Method and apparatus for adaptive real-time signal conditioning, processing, analysis, quantification, comparison, and control

• US 20040243659A1
• Filed: 10/04/2003
• Published: 12/02/2004
• Est. Priority Date: 10/07/2002
• Status: Active Grant

• Current Assignee: Alexei V. Nikitin
• Original Assignee: Alexei V. Nikitin
• Inventors: Nikitin, Alexei V.
• Granted Patent: US 7,107,306 B2
• Time in Patent Office: Days
• US Class Current: 708/819
• CPC Class Codes: G06F 2218/02 Preprocessing
• CPC Class Codes: G06G 7/02 Details not covered by G06G7/04 - G06G7/10, e.g. monitoring, construction, maintenance

Abstract

Various components of the present invention are collectively designated as Adaptive Real-Time Embodiments for Multivariate Investigation of Signals (ARTEMIS). It is a method, processes, and apparatus for measurement and analysis of variables of different type and origin. In this invention, different features of a variable can be quantified either locally as individual events, or on an arbitrary spatio-temporal scale as scalar fields in properly chosen threshold space. The method proposed herein overcomes limitations of the prior art by directly processing the data in real-time in the analog domain, identifying the events of interest so that continuous digitization and digital processing is not required, performing direct, noise-resistant measurements of salient signal characteristics, and outputting a signal proportional to these characteristics that can be digitized without the need for high-speed front-end sampling.

The application areas of ARTEMIS are numerous, e.g., it can be used for adaptive content-sentient real-time signal conditioning, processing, analysis, quantification, comparison, and control, and for detection, quantification, and prediction of changes in signals, and can be deployed in automatic and autonomous measurement, information, and control systems. ARTEMIS can be implemented through various physical means in continuous action machines as well as through digital means or computer calculations. Particular embodiments of the invention include various analog as well as digital devices, computer programs, and simulation tools.

Claims

1. Method for processing an input signal consisting of a Threshold Domain Filtering step utilizing a threshold domain and at least one of the following additional steps:

(a) Multimodal Pulse Shaping;

(b) Analog Rank Filtering; and

(c) Analog Counting.

2. Method for signal processing of claim 1 wherein said Threshold Domain Filtering step is used for extraction of one or more features of interest from said input signal to produce one or more output signals.

3. Method for signal processing of claim 1 wherein said threshold domain is defined by a Control Level Signal consisting of one or more components.

4. Method for signal processing of claim 1 wherein said Threshold Domain Filtering step outputs a first output value when said input signal is within said threshold domain and a second output value when said input signal is outside of said threshold domain.

5. Method for signal processing of claim 1 wherein said Threshold Domain Filtering step outputs a first output value when said input signal is within said threshold domain, a second output value when said input signal is outside of said threshold domain, and one or more values when said input signal is inside a finite interval at a boundary of said threshold domain.

6. Method for signal processing of claim 1 wherein at least one of said additional steps is Multimodal Pulse Shaping and wherein said input signal has one or more components and wherein said Multimodal Pulse Shaping transforms at least one of said components of said input signal into a plurality of signal components where at least one said component represents a derivative of another said component.

7. Method for signal processing of claim 6 wherein said derivative is a time-derivative.

8. Method for signal processing of claim 6 wherein said plurality of signal components is representative of one or more features of interest of said input signal.

9. Method for signal processing of claim 3 wherein one or more components of said Control Level Signal are outputs of a step of Analog Rank Filtering.

10. Method for signal processing of claim 3 wherein one or more components of said Control Level Signal are results of a step of Analog Rank Filtering of one or more output components of a step of Multimodal Pulse Shaping of said input signal.

11. Method for signal processing of claim 1 wherein at least one of said additional steps is Analog Counting, wherein said Threshold Domain Filtering step has an output with at least one component and where said Analog Counting comprises the steps of:

(a) time-differentiating said component to produce a time-derivative, and

(b) rectifying said time-derivative to produce an instantaneous count rate.

12. Method for signal processing of claim 11 wherein said Analog Counting further comprises the step of:

(a) time-integrating said instantaneous count rate to produce a count rate in a moving time window.

15. Method for signal processing of claim 3 wherein one or more components of said Control Level Signal are linear combinations of outputs of a step of Adaptive Analog Rank Filtering of one or more components of said input signal.

16. Method for signal processing of claim 3 wherein one or more components of said Control Level Signal are linear combinations of outputs of a step of Adaptive Analog Rank Filtering of one or more output components of a step of Multimodal Pulse Shaping of said input signal.

17. Method for signal processing of claim 3 wherein one or more components of said Control Level Signal are linear combinations of outputs of a step of Adaptive Analog Rank Selecting of a plurality of components of said input signal.

18. Method for signal processing of claim 3 wherein one or more components of said Control Level Signal are linear combinations of outputs of a step of Adaptive Analog Rank Selecting of a plurality of output components of a step of Multimodal Pulse Shaping of said input signal.

13. Method for Adaptive Analog Rank Filtering operable to transform an input scalar variable into an output Rank Filtered Variable comprising steps of:

(a) forming a plurality of outputs of delayed comparators by passing said input variable and a plurality of feedbacks of Offset Rank Filtered Variables through a respective plurality of said delayed comparators;

(b) forming a first difference which is a weighted difference of said outputs of the delayed comparators;

(c) forming a Density Function by time averaging said first difference;

(d) forming a second difference which is a weighted difference of said feedbacks of the Offset Rank Filtered Variables;

(e) forming a plurality of differences between the outputs of the delayed comparators and the respective Offset Quantile Parameters of said Offset Rank Filtered Variables;

(f) forming a plurality of time derivatives of said Offset Rank Filtered Variables by multiplying each of said plurality of differences by a ratio of said second difference and said Density Function;

(g) producing the plurality of the Offset Rank Filtered Variables by time-integrating said plurality of time derivatives; and

(h) producing said Rank Filtered Variable as a weighted average of said Offset Rank Filtered Variables.

14. Method for Adaptive Analog Rank Selecting operable to transform a plurality of input variables into an output Rank Selected Variable comprising steps of:

(a) forming a plurality of outputs of averaging comparators by passing said plurality of input variables and each component of a plurality of feedbacks of Offset Rank Selected Variables through a respective plurality of said averaging comparators;

(b) forming a first difference which is a weighted difference of said outputs of the averaging comparators;

(c) forming a Density Function by time averaging said first difference;

(d) forming a second difference which is a weighted difference of said feedbacks of the Offset Rank Selected Variables;

(e) forming a plurality of differences between the outputs of the averaging comparators and the respective Offset Quantile Parameters of said Offset Rank Selected Variables;

(f) forming a plurality of time derivatives of said Offset Rank Selected Variables by multiplying each of said plurality of differences by a ratio of said second difference and said Density Function;

(g) producing the plurality of the Offset Rank Selected Variables by time-integrating said plurality of the time derivatives; and

(h) producing said Rank Selected Variable as a weighted average of said Offset Rank Selected Variables.

19. A method for signal processing operable to transform an input signal into an output signal comprising a Threshold Domain Filtering step performing Threshold Domain Filtering and at least one additional step operable to perform a function selected from the group consisting of Multimodal Pulse Shaping, Analog Rank Filtering, and Analog Counting.

20. A method for signal processing as recited in claim 19 wherein said Threshold Domain Filtering step is operable to extract one or more features of interest from said input signal to produce one or more output signals.

21. A method for signal processing as recited in claim 19 wherein a threshold domain of said Threshold Domain Filtering is defined by a Control Level Signal, said Control Level Signal consisting of one or more components.

22. A method for signal processing as recited in claim 19 wherein said Threshold Domain Filtering step outputs a first output value when said input signal is within a threshold domain of said Threshold Domain Filtering and a second output value when said input signal is outside of said threshold domain.

23. A method for signal processing as recited in claim 19 wherein said Threshold Domain Filtering step outputs a first output value when said input signal is within a threshold domain of said Threshold Domain Filtering, a second output value when said input signal is outside of said threshold domain, and one or more values when said input signal is inside a finite interval at a boundary of said threshold domain.

24. A method for signal processing as recited in claim 19 wherein at least one of said additional steps is operable to perform Multimodal Pulse Shaping and wherein said input signal has one or more components and wherein said Multimodal Pulse Shaping transforms at least one of said input signal components into a plurality of output signal components where at least one of said output signal components represents a derivative of another of said output signal components.

25. A method for signal processing as recited in claim 24 wherein said derivative is a time-derivative.

26. A method for signal processing as recited in claim 24 wherein said plurality of output signal components is representative of one or more features of interest of said input signal.

27. A method for signal processing as recited in claim 19 wherein at least one of said additional steps is Analog Counting operable to perform an Analog Counting function and wherein said Threshold Domain Filtering step has an output with at least one component and where said Analog Counting comprises the steps of:

a. time-differentiating said component to produce a time-derivative, and

b. rectifying said time-derivative to produce an instantaneous count rate.

28. A method for signal processing as recited claim 27 wherein said Analog Counting further comprises the step of time-integrating said instantaneous count rate in a moving time window to produce a count rate.

29. A method for signal processing as recited in claim 21 wherein at least one of said additional steps is operable to perform Analog Rank Filtering and wherein said one or more components of said Control Level Signal are linear combinations of outputs of said Analog Rank Filtering.

30. A method for signal processing as recited in claim 29 wherein said input signal has one or more components and wherein said Analog Rank Filtering is performed on one or more components of said input signal.

31. A method for signal processing as recited in claim 21 wherein at least one of said additional steps is operable to perform Analog Rank Filtering and wherein at least one of said additional steps is operable to perform Multimodal Pulse Shaping and wherein said one or more components of said Control Level Signal are produced by applying said Analog Rank Filtering to one or more output components of said Multimodal Pulse Shaping.

32. A method for signal processing operable to transform an input signal into an output signal comprising the steps of:

a. forming a plurality of outputs of delayed comparators by passing said input signal and a plurality of feedbacks of Offset Rank Filtered Signals through a respective plurality of said delayed comparators, said Offset Rank Filtered Signals having Offset Quantile Parameters;

b. forming a first difference which is a weighted difference of said outputs of the delayed comparators;

c. forming a Density Function by time averaging said first difference;

d. forming a second difference which is a weighted difference of said feedbacks of the Offset Rank Filtered Signals;

e. forming a plurality of differences between the outputs of the delayed comparators and the respective Offset Quantile Parameters of said Offset Rank Filtered Signals;

f. forming a plurality of time derivatives of said Offset Rank Filtered Signals by multiplying each of said plurality of differences by a ratio of said second difference and said Density Function;

g. producing the plurality of said Offset Rank Filtered Signals by time-integrating said plurality of time derivatives; and

h. producing said output signal as a weighted average of said Offset Rank Filtered Signals.

33. A method for signal processing as recited in claim 32 wherein:

a. said plurality of outputs of delayed comparators consists of two outputs and said plurality of feedbacks of Offset Rank Filtered Signals consists of two feedbacks and said plurality of said delayed comparators consists of two delayed comparators;

b. said first difference is a difference of said two outputs of said two delayed comparators;

c. said second difference is a difference of said two feedbacks;

d. said plurality of differences consists of two differences;

e. said plurality of time derivatives consists of two time derivatives;

f. said plurality of the Offset Rank Filtered Signals consists of two Offset Rank Filtered Signals; and

g. said weighted average of said Offset Rank Filtered Signals is an average of said two Offset Rank Filtered Signals.

34. A method for signal processing operable to transform a plurality of input signals into an output signal comprising the steps of:

a. forming a plurality of outputs of averaging comparators by passing said plurality of input signals and a plurality of feedbacks of Offset Rank Selected Signals through a plurality of said averaging comparators, said Offset Rank Selected Signals having Offset Quantile Parameters;

b. forming a first difference which is a weighted difference of said outputs of the averaging comparators;

c. forming a Density Function by time averaging said first difference;

d. forming a second difference which is a weighted difference of said feedbacks of the Offset Rank Selected Signals;

e. forming a plurality of differences between the outputs of the averaging comparators and the respective Offset Quantile Parameters of said Offset Rank Selected Signals;

f. forming a plurality of time derivatives of said Offset Rank Selected Signals by multiplying each of said plurality of differences by a ratio of said second difference and said Density Function;

g. producing the plurality of said Offset Rank Selected Signals by time-integrating said plurality of the time derivatives; and

h. producing said output signal as a weighted average of said Offset Rank Selected Signals.

35. A method for signal processing as recited in claim 34 wherein:

a. said plurality of outputs of averaging comparators consists of two outputs and said plurality of feedbacks of Offset Rank Selected Signals consists of two feedbacks and said plurality of said averaging comparators consists of two averaging comparators;

b. said first difference is a difference of said two outputs of said two averaging comparators;

c. said second difference is a difference of said two feedbacks;

d. said plurality of differences consists of two differences;

e. said plurality of time derivatives consists of two time derivatives;

f. said plurality of the Offset Rank Selected Signals consists of two Offset Rank Selected Signals; and

g. said weighted average of said Offset Rank Selected Signals is an average of said two Offset Rank Selected Signals.

36. A method for image processing operable to transform an input image signal into an output signal comprising the steps of:

a. forming a plurality of outputs of delayed comparators by passing said input image signal and a plurality of feedbacks of Offset Rank Filtered Signals through a respective plurality of said delayed comparators, said Offset Rank Filtered Signals having Offset Quantile Parameters;

b. forming a first difference which is a weighted difference of said outputs of the delayed comparators;

c. forming a Density Function by time averaging said first difference;

d. forming a second difference which is a weighted difference of said feedbacks of the Offset Rank Filtered Signals;

e. forming a plurality of differences between the outputs of the delayed comparators and the respective Offset Quantile Parameters of said Offset Rank Filtered Signals;

f. forming a plurality of time derivatives of said Offset Rank Filtered Signals by multiplying each of said plurality of differences by a ratio of said second difference and said Density Function;

g. producing the plurality of said Offset Rank Filtered Signals by time-integrating said plurality of time derivatives; and

h. producing said output signal as a weighted average of said Offset Rank Filtered Signals.

37. A method for image processing as recited in claim 36 wherein:

a. said plurality of outputs of delayed comparators consists of two outputs and said plurality of feedbacks of Offset Rank Filtered Signals consists of two feedbacks and said plurality of said delayed comparators consists of two delayed comparators;

b. said first difference is a difference of said two outputs of said two delayed comparators;

c. said second difference is a difference of said two feedbacks;

d. said plurality of differences consists of two differences;

e. said plurality of time derivatives consists of two time derivatives;

f. said plurality of the Offset Rank Filtered Signals consists of two Offset Rank Filtered Signals; and

g. said weighted average of said Offset Rank Filtered Signals is an average of said two Offset Rank Filtered Signals.

38. A method for image processing operable to transform a plurality of input image signals into an output signal comprising the steps of:

a. forming a plurality of outputs of averaging comparators by passing said plurality of input image signals and a plurality of feedbacks of Offset Rank Selected Signals through a plurality of said averaging comparators, said Offset Rank Selected Signals having Offset Quantile Parameters;

b. forming a first difference which is a weighted difference of said outputs of the averaging comparators;

c. forming a Density Function by time averaging said first difference;

d. forming a second difference which is a weighted difference of said feedbacks of the Offset Rank Selected Signals;

e. forming a plurality of differences between the outputs of the averaging comparators and the respective Offset Quantile Parameters of said Offset Rank Selected Signals;

f. forming a plurality of time derivatives of said Offset Rank Selected Signals by multiplying each of said plurality of differences by a ratio of said second difference and said Density Function;

g. producing the plurality of said Offset Rank Selected Signals by time-integrating said plurality of the time derivatives; and

h. producing said output signal as a weighted average of said Offset Rank Selected Signals.

39. A method for image processing as recited in claim 38 wherein:

a. said plurality of outputs of averaging comparators consists of two outputs and said plurality of feedbacks of Offset Rank Selected Signals consists of two feedbacks and said plurality of said averaging comparators consists of two averaging comparators;

b. said first difference is a difference of said two outputs of said two averaging comparators;

c. said second difference is a difference of said two feedbacks;

d. said plurality of differences consists of two differences;

e. said plurality of time derivatives consists of two time derivatives;

f. said plurality of the Offset Rank Selected Signals consists of two Offset Rank Selected Signals; and

g. said weighted average of said Offset Rank Selected Signals is an average of said two Offset Rank Selected Signals.

40. An apparatus for signal processing operable to transform an input signal into an output signal comprising a Threshold Domain Filtering module operable to perform Threshold Domain Filtering and at least one additional module operable to perform a function selected from the group consisting of Multimodal Pulse Shaping, Analog Rank Filtering, and Analog Counting.

41. An apparatus for signal processing as recited in claim 40 wherein said Threshold Domain Filtering module is operable to extract one or more features of interest from said input signal to produce one or more output signals.

42. An apparatus for signal processing as recited in claim 40 wherein a threshold domain of said Threshold Domain Filtering is defined by a Control Level Signal, said Control Level Signal consisting of one or more components.

43. An apparatus for signal processing as recited in claim 40 wherein said Threshold Domain Filtering module outputs a first output value when said input signal is within a threshold domain of said Threshold Domain Filtering and a second output value when said input signal is outside of said threshold domain.

44. An apparatus for signal processing as recited in claim 40 wherein said Threshold Domain Filtering module outputs a first output value when said input signal is within a threshold domain of said Threshold Domain Filtering, a second output value when said input signal is outside of said threshold domain, and one or more values when said input signal is inside a finite interval at a boundary of said threshold domain.

45. An apparatus for signal processing as recited in claim 40 wherein at least one of said additional modules is a Multimodal Pulse Shaping and wherein said input signal has one or more components and wherein said Multimodal Pulse Shaping transforms at least one of said input signal components into a plurality of output signal components where at least one of said output signal components represents a derivative of another of said output signal components.

46. An apparatus for signal processing as recited in claim 45 wherein said derivative is a time-derivative.

47. An apparatus for signal processing as recited in claim 45 wherein said plurality of output signal components is representative of one or more features of interest of said input signal.

48. An apparatus for signal processing as recited in claim 40 wherein at least one of said additional modules is an Analog Counting module operable to perform an Analog Counting function and wherein said Threshold Domain Filtering module has an output with at least one component and where said Analog Counting module comprises:

a. a differentiator operable to time-differentiate said component to produce a time-derivative, and

b. a rectifier operable to rectify said time-derivative to produce an instantaneous count rate.

49. An apparatus for signal processing as recited claim 48 wherein said Analog Counting module further comprises an integrator operable to time-integrate said instantaneous count rate in a moving time window to produce a count rate.

50. An apparatus for signal processing as recited in claim 42 wherein at least one of said additional modules is operable to perform Analog Rank Filtering and wherein said one or more components of said Control Level Signal are linear combinations of outputs of said Analog Rank Filtering.

51. An apparatus for signal processing as recited in claim 50 wherein said input signal has one or more components and wherein said Analog Rank Filtering is performed on one or more components of said input signal.

52. An apparatus for signal processing as recited in claim 42 wherein at least one of said additional modules is operable to perform Analog Rank Filtering and wherein at least one of said additional modules is operable to perform Multimodal Pulse Shaping and wherein said one or more components of said Control Level Signal are produced by applying said Analog Rank Filtering to one or more output components of said Multimodal Pulse Shaping.

53. An apparatus for signal processing operable to transform an input signal into an output signal comprising:

a. a plurality of delayed comparators operable to form a plurality of outputs by passing said input signal and a plurality of feedbacks of Offset Rank Filtered Signals through said plurality of delayed comparators, said Offset Rank Filtered Signals having Offset Quantile Parameters;

b. a component operable to form a first difference which is a weighted difference of said outputs of said plurality of delayed comparators;

c. a component operable to form a Density Function by time averaging said first difference;

d. a component operable to form a second difference which is a weighted difference of said feedbacks of the Offset Rank Filtered Signals;

e. a component operable to form a plurality of differences between the outputs of said plurality of delayed comparators and the respective Offset Quantile Parameters of said Offset Rank Filtered Signals;

f. a component operable to form a plurality of time derivatives of said Offset Rank Filtered Signals by multiplying each of said plurality of differences by a ratio of said second difference and said Density Function;

g. a component operable to produce the plurality of said Offset Rank Filtered Signals by time-integrating said plurality of time derivatives; and

h. a component operable to produce said output signal as a weighted average of said Offset Rank Filtered Signals.

54. An apparatus for signal processing as recited in claim 53 wherein:

a. said plurality of delayed comparators consists of two delayed comparators and said plurality of outputs consists of two outputs and said plurality of feedbacks of Offset Rank Filtered Signals consists of two feedbacks;

b. said first difference is a difference of said two outputs of said two delayed comparators;

c. said second difference is a difference of said two feedbacks;

d. said plurality of differences consists of two differences;

e. said plurality of time derivatives consists of two time derivatives;

f. said plurality of the Offset Rank Filtered Signals consists of two Offset Rank Filtered Signals; and

g. said weighted average of said Offset Rank Filtered Signals is an average of said two Offset Rank Filtered Signals.

55. An apparatus for signal processing operable to transform a plurality of input signals into an output signal comprising:

a. a plurality of averaging comparators operable to form a plurality of outputs by passing said plurality of input signals and a plurality of feedbacks of Offset Rank Selected Signals through said plurality of averaging comparators, said Offset Rank Selected Signals having Offset Quantile Parameters;

b. a component operable to form a first difference which is a weighted difference of said outputs of the averaging comparators;

c. a component operable to form a Density Function by time averaging said first difference;

d. a component operable to form a second difference which is a weighted difference of said feedbacks of the Offset Rank Selected Signals;

e. a component operable to form a plurality of differences between the outputs of said plurality of averaging comparators and the respective Offset Quantile Parameters of said Offset Rank Selected Signals;

f. a component operable to form a plurality of time derivatives of said Offset Rank Selected Signals by multiplying each of said plurality of differences by a ratio of said second difference and said Density Function;

g. a component operable to produce the plurality of said Offset Rank Selected Signals by time-integrating said plurality of the time derivatives; and

h. a component operable to produce said output signal as a weighted average of said Offset Rank Selected Signals.

56. An apparatus for signal processing as recited in claim 55 wherein:

a. said plurality of said averaging comparators consists of two averaging comparators and said plurality of outputs consists of two outputs and said plurality of feedbacks of Offset Rank Selected Signals consists of two feedbacks;

b. said first difference is a difference of said two outputs of said two averaging comparators;

c. said second difference is a difference of said two feedbacks;