Nonlinear statistical signal processor

Claim 2

A signal processor comprising: an input for receiving an input signal y, wherein: said input signal y has in it a signal x; and said input signal y has a conditional density function given that the signal x is present, p(y/x); and said signal x has an amplitude-density function p(x); a memory for storing the amplitude-density function p(x) of the signal x, wherein said memory is coupled to said input so that said memory is addressed by said input signal y; an amplitude-density function generator coupled to said input and to said memory for generating a combined density function p(y|x)p(x) comprising a product of said conditional density function p(y|x) and said amplitude-density function p(x): a threshold scaler comprising; an integrator for integrating said combined density function to produce an integrated product; an input for receiving an externally supplied threshold-detection level; and means for combining said integrated product and said externally supplied threshold-detection level to produce a scaled detection threshold, where said combining means is connected to a comparator for applying said scaled detection threshold to said comparator.

Claim 3

The signal processor of claim 1 wherein said combined density function has a peak, and said signal processor additionally comprises means for detecting the peak of said combined density function.

Claim 4

The signal processor of claim 2 wherein said comparator determines whether said peak of said combined density function exceeds a defined threshold, indicating that the signal x is indeed contained in the input signal y, and for producing a "detection achieved" output signal.

Claim 5

The signal processor of claim 2 additionally comprising means for reconstructing an estimate of the signal x from the peak of said combined density function.

Claim 6

The signal processor of claim 4 wherein said reconstruction means comprises a hold circuit for receiving an output of a counter, and which is written by said peak of said combined density function.

Claim 7

The signal processor of claim 5 additionally comprising means for updating the amplitude density function p(x) stored in said memory based on said estimate of the signal x.

Claim 8

The signal processor of claim 6 wherein said estimate of the signal x has an amplitude-density function and said updating means comprises a second amplitude-density-function generator having as its input said estimate of the signal x, and producing as its output the amplitude-density function of said estimate.

Claim 9

The signal processor of claim 7, additionally comprising: a comparator for determining whether the peak of said combined density function exceeds a defined threshold, and for producing a "detection achieved" output signal when said peak exceeds said threshold; and means for coupling the output of said comparator to a second counter to control said second counter so that said second counter is enabled only when said "detection achieved" signal is produced.

Claim 10

A signal estimator for estimating a signal x present in an input signal y, wherein said signal x has an amplitude-density function p(x) and said signal y has a conditional density function p(y|x) given that the signal x is present, said signal estimator comprising: an input for receiving said input signal y; a counter; a first memory for storing said amplitude-density function p(x), wherein said first memory is addressed by said input signal y; a second memory; an incrementer for adding the output of said first memory to the value stored at a given location of said second memory when said input signal y has a predetermined amplitude; a scaler coupled to an output of said second memory and to said counter for scaling the value stored at a given location in said second memory in response to an output from said counter, and storing at least a plurality of most-significant bits of the scaled value back at said given location of said second memory to generate a combined amplitude-density function p(x) p(y|x); a peak-hold circuit coupled to said output of said second memory for detecting and storing a peak of said combined amplitude density function; and a hold circuit comprising: a write enable input coupled to an output of said peak hold circuit to enable writing into said hold circuit said location of a detected peak of said combined amplitude density function; an address input coupled to an output of said counter so that said detected peak is written at an address corresponding to an output of said counter, so that said hold circuit generates an estimate of the signal x; and a signal estimate output line, on which is output said estimate of the signal x.

Claim 11

A signal estimator for estimating a signal x present in an input signal y, wherein said signal x has an amplitude-density function p(x) and said signal y has a conditional density function p(y|x) given that the signal x is present, said signal estimator comprising: an input for receiving said input signal y; a counter; a first memory for storing said amplitude-density function p(x), wherein said first memory is addressed by said input signal y; a second memory; an incrementer for adding the output of said first memory to the value stored at a given location of said second memory when said input signal y has a predetermined amplitude; a scaler coupled to an output of said second memory and to said counter for scaling the value stored at a given location in said second memory in response to an output from said counter, and storing at least a plurality of most-significant bits of the scaled value back at said given location of said second memory to generate a combined amplitude-density function p(x) p(y|x); a peak-hold circuit coupled to said output of said second memory for detecting and storing a peak of said combined amplitude density function; and a hold circuit comprising: a write enable input coupled to an output of said peak hold circuit to enable writing into said hold circuit said location of a detected peak of said combined amplitude density function; an address input coupled to an output of said counter so that said detected peak is written at an address corresponding to an output of said counter, so that said hold circuit generates an estimate of the signal x; and a signal estimate output line, on which is output said estimate of the signal x.

Claim 12

The signal estimator of claim 9, additionally comprising a third memory, and wherein said plurality of most-significant bits of the output of said scaler are stored in said second memory and a plurality of least-significant bits of the output of said scaler are stored in said third memory.

Claim 13

The signal estimator of claim 10, wherein said first memory comprises random-access memory, and said signal estimator additionally comprises means for updating the amplitude-density function p(x) stored in said first memory based on the signal on said signal estimate output line.

Claim 14

The signal estimator of claim 11, wherein said updating means comprises: a fourth memory; a fifth memory; an incrementer for incrementing the value stored at a given location in said fourth memory when said signal estimate has a predetermined amplitude; a scaler for scaling the value stored at a given location in said fourth memory in response to an output from a second counter, and for storing at least a plurality of the most-significant bits of the scaled value back at said given location of said fourth memory to generate the amplitude density function of said signal estimate.

Claim 15

The signal estimator of claim 9, wherein said first memory comprises random-access memory, and said signal estimator additionally comprises means for updating the amplitude-density function p(x) stored in said first memory based on the signal estimate on said signal estimate output line.

Claim 16

The signal estimator of claim 13, wherein said updating means comprises: a third memory; a second counter; an incrementer for incrementing the value stored at a given location in said third memory when said signal estimate has a predetermined amplitude; a scaler for scaling the value stored at a given location in said third memory in response to an output from said second counter, and storing at least a plurality of most significant bits of the scaled value back at said given location of said third memory to generate the amplitude density function of said signal estimate.

Claim 17

The signal estimator of claim 14, additionally comprising: a comparator for comparing the output of said peak-hold circuit with a predetermined threshold and determining whether the peak value of said combined amplitude-density function p(x) p(y|x) exceeds a predetermined threshold; and means for coupling the output of said comparator to said second counter to control said second counter so that said second counter is enabled only when said combined amplitude-density function exceeds said threshold.

Claim 18

The signal estimator of claim 15, additionally comprising a threshold scaler comprising: an integrator for integrating said combined density function p(x) p(y|x); an input for receiving an externally supplied threshold-detection level; and means for combining said integrated combined density function and said externally supplied detection threshold to generate a scaled detection threshold, wherein said combining means is connected to said comparator for applying said scaled detection threshold to said comparator.

Claim 19

A minimum-mean-squared-error signal estimator comprising: an input for receiving an input signal y, wherein: said input signal y has in it a signal x; said signal x has an amplitude density function p(x); said input signal y has a conditional density function, given that the signal x is present, p(y|x); a counter; a memory for storing the amplitude-density function p(x) of the signal x, wherein said memory is coupled to said input so that said memory is addressed by said input signal y; an amplitude-density-function generator coupled to said input and to said memory for generating a combined density function p(x) p(y|x) comprising a product of said amplitude-density function p(x) of said signal x and said conditional density function p(y|x); a first integrator coupled to an output of said amplitude-density-function generator for generating a combined distribution function by accumulating said combined density function over time; a multiplier coupled to the output of said amplitude-density-function generator and to an output of said counter for generating the product of said combined density function and a count; a second integrator coupled to an output of said multiplier for accumulating over time said product of said combined density function and said count; a divider coupled to an output of said first integrator and to an output of said second integrator for dividing the output of said second integrator by the output of said first integrator; and an output line coupled to an output of said divider.

Claim 20

A minimum-mean-squared-error signal estimator comprising: an input for receiving an input signal y, wherein: said input signal y has in it a signal x; said signal x has an amplitude density function p(x); said input signal y has a conditional density function, given that the signal x is present, p(y|x); a counter; a memory for storing the amplitude-density function p(x) of the signal x, wherein said memory is coupled to said input so that said memory is addressed by said input signal y; an amplitude-density-function generator coupled to said input and to said memory for generating a combined density function p(x) p(y|x) comprising a product of said amplitude-density function p(x) of said signal x and said conditional density function p(y|x); a first integrator coupled to an output of said amplitude-density-function generator for generating a combined distribution function by accumulating said combined density function over time; a multiplier coupled to the output of said amplitude-density-function generator and to an output of said counter for generating the product of said combined density function and a count; a second integrator coupled to an output of said multiplier for accumulating over time said product of said combined density function and said count; a divider coupled to an output of said first integrator and to an output of said second integrator for dividing the output of said second integrator by the output of said first integrator; and an output line coupled to an output of said divider.

Claim 21

The minimum-mean-squared-error signal estimator of claim 17, wherein the output fo said divider comprises an estimate of the signal x, and said minimum-mean-squared error signal estimator additionally comprises means for updating the amplitude-density function p(x) stored in said memory based on the signal on said output line.

Claim 22

The minimum-mean-squared-error signal estimator of claim 17, wherein said amplitude-density-function generator comprises: a second memory; an incrementer coupled to said first memory for adding the output of said first memory to the value stored at a given location of said second memory in response to an input signal y of a predetermined amplitude; a scaler for scaling the value stored at a given location in said second memory in response to an output from said counter, and for storing at least a plurality of most-significant bits of the scaled value back at said given location of said second memory to generate said combined amplitude-density function p(x) p(y|x); and a density function output line coupled to said second memory for said combined density function.

Claim 23

The minimum-mean-squared-error signal estimator of claim 19, wherein said first memory comprises random-access memory, and said signal estimator additionally comprises means for updating the amplitude-density function p(x) stored in said first memory based on the signal on said signal estimate output line, said updating means comprising: a third memory; a second counter; an incrementer for incrementing the value stored at a given location in said third memory when said signal estimate has a predetermined amplitude; a scaler for scaling the value stored at a given location in said second memory in response to an output from said second counter, and storing at least a plurality of most-significant bits of the scaled value back at said given location of said third memory to generate the amplitude-density function of said signal estimate.

Claim 24

A minimum-mean-squared-error signal estimator for estimating a signal x present in an input signal y, wherein said signal x has an amplitude-density function p(x) and said signal y has a conditional density function p(y|x) given that the signal x is present, said signal estimator comprising: an input for receiving said input signal y; a counter; a first memory for storing said amplitude-density function p(x), wherein said first memory is addressed by said input signal y; a second memory; an incrementer coupled to said first memory for adding an output of said first memory to the value stored at a given location of said second memory in response to an input signal y of a predetermined amplitude; a scaler for scaling the value stored at a given location in said second memory in response to an output from said counter, and for storing at least a plurality of most-significant bits of the scaled value back at said given location of said second memory to generate a combined amplitude-density function p(x) p(y|x); a density-function output line coupled to said second memory for said combined amplitude density function p(x) p(y|x); a first integrator coupled to said density-function output line for generating a combined distribution function by accumulating said combined density function over time; a multiplier coupled to said density function output line and to said counter for generating a product of said combined density function and a count; a second integrator coupled to an output of said multiplier for accumulating said product over time; a divider coupled to an output of said first integrator and to an output of said second integrator for dividing the output of said second integrator by the output of said first integrator to generate an estimate of said signal x; and an estimate output line coupled to the output of said divider for said signal estimate.

Claim 25

A minimum-mean-squared-error signal estimator for estimating a signal x present in an input signal y, wherein said signal x has an amplitude-density function p(x) and said signal y has a conditional density function p(y|x) given that the signal x is present, said signal estimator comprising: an input for receiving said input signal y; a counter; a first memory for storing said amplitude-density function p(x), wherein said first memory is addressed by said input signal y; a second memory; an incrementer coupled to said first memory for adding an output of said first memory to the value stored at a given location of said second memory in response to an input signal y of a predetermined amplitude; a scaler for scaling the value stored at a given location in said second memory in response to an output from said counter, and for storing at least a plurality of most-significant bits of the scaled value back at said given location of said second memory to generate a combined amplitude-density function p(x) p(y|x); a density-function output line coupled to said second memory for said combined amplitude density function p(x) p(y|x); a first integrator coupled to said density-function output line for generating a combined distribution function by accumulating said combined density function over time; a multiplier coupled to said density function output line and to said counter for generating a product of said combined density function and a count; a second integrator coupled to an output of said multiplier for accumulating said product over time; a divider coupled to an output of said first integrator and to an output of said second integrator for dividing the output of said second integrator by the output of said first integrator to generate an estimate of said signal x; and an estimate output line coupled to the output of said divider for said signal estimate.

Claim 26

The minimum-mean-squared-error signal estimator of claim 21, wherein said first memory comprises random access memory, and said signal estimator additionally comprises means for updating the amplitude-density function p(x) stored in said first memory based on the signal estimate on said estimate output line.

Claim 27

The minimum-mean-squared-error signal estimator of claim 22, wherein said updating means comprises: a second counter; a fourth memory; a fifth memory; an incrementer for incrementing the value stored at a given location in said fourth memory when said signal estimate has a predetermined amplitude; a scaler for scaling the value stored at a given location in said fourth memory in response to an output from said second counter, and for storing a plurality of most-significant bits of the scaled value back at said given location of said fourth memory to generate the amplitude-density function of said signal estimate, and for storing a plurality of least-significant bits of said scaled value at a corresponding location of said fifth memory.

Claim 28

The minimum-mean-squared-error signal estimator of claim 23, additionally comprising: a comparator coupled to said density function output line for comparing said density function with a threshold and producing a "detection achieved" signal when the value of said density function exceeds said threshold; means for coupling the output of said comparator to said second counter to control said second counter, so that said second counter is enabled only when said "detection achieved" signal is produced.