Least mean square (LMS) normalizer for active sonar

Claim 2

In an active sonar including a matched filter processor for processing received waveform echoes, an improved normalizer responsive to a sequence of matched filter output data to provide a normalizer output, comprising: means responsive to said matched filter output data for providing a logarithm of said output data; and normalizer processing means for processing said logarithm of said output data to dynamically adapt to locally non-stationary noise, said processing means comprising means for applying an adaptive Least Means Square (LMS) algorithm to said logarithm of said output data.

Claim 3

The normalizer of claim 1 wherein said processing means comprises a tapped delay line into which said logarithm of said sequence of matched filter output data is input, each tap including a corresponding tap weight.

Claim 4

The normalizer of claim 2 wherein said taps of said delay line include a signal window comprising one or more signal taps, and first and second noise tap windows each comprising one or more noise taps, said first and second noise windows straddling said signal tap window.

Claim 5

The normalizer of claim 3 wherein said means for applying said LMS algorithm comprises LMS weight update means responsive to an output of said normalizer for adaptively adjusting said tap weights of said first and second tap windows in accordance with said LMS algorithm, so that outputs of said taps comprising said first and second windows are weighted by tap weights which are adjusted in accordance with said LMS algorithm.

Claim 6

The normalizer of claim 4 wherein said processing means further comprises means for combining said adjusted tap outputs of said first noise window to form a first noise window signal, mean for combining said adjusted tap outputs of said second noise window to form a second noise window signal, and means for combining said first and second noise window signals to provide a combined adjusted noise signal.

Claim 7

The normalizer of claim 5 further comprising means for subtracting said combined noise signal from outputs from said one or more signal taps to provide said normalizer output.

Claim 8

The normalizer of claim 6 wherein no adaptation is applied to outputs of said signal taps.

Claim 9

The normalizer of claim 4 wherein said LMS weight update means further comprises means for adaptively adjusting said tap weights of said signal window taps in accordance with said LMS algorithm, to thereby adapt said normalizer to time-spreading of said signal.

Claim 10

The normalizer of claim 8 wherein said signal window comprises a plurality of signal taps, and said processing means further comprises means for combining said taps of said signal window.

Claim 11

The normalizer of claim 8 wherein the sum of said tap weights of said signal window taps equals one.

Claim 12

The normalizer of claim 8 wherein said means for applying said LMS algorithm includes a noise tap weight update coefficient applied to each noise tap output, and a signal tap weight update coefficient applied to each signal tap output, and wherein said noise coefficient is different from said signal coefficient.

Claim 13

The normalizer of claim 11 wherein said noise coefficient is a first time constant parameter which is set fast enough to account for noise variability, yet is relatively unresponsive to signals as they pass through said noise windows.

Claim 14

The normalizer of claim 12 wherein said signal coefficient is a second time constant parameter which is set shorter than said first time constant parameter in order to adjust to a narrow signal time spread as said signal passes through said signal window.

Claim 15

The normalizer of claim 1 wherein said sequence of matched filter data comprises a sequence of range bin outputs.

Claim 16

In an active sonar including a matched filter processor for processing received waveform echoes, an improved normalizer responsive to a sequence of matched filter output data to provide a normalizer output, comprising: means responsive to said matched filter output data for providing the logarithm of said output data; a normalizer means responsive to said logarithm of said matched filter output data to provide a normalizer output, said normalizer means comprising tapped delay line means having a plurality of tap outputs, wherein said first group of said tap outputs comprise a signal window, and second and third groups of said tap outputs on either side of said signal window taps comprise first and second noise windows, adaptive filter means responsive to said normalizer output for adapting respective tap weights applied to said tap outputs, means for respectively combining said signal window tap outputs, said first noise window tap outputs and said second noise window tap outputs to provide respective combined signal, first noise and second noise components, means for combining said first and second noise components to provide a composite noise component, and means for subtracting said composite noise component from said signal component to provide said normalizer output, wherein said normalizer adapts to locally non-stationary noise and time-spreading of said signal component.

Claim 17

In an active sonar including a matched filter processor for processing received waveform echoes, an improved normalizer responsive to a sequence of matched filter output data to provide a normalizer output, comprising: means responsive to said matched filter output data for providing the logarithm of said output data; a normalizer means responsive to said logarithm of said matched filter output data to provide a normalizer output, said normalizer means comprising tapped delay line means having a plurality of tap outputs, wherein said first group of said tap outputs comprise a signal window, and second and third groups of said tap outputs on either side of said signal window taps comprise first and second noise windows, adaptive filter means responsive to said normalizer output for adapting respective tap weights applied to said tap outputs, means for respectively combining said signal window tap outputs, said first noise window tap outputs and said second noise window tap outputs to provide respective combined signal, first noise and second noise components, means for combining said first and second noise components to provide a composite noise component, and means for subtracting said composite noise component from said signal component to provide said normalizer output, wherein said normalizer adapts to locally non-stationary noise and time-spreading of said signal component.

Claim 18

The normalizer of claim 15 wherein said adaptive filter means performs Least Means Square (LMS) filtering, and comprises LMS weight update means responsive to said normalizer output for adaptively adjusting said tap weights.

Claim 19

The normalizer of claim 15 wherein said adaptive filter means comprises Finite Impulse Response filter means.

Claim 20

The normalizer of claim 15 wherein the sum of said tap weights applied to said tap outputs of said signal window equals one.

Claim 21

The normalizer of claim 15 wherein said adaptive filter means further comprises means for applying a noise time constant coefficient to each of said tap outputs comprising said first and second noise windows, and means for applying a signal time constant coefficient to each of said tap outputs comprising said signal window, and wherein said noise time coefficient is a different value than the value of said signal coefficient.

Claim 22

The normalizer of claim 19 wherein noise time constant coefficient is set fast enough to account for noise variability, yet is relatively unresponsive to signals as they pass through said noise windows.

Claim 23

The normalizer of claim 19 wherein said signal time constant coefficient is set shorter than said noise time constant coefficient in order to adjust to a narrow signal time spread as said signal passes through said signal window.

Claim 24

An active sonar, comprising: a transmitter for generating a transmit waveform signal; a transmit transducer array for transmitting said transmit signal; a receiver transducer array; a beamformer responsive to outputs of said receiver array for producing a beamformer output; matched filter processing means responsive to a replica of said transmit waveform signal and to said beamformer output for producing a sequence of matched filter output signals; a normalizer for processing said matched filter output signals; threshold test means responsive to normalizer outputs for comparing said normalizer outputs to a threshold and declaring a detection when said threshold is exceeded; and wherein said normalizer comprises: means responsive to said matched filter output signals for providing the logarithm of said matched filter output signals; tapped delay line means having a plurality of tap outputs, wherein said first group of said tap outputs comprise a signal window, and second and third groups of said tap outputs on either side of said signal window taps comprise first and second noise windows, adaptive filter means responsive to said normalizer output for adapting respective tap weights applied to said tap outputs, means for respectively combining said signal window tap outputs, said first noise window tap outputs and said second noise window tap outputs to provide respective combined signal, first noise and second noise components, means for combining said first and second noise components to provide a composite noise component, and means for subtracting said composite noise component from said signal component to provide said normalizer output, wherein said normalizer adapts to locally non-stationary noise and time-spreading of said signal component.

Claim 25

An active sonar, comprising: a transmitter for generating a transmit waveform signal; a transmit transducer array for transmitting said transmit signal; a receiver transducer array; a beamformer responsive to outputs of said receiver array for producing a beamformer output; matched filter processing means responsive to a replica of said transmit waveform signal and to said beamformer output for producing a sequence of matched filter output signals; a normalizer for processing said matched filter output signals; threshold test means responsive to normalizer outputs for comparing said normalizer outputs to a threshold and declaring a detection when said threshold is exceeded; and wherein said normalizer comprises: means responsive to said matched filter output signals for providing the logarithm of said matched filter output signals; tapped delay line means having a plurality of tap outputs, wherein said first group of said tap outputs comprise a signal window, and second and third groups of said tap outputs on either side of said signal window taps comprise first and second noise windows, adaptive filter means responsive to said normalizer output for adapting respective tap weights applied to said tap outputs, means for respectively combining said signal window tap outputs, said first noise window tap outputs and said second noise window tap outputs to provide respective combined signal, first noise and second noise components, means for combining said first and second noise components to provide a composite noise component, and means for subtracting said composite noise component from said signal component to provide said normalizer output, wherein said normalizer adapts to locally non-stationary noise and time-spreading of said signal component.