Method and apparatus for quantizing a signal in a digital system

• US 6,128,346 A
• Filed: 04/14/1998
• Issued: 10/03/2000
• Est. Priority Date: 04/14/1998
• Status: Expired due to Fees

• Current Assignee: Motorola Solutions, Inc.
• Original Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Inventors: Suarez, Jose' I., Prieto, Yolanda
• Primary Examiner(s): Pham, Chi H.
• Assistant Examiner(s): WEBSTER, BRYAN E
• Application Number:
  US09/059,823
• Time in Patent Office: 903 Days
• Field of Search: 375/242, 375/243, 375/245, 375/254, 341/77, 341/143
• US Class Current: 375/254
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/63 using sub-band based transform, e.g. wavelets
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

An input data signal (102) in a digital signal processing system gets adaptively quantized by performing a LMS error algorithm on a signal of transformed coefficients (113), determining paths of minimum error (202), and forming a three-dimensional matrix (206) representing these paths. Centroids are computed (208) based on the paths of minimum error (206), and these computed centroids are used as part of a next iteration to represent new quantization states and levels.

Claims

1. A method of quantizing a data signal in a digital signal processing system, comprising the steps of:

receiving a series of transformed coefficients as the data signal;

performing a minimum error algorithm on the transformed coefficients to determine paths of minimum error;

forming a three-dimensional matrix representing the paths of minimum error;

computing centroids based on the paths of minimum error;

using the computed centroids as part of a next iteration where the computed centroids are used to represent new quantization states and levels;

repeating the steps of performing through using until a predetermined optimization of centroids is met; and

transmitting the centroids as the quantized signal.

2. The method of claim 1, wherein the minimum error algorithm is characterized by a trellis least mean square (LMS) algorithm.

23. The method of claim 1, wherein the optimization of centroids is determined based on a predetermined signal to noise ratio (SNR) threshold.

24. The method of claim 1, wherein the optimization of centroids is based on an entropy threshold.

25. The method of claim 1, wherein the optimization of centroids is based on a bit rate threshold.

3. A method of quantizing a signal, comprising the steps of:

receiving an input data signal having a data range;

dividing the input data range into a predetermined number of quantization steps;

allocating the quantization steps to a number of quantization states and levels;

running the quantized states and levels through an error minimization algorithm;

adaptively computing centroid values from the input data of each quantizer state and level;

updating the quantizer states and levels based on the computed centroids;

repeating the steps of dividing through updating until a predetermined optimization of centroids is achieved; and

transmitting the updated centroids as quantizer outputs once the predetermined optimized centroids are achieved.

4. The method of claim 3, wherein the step of running the quantized states and levels through a minimum error algorithm includes the steps of:

storing an index pointing to the quantizer state and level for each input of sampled data; and

forming a matrix of the stored index locations for each quantization state and level.

5. The method of claim 3, wherein the minimum error algorithm consists of a least mean square (LMS) algorithm.

6. The method of claim 3, wherein the predetermined optimization of centroids is based on a predetermined error threshold.

7. The method of claim 6, wherein the step of receiving the input data signal includes the step of decimating the input data signal prior to the step of dividing.

8. The method of claim 7, wherein the step of decimating comprises the step of taking every other data sample from the data range.

9. The method of claim 8, further including the step of receiving a full string of input data to run through a final iteration of the steps of receiving through updating.

10. The method of claim 3, wherein a signal-to-noise ratio (SNR) value or change in value is determined based on the computed centroids.

11. The method of claim 10, wherein the optimization of centroids is based on comparing the SNR value to a predetermined threshold;

and

repeating the steps of re-running the minimum error algorithm, adaptively computing centroid values, updating the states and levels, and obtaining a SNR value until the SNR value or change in value reaches the predetermined threshold.

12. The method of claim 3, where the step dividing comprises the step of uniformly dividing the data range.

13. The method of claim 3, wherein the LMS algorithm is a Viterbi algorithm.

14. The method of claim 3, wherein a shrink factor is applied to the quantization steps.

15. The method of claim 14, wherein the step of dividing further includes the step of decimating the input data signal.

16. A method of quantizing an input data signal in a digital processing system, comprising the steps of:

receiving the data signal;

running a trellis path optimization on the input data signal to generate a history of minimum error paths;

forming a path matrix based on the minimum error paths;

computing a centroid based on the path matrix;

re-running the trellis path optimization to produce non-uniform allocation of quantizer states and levels; and

transmitting the computed centroids as quantizer outputs.

17. The method of claim 16, wherein the data signal is a decimated data signal.

18. The method of claim 16, where the trellis path optimization is mirror imaged.

26. The method of claim 16, wherein the computed centroids are transmitted to a coding stage.

19. A data CODEC system for a digital communication system, comprising:

a decomposition stage for receiving an input data signal and generating a signal of transformed coefficients;

an adaptive centroid based quantizer for receiving the transformed coefficients and generating optimal non-uniform spaced quantizers, wherein the adaptive centroid based quantizer includes a least means square (LMS) stage to generate a matrix of minimum error paths of the quantizers and a centroid computation stage for computing and updating a centroid calculation of the matrix of minimum error path;

a coding stage for coding the quantizers into a coded signal;

a channel for transmitting the coded signal; and

a decoder for decoding the transmitted signal.

20. A data CODEC as described in claim 19, wherein the input data signal is first converted into a one dimensional (1-D) data input stream meeting a predetermined threshold in which unwanted or repetitive data is discarded.

21. A data CODEC as described in claim 19, wherein the decoder includes an inverse of the adaptive centroid based quantizer for converting the transmitted signal into a decoded signal.

22. A data CODEC as described in claim 19, wherein non-uniformly spaced quantizers are generated until a predetermined error measurement or bit rate is achieved.

27. A data CODEC system for a digital communication system, comprising:

a decomposition stage for receiving an input data signal and generating a signal of transformed coefficients;

an adaptive centroid based quantizer having a matrix generation stage and a centroid computation stage for receiving the transformed coefficients and generating optimal non-uniform spaced quantizers;

a coding stage for coding the quantizers into a coded signal;

a channel for transmitting the coded signal; and

a decoder for decoding the transmitted signal.

28. A data CODEC as described in claim 27, wherein the input data signal is first converted into a one dimensional (1-D) data input stream meeting a predetermined threshold in which unwanted or repetitive data is discarded.

29. A data CODEC as described in claim 27, wherein the decoder includes an inverse of the adaptive centroid based quantizer for converting the transmitted signal into a decoded signal.

30. A data CODEC as described in claim 27, wherein non-uniformly spaced quantizers are generated until a predetermined error measurement or bit rate is achieved.

31. A data CODEC as described in claim 27, wherein the channel is a wireless channel.

32. A data CODEC as described in claim 27, wherein the channel is a wireline channel.

33. A data CODEC as described in claim 27, wherein the channel is a optically linked path.

34. A data CODEC as described in claim 27, wherein the channel is an acoustic path.

35. A data CODEC as described in claim 27, wherein the channel is a fiber optic transmission path.

36. A data CODEC as described in claim 27, wherein the channel is a radio frequency (RF) channel.