein the abnormal state to be detected at the detecting step is a lowering of a signal strength of the signal sequence below a predetermined level.

15. The method of claim 13, wherein the abnormal state to be detected at the detecting step is a start of a limit cycle state in an equalizer for obtaining said one of the forward equalization and the backward equalization in which a sum of absolute values of feed-forward tap coefficients of the equalizer becomes lower than a predetermined number.

16. A method of equalization, comprising the steps of:

obtaining a signal sequence from received signals by applying frequency transformation and sampling, where the signal sequence comprises data to be equalized, and first and second training sequences which sandwich the data to be equalized;

carrying out a forward equalization of the signal sequence to obtain a forward equalization output in which the signal sequence is equalized in an order in which the received signals are received, from a top of the first training sequence to an end of the second training sequence;

detecting a presence of errors in the forward equalization output for the second training sequence;

in a case the presence of the errors in the forward equalization output for the second training sequence is not detected, obtaining a final equalization output as the forward equalization output for the data to be equalized; and

in a case the presence of the errors in the forward equalization output for the second training sequence is detected,

calculating a forward evaluation function indicating a reliability of the forward equalization by using the signal sequence;

carrying out a backward equalization of the signal sequence to obtain a backward equalization output in which the signal sequence is equalized in an order opposite to that in which the received signals are received, from an end of the second training sequence to a top of the data to be equalized;

calculating a backward evaluation function indicating a reliability of the backward equalization by using the signal sequence; and

obtaining a final equalization output by selecting for each bit of the data to be equalized of the signal sequence one of the forward equalization output for the data to be equalized and the backward equalization output for the data to be equalized for which the reliability indicated by a corresponding one of the forward evaluation function and the backward evaluation function is higher as the final equalization output for each bit, such that the final equalization output is given as a composition of those parts of the forward equalization output for the data to be equalized for which the reliability indicated by the forward evaluation function is higher than the reliability of the backward equalization output for the data to be equalized indicated by the backward evaluation function and those parts of the backward equalization output for the data to be equalized for which the reliability indicated by the backward evaluation function is higher than the reliability of the forward equalization output for the data to be equalized indicated by the forward evaluation function.