Method and system for noise measurement with combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal

2006200119 12 Jan 2006

Australia

P/00/011

Regulation 3.2

Patents Act 1990

COMPLETE SPECIFICATION

STANDARD PATENT

Invention Title:

Method and system for noise measurement with combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal

The following statement is a full description of this invention, including the best method of performing it known to us:

2006200119 28 Sep 2010

Method and system for noise measurement with combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal

The invention relates to a method and a system for noise measurement with combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noiss signal.

As illustrated in Figure 1, measured noise signals conventionally comprise superimposed sinusoidal interference signals. The sources of these sinusoidal interference signals are to be found either internally within the device or externally. Some of the frequencies 20 and interference lines are known in advance (network hum up to 10 harmonics, subsidiary lines of an internal frequency synthesiser, crosstalk from frequency dividers, microphone effects, for example, from fans) and some must still be determined.

These spectral lines associated with sinusoidal interference signals can be disturbing, for example, in the measurement of phase noise and must therefore be identified and removed as well as possible from the noise 30 measuring curve. However, in some applications, it is also important just to measure the frequencies and power levels of the sinusoidal interference signals as accurately as possible or to compare them with known frequencies and power levels of sinusoidal, reference

2006200119 12 Jan 2006 interference signals within the framework of a reference measurement.

In the context of a high-precision spectral analysis of a 5 measured signal, it is therefore desirable to identify the individual spectral lines associated with the sinusoidal interference signals from the other spectral components of the measured signal, to measure the frequencies and power level of the identified spectral 10 lines and, if required, for example, in the case of a noise measurement, to remove the identified spectral lines from the measured spectrum completely.

According to the prior art, graphic methods are used for 15 the identification of sinusoidal interference signals in a noise signal.

In one graphic method, as shown in Figure 2, a thresholdvalue line is placed over the noise curve. All of the 20 components of the spectrum disposed above this thresholdvalue line are recognised as spectral lines. The left and right intersection of the peak lines identified respectively by the threshold-value line with the peakfree noise curve is determined and linked by means of an 25 appropriate interpolation.

The identification of spectral lines associated with sinusoidal interference signals in a noise spectrum requires a high-resolution Fourier Transform. Spectral 30 lines disposed close together cannot be identified separately from one another using graphic means. The use of a constant threshold-value line is not appropriate in the case of a phase-noise curve, which provides a monotonously descending course. Consequently, a

6215839

2006200119 28 Sep 2010 threshold-value line must be used, which provides a course, which is constant only in very small regions, or a course which is complementary to the phase noise curve. In such a case, measuring the position of spectral lines is disadvantageously still only possible in a heuristic manner. A further disadvantage of the graphic method is the fact that, because of the interpolation of the noise curve, the precise information about the noise curve in the region of the identified spectral lines is lost. Furthermore, with numerous spectral lines in the noise curve, the graphic imprint of the spectrum curve is destroyed as a result of the numerous interpolations. Finally, the graphic method disadvantageously also does not allow a separation between the individual subroutines of measurement, identification, 15 nor does it therefore allow a selection of one or more of these subroutines dependent upon the respective requirements of the measuring task.

It would therefore be desirable to provide a method and a system for the identification and/or removal of sinusoidal 20 interference signals in a noise signal, which is based upon an efficient numerical method using a Fast Fourier Transform with a practicable frequency resolution and which allows a selection of one or more of the subroutines for noise measurement, identification and removal dependent upon the 25 measurement task.

Alternatively or additionally, it would be desirable to provide the public with a useful choice.

Reference to any prior art in the specification is not, and should not be taken as, an acknowledgment or any form of 30 suggestion that this prior art forms part of the common general knowledge in Australia or any other

6215839

2006200119 28 Sep 2010 jurisdiction or that this prior art could reasonably be expected to be ascertained, understood and regarded as relevant by a person skilled in the art.

As used herein, except where the context requires otherwise, the term comprise and variations of the term, such as comprising, comprises and comprised, are not intended to exclude further additives, components, integers or steps.

Summary of the Invention

In one aspect of the invention there is provide a method with subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal, which can be combined in a system for noise measurement, wherein:

the procedural stage of splitting of the frequency range to be measured into several frequency bands via an FFT filter bank and the procedural stage of determining autocorrelation matrices respectively associated with the frequency bands is implemented in a combined manner for the respectively selected subroutines of measurement, identification and removal of sinusoidal interference signals in a noise signal;

the parameters of the autocorrelation matrices a readjusted in a variable manner dependent upon the respectively selected subroutine(s) and upon the required quality of the result; and in the FFT filter bank the stages windowing, down mixing and under sampling are performed; and the FFT filter bank comprises several signal paths 30 each performing the stages windowing, down mixing and under sampling.

6215839

2006200119 28 Sep 2010

In another aspect of the invention there is provided a system for noise measurement with the combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal with an FFT filter bank for splitting a frequency range to be measured into several frequency bands, which is shared for each subroutine, and respectively a unit for the calculation of an autocorrelation matrix associated with the each frequency band, which is shared 10 for each subroutine and of which the parameters can be adjusted dependent upon the required subroutine and attainable quality of the result, wherein:

the FFT filter bank comprises units for windowing, units for downing mixing and units for under sampling;

and the FFT filter bank comprises several signal paths each comprising units for windowing, units for down mixing, and units for under sampling.

Also disclosed herein is a method with the subroutines, 20 which can be combined in a system for noise measurement, for the measurement, identification and removal of sinusoidal interference signals in a noise signal, as well as a system for noise measurement with the combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal.

With the subroutine for the identification of sinusoidal interference signals in a noise signal, the entire frequency range of the noise signal to be measured is subdivided via a Fast Fourier Transform filter bank into several frequency bands, in which the respective noise signal is approximately white and additionally contains

6215839

2006200119 28 Sep 2010 only a limited, comparatively small number of spectral lines associated with sinusoidal interference signals.

Accordingly, the conditions are fulfilled for a determination of the frequencies and power levels of the 5 sinusoidal interference signals, using a method for eigenvalue analysis of autocorrelation matrices, which are obtained from the measured signal comprising the noise signal and superimposed sinusoidal interference signals. In this context, the eigenvalues for the autocorrelation matrix obtained for each frequency band are analysed respectively into the eigenvalues associated with the noise components and the eigenvalues associated with the signal components. The eigenvalues associated with the noise components are the respectively lowest eigenvalues and provide a significant value-difference relative to the eigenvalues associated with the signal components. Using the eigenvectors associated with the noise eigenvalues, an estimation function is formed, of which the maxima represent the power levels for the sought sinusoidal interference signals associated with the respective frequencies .

By way of example, the estimation function based upon the MUSIC (Multiple Signal Classification) method can be used as the estimation function. However, other frequency25 estimation methods based upon the analysis of eigenvalues of autocorrelation matrices can also be used as an alternative. A Fast Fourier Transform filter bank is used to generate frequency bands with a frequency bandwidth, in which the noise signal can be assumed to be white.

On the basis of the windowing of the Fast Fourier Transform filter bank, which is realised, for example, via a Chebyshev filter, the spectral lines of sinusoidal

6215839

2006200119 28 Sep 2010 interference signals appear not only in one frequency band, but also in several adjacent frequency bands on the left and the right. These additional spectral lines are undesirable; they must be identified as such and must therefore be taken into consideration in the identification and removal of the spectral lines.

On the basis of the window function, the spectral lines of sinusoidal interference signals extend over several frequency bands. With a scaling of the frequency difference of each of these spectral lines genera ted from the window function in the respective frequency band to the mid-frequency of the respective frequency band with the bin width of the respective FFT used, it is comparatively simple, by deconvolution, to re-combine the power levels determined via the estimation function and associated with the individual interference spectral lines, which are distributed over several frequency bands .

If two or more sinusoidal interference signals have a frequency, which is disposed respectively in two or more adjacent frequency bands, and at the same time, have the same scaled frequency on the basis of their frequency position relative to the mid- frequency of the respective frequency band, and accordingly come to be disposed at the same frequencies over several adjacent frequency bands, the power levels of the spectral lines associated with the individual sinusoidal interference signals can be separated by means of a deconvolution. As an important condition for the use of the Welch method associated with the prior art, the individual FFTs of the Fast Fourier Transform filter bank are provided with a mutual time offset in the time domain.

6215839

2006200119 28 Sep 2010

With the subroutine for the removal of sinusoidal interference signals from a noise signal, in exactly the same manner as in the subroutine for the identification of sinusoidal interference signals in a noise signal, the 5 entire frequency range of the measured signal is divided into several frequency bands by means of a Fast Fourier Transform filter bank. The frequencies of the sinusoidal interference signals, which are determined either through the subroutine for the identification of sinusoidal interference signals in a noise signal described above or which are specified by the user, are assigned to the individual frequency bands. In this context, it is also taken into consideration that, spectral lines from a sinusoidal interference signal occur not only in a single 15 frequency band, but in several adjacent frequency bands as a result of the windowing of the FFT.

Starting from the number of spectral lines to be removed per frequency band, the noise power of the respective frequency band is determined from a given number of 20 lowest eigenvalues of an autocorrelation matrix associated with the respective frequency band. The number of lowest eigenvalues of the autocorrelation matrix associated with the respective frequency band, which are associated with the individual noise components of the measured signal comprising the noise signal and the sinusoidal interference signals, is determined, respectively via an analysis of all eigenvalues of the autocorrelation matrix associated with the respective frequency band, in that all of the eigenvalues associated with the respective noise components are minimal and provide a significant value difference relative to the signal components - the sinusoidal interference signals. With a specified number of interference-signal spectral lines to be removed, the number of eigenvalues associated

6215839

2006200119 28 Sep 2010

8a with the noise signal can, also be determined from the matrix dimension minus the specified number of interference-signal spectral lines to be removed. The spectrum of the noise signal after the removal of the 5 sinusoidal interference signals is derived from the combination of all of the noise spectra associated with the individual frequency bands.

Since the two subroutines for the identification and removal of sinusoidal interference signals in a noise 10 signal respectively contain the procedural stages of splitting the frequency range via an FFT filter bank into several frequency bands and determining the autocorrelation matrices associated with the respective frequency bands, and, as will be shown in greater detail 15 below, the subroutine for noise measurement can also be realised via these two procedural stages, these two procedural stages are implemented jointly for all three subroutines .

In this context, it must be taken into consideration that 20 the matrix dimension of the autocorrelation matrices is of a different size for each of the three subroutines. Accordingly, in a first embodiment of the method and the system for noise measurement with the combinable subroutines for the measurement, identification and 25 removal of sinusoidal interference signals in a noise signal, the respective maximum matrix dimension is used for each of the autocorrelation matrices. In this embodiment, the computational cost does, in fact, represent a maximum, however, evaluation data are 30 available for processing at all times for all three subroutines without interruption. In a second embodiment of the method and the system, in the normal operating mode, the autocorrelation matrix is calculated with the

6215839

2006200119 28 Sep 2010

8b minimum matrix dimension - the matrix dimension required for noise measurement at the level of the value 1 - and, in the case of an implementation of the subroutine for identification or removal, the respective autocorrelation matrix is re-initialised with the matrix dimension required for the respective subroutine for identification or removal. Accordingly, in the case of the second embodiment, each subroutine is implemented at its respectively maximum rate; however, in the subroutine for identification or removal, it is not possible to access the evaluation data from before the time of initialisation, because of the re-initialisation of the autocorrelation matrix.

Alongside the parameter of the matrix dimension for the 15 respective autocorrelation matrices, the parameter of the number of averagings can also be varied within the framework of a re-initialisation of the autocorrelation matrices. In the case of a large noise variance of the measured signal, sinusoidal interference signals, which are difficult to identify, can advantageously be identified in the noise spectrum with less ambiguity and greater security through an additional averaging up to the present time. The two embodiments of the method and system for noise measurement with the combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal are explained in greater detail below with reference to the drawings.

Brief Description of the Drawings

Figure 1 shows a spectral display of a measured spectrum of a noise signal with

6215839

Sep 2010

Figure 2 superimposed interference-signal spectral lines;

shows a spectral display of a noise spectrum with the interference-signal spectral lines removed on the basis of graphic methods;

<td> Figure 3</td><td> shows a block diagram of a Fast Fourier Transform filter bank;</td>

Figures 4A, 4B show a frequency spectrum of a Chebyshev

<td></td><td> window and a rectangular window;</td>

<td> Figure 5</td><td> shows a frequency-time diagram of an application according to the invention of a Fast Fourier Transform filter bank;</td>

<td> Figure 6</td><td> shows a schematic display of a FFT filter bank consisting of FFTs with a timeoverlap;</td>

<td> Figure 7</td><td> shows a spectral display of power-level curves;</td>

<td> Figure 8</td><td> shows a spectral display of the analysis range for interference-signal spectral lines at half-decade thresholds;</td>

<td> Figure 9</td><td> shows a flow chart for a subroutine according to the prior art for the measurement of a noise-power spectrum;</td>

<td> Figure 10</td><td> shows a flow chart for a subroutine according to the invention for the detection of sinusoidal interference signals in a noise signal;</td>

<td> Figure 11</td><td> shows a flow chart for a subroutine according to the invention for the removal of sinusoidal interference signals from a noise signal;</td>

<td> Figure 12</td><td> shows a flow chart for a first embodiment of the method according to the invention for noise measurement with the combinable subroutines according to the invention for the measurement, identification and removal of sinusoidal interference signals from a noise signal;</td>

<td> Figure 13</td><td> shows a flow chart for a second embodiment of the method according to the invention for noise measurement with the combinable subroutines according to the invention for the measurement, identification and removal of sinusoidal interference signals from a noise signal;</td>

<td> Figure 14</td><td> shows a block diagram of a system according to the invention for noise measurement with the combinable functions according to the invention for the measurement, identification and removal of sinusoidal interference signals in a noise signal;</td>

<td> Figure 15</td><td> shows a spectral display of a noise spectrum contaminated with spectral lines of sinusoidal interference signals; and</td>

<td> Figure 16</td><td> shows a spectral display of a noise spectrum with spectral lines of sinusoidal interference signals removed.</td>

Before describing the subroutine according to the invention for the detection of sinusoidal interference signals in a noise signal, the subroutine according to the invention for the removal of sinusoidal interference 5 signals from a noise signal and the method according to the invention for noise measurement with the combinable subroutines according to the invention for the measurement, identification and removal of sinusoidal interference signals in a noise signal in greater detail 10 with reference to Figures 9, 10, 11, 12, 13, and 14, the following section presents the theoretical background, on which the invention is based.

According to the invention, the entire frequency range of 15 the measured signal comprising the noise signal and sinusoidal interference signals is split into several frequency bands, in which the respective noise signal is approximately white and which each contain only a limited, comparatively small number of sinusoidal interference signals. The subdivision into several frequency bands takes place via a filter bank, which is realised in the form of a Fast Fourier Transform (FFT) according to Figure 3.

An FFT filter bank 1, which, according to the invention, generates a number of frequency bands corresponding to its FFT length N_FFT, accordingly comprises a total of N_FFTsignal paths, which, from a measured signal x(t), leads to a total of N_FFT time-discrete output signals x.(g-At), χ,(μΔΕ),..., Xim(g-At) . With a real measured signal x(t) and therefore with a symmetrical spectrum X(f), only the FFT in one sideband needs to be evaluated. In this case, the number of frequency bands to be observed is reduced to N_fft/2. The input signal x(t) is subjected in each of the N_FFT signal paths first to a windowing 2,, 2,,..., 2™ with the respectively associated impulse response h..^, (t), h.i^o.1 (t), . . ., h«inda««m (t) . The mίd-frequencies of the respective window transmission function H.^.(f),

,..., H.IndowMrrr (f) agree with the frequencies vf. of the individual FFT bins and also form the main frequencies of the individual frequency bands v. The bandwidth of each of the window transmission functions Η.^.(£), (f), . . ,, is therefore derived as follows - under the ideal condition of rectangular frequency spectra of the window transmission functions Hpiade«i (f), H«.^«m(f) - from the bin width f₀ of the FFT and also corresponds to the bandwidth Af of each of the frequency bands.

the windowing 2,, 2,, ..., 2am, each windowed signal is mixed in subsequent down mixings 3,,3,,..., 3_m, with the frequency f,, 2-fo, . . ., Nmf, into the respective frequency band in the baseband.

Finally, with the signals windowed and mixed down into the baseband, an undersampling 4.,4,,..,4^ takes place in the individual signal paths with the respective sampling ^rate v ¹ A/ ' ^which leads to the individual time-discrete FFT ' , output signals x. (μ-At), x, (μ-At) , . . ., x^(p At) at the respective outputs of the individual signal paths. Each of the time-discrete output signals χ.(μ·Δί:), x,(μ-At), . . ., Xmr^-At) is assigned to one of the total of N_FFT frequency bands and each provides an approximately white noise

2006200119 12 Jan 2006 signal and has a limited, comparatively small number of sinusoidal interference signals.

In reality, the window transmission functions (f),

H.uao.2 (f), . . ., do not provide rectangular frequency spectra, but, with a Chebyshev filter as shown in Figure 4A or with a rectangular filter as shown in Figure 4B, provide several subsidiary lobes disposed to the left and right of a main lobe. While a rectangular filter 10 provides a narrow main lobe and therefore most nearly fulfils the requirement for the realisation of a highfrequency resolution corresponding to the bin width f₀ of the FFT, one spectral line disadvantageously continues periodically in the adjacent subsidiary lobes 15 corresponding to the attenuation of the respective subsidiary lobe (so-called leakage effect). In the case of the Chebyshev filter, by contrast, the periodic continuation in the individual subsidiary lobes of spectral lines occurring in the main lobe can be 20 reduced in a targeted manner to a negligible value by attenuating the subsidiary lobes, while the bandwidth of the main lobe disadvantageously extends over several FFT bins and therefore over several frequency bands. The removal according to the invention of the disadvantageous 25 influence of the window function over several frequency bands is explained in greater detail below.

The output signal χ(μ · St · N_FFT) in the v-^th frequency band generated by the FFT filter bank is derived, in the case 30 of a time-discrete input signal χ(μ·Δϋ), according to equation (1) :

2006200119 12 Jan 2006

- j----νμ χ(μ Δί N_FFT) = Δΐ·Σ^χ(μΔί)·ΚμΔ()·β (1) μ

If the time-discrete input signal x(g-At) contains a sinusoidal interference signal with a frequency f_k, which is disposed exactly on the mid-frequency of the FFT bin then, with χ(μΔί) = and the relationship , ¹

Jo =——— &lt; an output signal χ(μ · Δί · N_FFT) of the FFT filter

IN bank is derived according to equation (2):

—j -vu χ(μ·Δί·Ν_ΕΡΤ) = Δί·^^2πν^ ·Η(μΔΐ)·_β = μ

= Δί·^^μΔί) = const.

μ (2)

If the frequency f_k of the sinusoidal interference signal is disposed centrally between two FFT bins &lt;/,=v/₀±A), then the output signal χ(μ·Δί·N_FFT) of the FFT filter bank in the v-^th frequency band provides a rotating phasor with maximum rotational velocity y- according to equation (3) x(p.^.N_FFT) = ^.Ye^j2n(V/^ ·ΚμΔί)·β^^μ = μ

= Δί ^Η(μΔί) e ² (3) μ

As a result of the windowing, spectral lines of sinusoidal interference signals, of which the frequency f_kcomes to be disposed in a given frequency band, are also detected in adjacent frequency bands. The amplitude of 25 these subsidiary spectral lines is derived from the attenuation of the window transmission function.

According to equation (4), the phases of the subsidiary spectral lines are constant, subject to the condition of a real and symmetrical (even) window function h*uao.(t) and therefore also of a real and even window transmission function H>i»a>.(f) relative to the phase of the main spectral line generated by the main lobe of the window transmission function:

&lt;(A ·&lt;5(/-Λ)*/7^(/)) = &lt;(Λ ’H_Window(f-f_k)) = &lt;A_k = const. (4)

By way of summary, the frequency-time diagram of the FFT filter bank in Figure 5 once again shows the connections between the time domain and the frequency domain with an FFT filter bank according to the invention, which, via several Fast Fourier transformers FFTi, FFT,, FFT,, FFT« and FFT», generates a total of N_m=8 frequency bands, which correspond respectively to the individual signal paths of the FFT filter bank in Figure 3, with a bandwidth at the level of the bin width f₀ of the FFTs.

The individual Fast Fourier transformers FFTi, FFT», FFT», FFT« and FFT. of the FFT filter bank need not be connected cyclically one after the other, but can also provide a time overlap as shown in Figure 6. This arrangement corresponds to the Welch method known from the prior art. The respective N_FFt sampling values at the outputs of the respective N_FFT signal paths of the FFT filter bank 1 are present according to Figure 3 at the positions

Xi (μ·Δ±·Ν™· (1-overlap)), x» (g-At N™· (1^overl^aP)) ' · · , x^» (μ·Δ±·Ν.»τ· (1—overlap)) . By comparison with the non-overlapping FFTs, there is therefore an oversampling by the factor -------.

- overlap

Additionally, the overlap in the case of a sinusoidal interference signal with a frequency fx, which provides a frequency difference Af« from the mid-frequency of the FFT bin (f_k = v-/θ+Δίο), brings about a rotating phasor, which provides a faster rotational velocity --—--.

overlap

The overlapping of the individual FFT window of the FFT filter bank 1 leads to a correlation between the values of the measured signal. As a result of the overlap, oversampled FFT results are obtained. The overlapping factor ov is obtained with an overlap according to equation (5):

ov =---------- / 5 j

- overlap

Only FFT results with a difference of ov or more are formed from non-overlapping FFT windows of the FFT filter bank 1. The oversampling is compensated by an undersampling by a factor ------- in the individual overlap

FFTs. In this manner, un-correlated noise sampling values, and therefore also a white noise signal necessary for the eigenvalue analysis of autocorrelation matrices, is provided.

If only an undersampling by the factor N_FFT is implemented in the individual undersamplings of the FFTs of the FFT filter bank, as shown in Figure 3, then a data sequence χ_ν(μ)= χ_ν(μ·Δϋ) is obtained at the output of the FFTs for the frequency band v as shown in equation (6), wherein w(g) models the noise, and the summated term models the total of p sinusoidal interference signals:

x_v(p) = w(p) + ^A_k-e^j^⁺^ (6) *=]

An undersampling by the factor Am(l - overlap) in the individual FFTs of the FFT filter bank 1 leads to a nonoversampled data sequence Χι(μον) according to equation (7) only for every ov-^th value:

x_v (μ ον) = w(p -ov) + ^A_k- _ej^^va>^ > (7) *=1

According to equation (8), the scaled angular frequency ^ωηοπηΛ ^{is the} non-oversampled angular frequency ox scaled to the bin width f₀ of the FFT filter bank of the spectral line associated with a sinusoidal interference signal and represents the frequency difference Af. of the spectral line at the frequency ν-f» of the nearest FFT bin, which corresponds to the mid-frequency ν-f» of the respective frequency band.

2π A (8)

Jo

The scaled angular frequency ω_ηοπηΛ therefore provides a value range [-π,+π] . At &>_ηοπηΛ=±π, the associated spectral line is disposed exactly at the right-hand or left-hand edge of the respective FFT bin and can also be found with the same strength in the adjacent frequency band.

According to equation (9) the ideal autocorrelation matrix R, which provides, for example, the dimension MxM, is obtained from a non-oversampled output signal χ_ν(μ·ον) at the output of the FFT associated with the frequency band v. Sampled values disposed at a past time of the non-oversampled output signal χ_ν(μ·ον) are used for this purpose:

R = £·{χ ·χ⁺} mit χ = [χ(μ),χ(μ-ον),...,χ(μ- Μ -ov)]^T (9)

[mit=with]

The autocorrelation matrix R is a square, positive and definite matrix, that is to say, its eigenvalues are real and positive. The eigenvectors, associated with the nonequal eigenvalues, are also orthogonal.

As a result of the stochastic of the non-oversampled output output of the FFT filter bank the autocorrelation matrix R_vis determined through multipli times :

- noise-laden - character signal χ»(μ·ον) at the

1, a reliable estimate for according to equation (10) averaging - a total of ι (Λί-Ο^ν+Λ^-Ι avg μ=(Λ/-1)·ον χ(μ) χ(μ-ον) χ(μ-(Λ/-1)·ον)

[**(M x’(A - ον),..., x* (μ - (Μ -1) · ον)] (10)

According to equation (11), the dimension M of the estimate (R_v) of the autocorrelation matrix associated with the frequency band (v) for the identification according to the invention of sinusoidal interference signals in a noise signal must correspond at least to the maximum expected number p_max of spectral lines per frequency band v with the addition of the value 2. The value 2 results from the fact that the MUSIC method requires at least two noise eigenvalues.

M^Pmax+2 (11)

In the special case, in which the matrix dimension M similarly adopts the value 1, starting from equation (10), the Welch method, known from the prior art for the calculation of the noise spectrum in the case of timeoverlapping FFTs, is derived according to equation (12) for the autocorrelation matrix (R_v) .

Σχ(Λ).χ,*(Λ)=5(ν /ο) (12) /Vflvt jt=o

According to the MUSIC (Multiple Signal Classification) method, the M eigenvalues associated with the estimated value R_v of the autocorrelation matrix and the associated eigenvectors v», ...,v» of the respective frequency band v are calculated after a total of N_avgaveraging stages according to equation (10).

The eigenvalues of the autocorrelation matrix can be subdivided into two groups. The first group of the lowest eigenvalues, of which the number M-p is equal to the dimension M of autocorrelation matrix R_v reduced by the number p of sinusoidal interference signals present in the frequency band v, are associated with the noise components of the measured signal x(n-ov) . The second group of the remaining eigenvalues, which are associated with the signal components - in the present case with the sinusoidal interference signals - and of which the number therefore corresponds to the number p sinusoidal interference signals, each provide a significant value difference relative to each of the respective lowest eigenvalues associated with the noise components.

No deterministic method exists for separating all eigenvalues ..,λ_Μ of the estimate R_v of the autocorrelation matrix associated with the frequency band v into the first and second group of eigenvalues and therefore for the determination of the number p of sinusoidal interference signals in the measured signal x(nov). According to the prior art, only a statistical analysis of the eigenvalues by means of histogram can be used in order to separate the eigenvalues .

In the case of an ideal autocorrelation matrix R, the lowest eigenvalues ,λ_Μ__ρ associated with the noise components are identical and, according to equation (13), equal to the noise power σ² .

σ² =/1,. for ie [1,..,M-p] (13)

In the case of a numerically-estimated autocorrelation matrix Λ,, the noise eigenvalues are distributed about the actual noise power σ² as a mean value. In this case, the noise power σ² is determined according to equation (14) :

<img file="AU2006200119B2_D0001.tif" />

(14)

The variance σ of the noise eigenvalues according to equation (15) declines with an increasing averaging duration N_mg .

^σ=^ΣΐΑ-Α|^! (15)

The scaled angular frequencies ω_ηοπηΛ of the spectral lines associated with the individual sinusoidal interference signals of a frequency band v are determined with the assistance of an estimation function, which is based upon the eigenvalue analysis of autocorrelation matrices.

These methods, which form the prior art, are described in detail in Manon H. Hayes: Statistical digital signal processing and modelling, John Wiley &amp; sons. Inc., 1996, page 451 to 469. The MUSIC (Multiple Signal Classification) method and respectively the Root MUSIC method are presented here briefly, without going into the details, only by way of example.

With the pure MUSIC method, on the basis of the eigenvectors v. associated with the noise components and an arbitrary column vector e((ik>™») of the signal correlation matrix R_s according to equation (16), an estimation function P_uu{e^J<“'^m*^ according to equation (17) is formed:

_ej(M~\)0)_normk (16)

Μ*'*' )=^r---^!----- (17) , ι=1

As shown in detail in Manon H. Hayes: Statistical digital signal processing and modelling, John Wiley &amp; sons. Inc., 1996, page 451 to 469, each of the eigenvalues y. associated with the individual noise components is orthogonal to an arbitrary column vector e((o«na) of the signal autocorrelation matrix R_s with a scaled angular frequency ω_ποπηΛ of a sinusoidal interference signal in the non-oversampled output signal Χι(η·ον) at the output of the FFT filter bank 1. In this case, the scalar product e(w_TOnnJk)·v, is zero, and accordingly, the estimation function is at a maximum. The scaled angular frequencies ω_ηοπηΛ at the respective p largest maxima of the estimation function therefore represent the scaled angular frequencies ω_ηοπη^ of the sinusoidal interference signals in the non-oversampled output signal x.(nov) at the output of the FFT filter bank 1. In this manner, with- the estimation function j, an estimation function for the identification of the scaled angular frequencies ω_ηοπηΛof the sinusoidal interference signals in the nonoversampled output signal x.(n-ov) is obtained at the output of the FFT filter bank 1.

The calculation of the estimation function (β^ίω’°·^Λ) via the angular frequency ω_ηοπηΛ, which represents a pseudo spectrum, can be advantageously implemented via an FFT.

In this context, the FFT length determines the frequency resolution of the calculated estimation function P_w ‘).

In the'case of the Root MUSIC Method, the Z-transform Vi(z) of the individual eigenvectors vi respectively associated with the noise components is determined. For this purpose, the individual components Vi(l) of the respective noise eigenvectors _Vi are subjected to a Ztransformation according to equation (18):

M-l ^(^ζ) = Σ^ν^)·^ζ~' for all i=i, . . ,_zM_p ₍₁₈₎

The estimates for the total of p scaled angular frequencies ω_ηοπιι> of the sinusoidal interference signals result from the angles of the p zero points of the polynomial D(z) calculated from the Z-transform Vi(_z) according to equation (19), which are disposed nearest on the unit circle of the complex z-plane:

Λ/ —p a//) ₍₁₉₎

The sealed angular frequencies ω__ can only be determined with an accuracy of modulus 2π, both with the pure MUSIC method via the FFT calculation of the pseudo spectrum and also with the Root-MUSIC method using the zero-point search on the unit circle of the complex z-plane, so that, in particular, the scaled angular frequencies ..v at the edges of the respective frequency band cannot be identified unambiguously. An unambiguous identification can only be achieved by combining the results determined in the individual frequency bands.

The power levels P™,·. of the respective p sinusoidal interference signals at the scaled angular frequencies co.™, of a frequency band v are obtained from the solution of the equation system (20), as is also shown in detail in Manon H. Hayes: Statistical digital signal processing and modelling, John Wiley &amp; sons. Inc., 1996, page 459 to 463, = for all i=l,...,p (20) *=i

The Z-transform of the eigenvectors y_t associated with the respective p interference signal components are obtained from equation (21) with reference to equation (18) for the M-p noise-component eigenvectors:

M-\

V_l(e^J‘“^t)='^_iv_l(l)e'^ja’^ar’*‘ for all i=l, ...,p (22) /=0

The eigenvectors λ»^ι, .. ., λ. associated with the p sinusoidal interference signals are the largest eigenvectors, arranged respectively in ascending order, of the estimated value Rv of the autocorrelation matrix for the frequency band v. σ. is the noise power predominating in the respective frequency band v. The vectorial presentation of equation system (20) is derived from equation system (22):

<img file="AU2006200119B2_D0002.tif" />

After the power level P«v.. and the scaled angular frequencies to»-» of the respective p (v) sinusoidal interference signals have been determined for every frequency band v, for example, using the MUSIC method, as presented above, the results of the individual frequency bands v must be combined to form a combined result. In this context, it must be taken into consideration that spectral lines of sinusoidal interference signals are obtained in adjacent frequency bands because of the windowing of the FFT filter bank 1.

The problem of frequency-band-wide superimposition is already known from the prior art in the context of the time overlap of several FFTs (overlapped FFT). The time overlap in this context serves to compensate the nonconstant window-transmission function with regard to an approximately constant evaluation of all spectral values in the entire frequency range. According to the prior art, an estimation of the frequency spectrum generated through an FFT filter bank 1 with overlapping in the time domain is realised by means of the Welch method.

As shown in detail in Manon H. Hayes: Statistical digital signal processing and modelling, John Wiley &amp; sons. Inc., 1996, page 415 to 420, the expected value

2006200119 12 Jan 2006 ^Epwckh(e^J )} of a frequency spectrum generated from several time-overlapped FFTs according to equation (23) is derived from the convolution of the frequency spectrum S(e^ja>) generated through an FFT filter bank 1 with the 5 modulus-squared window transmission function :

^£pwd=h (^)) = 5(^) *l//(_e^)l^{2 J} I I

The variance of a frequency spectrum of this kind generated via the Welch method tends toward zero «with an increasing averaging length.

The above Welch method is used according to the invention, to achieve an unambiguous identification of 15 main and subsidiary lines within one of the frequency bands v, which can be assigned to several sinusoidal interference signals with frequencies in different frequency bands. The individual frequency bands are overlapped in the time domain for this purpose as shown 20 in Figure 6.

The expected value for a frequency spectrum of an FFT filter bank 1 consisting of several spectral lines is derived starting from equation (23) using the Welch 25 method according to equation (24):

^£Ρ\νποκ(^^ω)} = Σ(^ρ* ·£(ω-ω^)*|//(β><^ω>)|² =^_{Pk (24)}

Since the frequency spectrum generated by the FFT filter 30 bank 1 and then modified using the Welch method is only calculated at the discrete frequencies f₀ of the FFT bins, the expected value of a frequency spectrum of an FFT ' filter bank 1 consisting of several spectral lines is obtained according to equation (25) :

(25) k

For the frequencies fk of the individual spectral lines, the relationship in equation (26) applies, in which the frequency f_k of the spectral line is classified in the FFT frequency grid by specifying the frequency offset Afo relative to the nearest FFT bin.

fk = V · /o +Af~=V · f₀ + (26)

2π

Accordingly, for every scaled angular frequency OU™., there is an individual expected value ^^,(^²^⁰) for the frequency spectrum at the discrete frequency fo of the FFT bin in the frequency band v.

Accordingly, the power level P_MUv, determined, for example, by means of the MUSIC method, of a spectral line at the scaled angular frequency co»™» in the frequency band v is derived according to equation (27) as the expected value 5_Wckhfor the frequency spectrum, which is calculated from the sum of all respective power levels Pk multiplicatively linked with the modulus-squared, window transmission function frequency-displaced by the frequency fk with approximately identical scaled angular frequency and accordingly identical frequency offset Af. relative to the respective FFT bin frequency:

2006200119 12 Jan 2006 (28) given (28), ^Pmuv = E{$Welch(e^w·)} = ΣΛ ·\H(e^J(v2n^-^ln^)|² (27} k

All spectral lines, of which the scaled angular frequencies provide an angular-frequency difference

Δω_ according to equation (28), which is smaller than a maximum angular-frequency difference Δω_, are counted as spectral lines in different frequency bands v with approximately identical scaled angular frequency (0..-., θ ^^normk &lt; ^^normMax

The power levels P^ at the individual FFT bin frequencies v · f₀, which can be calculated for each scaled angular frequency (o.-. according to equation 15 are obtained in sum for each given scaled angular frequency (O.^ in a power-level curve according to Figure 7.

The number p(v) of sinusoidal interference signals, the 20 scaled frequencies au- and the power levels P_Mr, of the respective spectral lines associated with the p(v) sinusoidal interference signals are derived using the MUSIC method and the subsequent Welch method for each frequency band v. Ambiguity continues to exist with 25 regard to the frequencies f_k and the power level P_k of those spectral lines, which contribute to the power-level curve with approximately identical scaled angular frequency (O.« in adjacent frequency bands v at the respective FFT bin frequency v·f .

For the determination presented below of the individual power levels P_k of those spectral lines, which contribute to the power-level curve with approximately identical scaled angular frequency Ok.™» in adjacent frequency bands v at the respective FFT bin frequency v · f₀, a weighted, scaled angular frequency w_normk according to equation (29) is introduced in the individual frequency bands v instead of the scaled angular frequency normk = &lt;Σ ^PMUv ’ ( 2 9 )

V

By taking into consideration the power levels P_MUv in the weighted, scaled angular frequency w_nonnk, the scaled angular frequencies (i)_normkv of spectral lines, which provide a higher power level P_MUv, are more heavily weighted. Through the use of the complex exponential function for the scaled angular frequency d)_narm,, in particular, at the frequency band edges (w_normiv = ±π), the value of the scaled angular frequency ω_{ηαηηΙίν} is preserved in the averaging.

Starting from equation (27), the relationship between the power levels P«v determined, for example, using the MUSIC method in the frequency band v at a given weighted, scaled angular frequency w_normk and the linear combination of the sought power levels P_k of spectral lines, which result from sinusoidal interference signals with angular frequencies io_k in adjacent frequency bands v+i and which are superimposed at the power level P»v of the power-level curve in the frequency band v, is described by the equation system (30). The angular frequencies (¾ of the main lines and subsidiary lines associated respectively with a sinusoidal interference signal, which are associated in sum with a common power-level curve, are disposed respectively in adjacent frequency bands v+i and all provide the identical frequency difference Δί. at the respective frequency band mid-frequency or respective FFT bin frequency (v+i)-f«. The power-level curves begin respectively in the frequency band n,..„ and extend over a total of N_lpfrequency bands. In the equation system (30), it is assumed that a main line of a sinusoidal interference signal could be disposed in each of the total of N_LP frequency bands.

-1)2π/„ +ω^ ·/„)( |//((Λί„-2)2π/₀+®„_ ./,)|^!

<td> |//(-(Λ/„ -2)2®/, +®_, ·Λ)|^!</td><td></td><td> ’ A ' p ⁿ&gt;tart⁺l</td><td> —</td><td> P P</td><td> +</td><td> e„</td>

<td> -/.)( .</td><td></td><td> P</td><td></td><td> P / MU.n^Nu.-X _</td><td></td><td></td>

(30)

The equation system (30) can be presented in an equivalent manner via the short form of equation (31).

HP = P_MU+e (31)

The error vector e = ^,e_n^,...,e_n^.J represents the error of the MUSIC algorithm in the respective powerlevel curve. If the error vector e is set to zero, then an unambiguous solution is obtained for the estimated vector P of the sought power level Ρ» according to equation (32):

£ = H-‘.P_MU (32)

Through the error vector e, error power-level values P* can occur, some of which can also be negative. In a case of this kind, by resolving the optimisation problem according to equation (33), which guarantees a nonnegative estimated vector P for the sought power-level values Pk, a better solution can be achieved than by resolving the equation system (32).

I~ II²

H P-P_mu|| under subsidiary condition p >0 (33)

The frequency f_k of the spectral line found, for example, by means of the MUSIC algorithm, is determined according to equation (34) from the number v.™ of the associated frequency band (counted from 1 to N_FFT) with the addition of the weighted, scaled angular frequency ω_ηοπη1ί .

Λ + A (34)

If only one sideband - in the case of a real measured signal x(t) - is evaluated, then the respectivelydetermined power-level value P* must still be multiplied by the factor 2.

After all of the main lines and subsidiary lines, associated with all of the sinusoidal interference signals present in the noise signal have been identified, either a frequency and power level comparison of the identified sinusoidal interference signals can be carried out with reference interference signals in the framework of an evaluation, or, as described below, a removal according to the invention of the identified spectral lines associated with sinusoidal interference signals can be implemented. Alternatively, spectral lines, of which the frequencies are known, for example, in the case of the disturbance of the noise signal by system hum, can also be removed without the identification according to the invention described above using the method according to the invention for the removal of sinusoidal interference signals from a noise signal.

The starting point is a list with frequencies f_k, at which sinusoidal interference signals occur. The removal of the spectral lines associated with the sinusoidal interference signals takes place, once again, by analogy with the detection of interference-signal spectral lines, in several frequency bands generated by an FFT filter bank 1. The list can contain positive frequencies f_kassociated with real sinusoidal interference signals or positive and negative frequencies f_k associated with complex interference-signal rotating phasors. For reasons of simplicity, the removal of real interference signals will be considered below.

After determining the number L_H of frequency bands or respectively FFT bins, which are covered by a window function |//(/)|² (corresponding, for example, to the frequency range, in which the window function |//(/)|²exceeds a given threshold value), the variables

P(v-1, . . ., N _FFT/2 ), which count the number of interferencesignal spectral lines per frequency band v, are initialised - when observing one sideband - with the value zero for every frequency band from 1 up to N_FFT!2.

For every frequency f_k, at which a spectral line can occur, a test of the Nyquist criterion is carried out as a selection criterion. If the frequency f_k does not satisfy the Nyquist criterion (condition in equation (35)), then the frequency f_k is discarded and no longer pursued.

A>/,/2=l/(At2) (35)

Following this, the number Vo_«„ of the FFT bin or the respective frequency band according to equation (36), in which the main line of the sinusoidal interference signal with the frequency f_k, is disposed, is determined:

^center = round · N_FFT +1 < J’ >

(36)

Starting from the determined number of the FFT bins or the respective frequency band, in which the main line of the sinusoidal interference signal with the frequency f_k falls, the counting variables p(v) of those frequency bands according to equation (38), which are disposed within the window function in the frequency range, are incremented.

p(v) = p(v) + l for all v = v - -¼ ~¹ ,.,.,ν + tenter 2 ⁷ center (37)

If the list with frequencies f_k, at which sinusoidal interference signals occur, contains an estimate for the difference of the power-level value P* of the spectral line of the noise level associated with the sinusoidal interference signal in addition to the frequencies f_k, the number L_H of the frequency bands or respectively FFT bins for those spectral lines, of which the power levels P* are disposed only slightly above the noise level, could be reduced in order to cut down the calculation time.

When considering both sidebands or one respective sideband, the estimate R_v of the autocorrelation matrix associated with the frequency band v is determined for each of the total of N_FFT or respectively N_FFT/2 frequency bands, starting from the time-discrete output signal χ_ν(μ·ον) of the FFT filter bank 1 associated with the respective frequency band v.

According to equation (38), for the removal according to the invention of sinusoidal interference signals in a noise signal, the dimension M(v) of the estimate R_v of the autocorrelation matrix associated with the frequency band v must correspond at least to the number p(v), identified above, of spectral lines per frequency band v with the addition of the value 1:

M(v)>p(v)+1 (38)

In this manner, the dimension M(v) of the estimate R_v of the autocorrelation matrix associated with a frequency band v for the method according to the invention for the removal of sinusoidal interference signals from a noise signal can generally be interpreted as smaller than the dimension M of the estimate R_v for the method according to the invention for the detection of sinusoidal interference signals in a noise signal according to equation (11).

Moreover, the averaging lengths N.„, in the case of the determination of the estimate R_v of the autocorrelation matrix associated with a frequency band v, can be designed to be smaller for the removal of spectral lines by comparison with the detection of spectral lines.

The determination of the estimate R_v of the autocorrelation matrix associated with the frequency band v for the method according to the invention for the removal of spectral lines corresponds to the procedure in the method according to the invention for the identification of spectral lines according to equation (10) .

Starting from the estimate R_v of the autocorrelation matrix associated with the respective frequency band v, the noise power σ.._ν associated with the frequency band v is then determined.

If no spectral line is present within the frequency band v - p(v)=0 -, then the noise-power spectrum S(v) is derived from the averaging of the spur elements of the estimate R_v of the autocorrelation matrix associated with a frequency band v according to the equation (39):

a 1 Μ (39)

This corresponds to the Welch method for spectral estimation known from the prior art.

With at least one spectral line per frequency band v p(y)>0 - the total of M eigenvalues Α^,...,Α_Μ of the Mdimensional estimate R_v of the autocorrelation matrix associated with a frequency band v is sorted via an analysis of eigenvalues of R_v and in ascending order of their value - λ_[<λ₁<...<λ_Μ . According to equation (40), the noise-power spectrum 5(v) is once again derived with reference to equation (14) from the averaging of the MP(v) lowest eigenvalues A, < < ... < A_M__p(v) of the estimate

Rv of the autocorrelation matrix associated with the frequency band v, which correspond to the noise eigenvalues of the matrix R_v :

^{ί(ν) = σ} ί Λ (40)

On the basis of the mathematical background described above, the subroutines for measurement of a noise-power spectrum, identification of sinusoidal interference signals in a noise signal and removal of sinusoidal interference signals from a noise signal and the method according to the invention for noise measurement with the combinable subroutines of measurement, identification and removal of sinusoidal interference signals in a noise signal are described below.

The subroutine for the measurement of a noise-power spectrum according to Figure 9 begins in procedural stage S10 with the splitting of a continuous or time-discrete measured signal x(t) or χ(μ·ΔΕ), which represents a noise signal w(t) or w^-At) with superimposed sinusoidal interference signals A_t e^J(a,t+9,) or A_t , by means of an FFT filter bank 1 according to equation (1) into a total of N_FFT measured signals, of which the frequency spectrum is bandpass-filtered respectively to a frequency band v. Each of these measured signals at the output of the FFT filter bank 1 is generated by one FFT of the FFT filter bank 1, wherein the respective bandpass-filtering of the measured signal is implemented via a windowing 2,,2,,...,2^ and subsequent down mixing 3,, 3,,...,3ott via. the FFT. The individual FFTs of the FFT filter bank 1 read in the respective measured signal at the input of the FFT filter bank 1 with a time offset and therefore operate with a time overlap with reference to the measured signal at the input. The oversampling of the individual FFT results caused because of this time overlap of the individual FFTs according to equation (7) is compensated by a complementary undersampling in subsequent undersamplings 4,,4,,...,4^,. In order to generate a total of N_FFT measured signals of a given frequency band v respectively bandpass-filtered with regard to their frequency spectrum at the output of the FFT filter bank 1, the bandwidth of the output signal of the individual FFTs is limited respectively to one FFT bin length f₀.

In the next procedural stage S20, starting from a total of N_fft measured signals, respectively bandpass-filtered with reference to their frequency spectrum to a given frequency band v, an estimate R_v of the autocorrelation matrix associated with each frequency band v is determined according to equation (10) . As a result of the stochastic characteristic of the noise signal, in this context, the unbiased quality of the estimate R_v of the respective autocorrelation matrix is increased by multiple averaging. According to equation (10), the maximum dimension M of the autocorrelation matrices for the calculation of the noise-power spectrum S(y · f₀) with superimposed spectral lines of sinusoidal interference signals associated with the respective frequency bands v is reduced according to the Welch method to the value 1.

The subroutine according to the invention for the identification of the interference-signal spectral lines in a noise-power spectrum according to Figure 10 provides m its two first procedural stages S110 and S120 identical procedural stages to the two procedural stages S10 and S20 of the subroutine for the measurement of a noise-power spectrum according to Figure 9. The only difference relative to the subroutine for the measurement of a noise-power spectrum is that the matrix dimension of the respective autocorrelation matrices R_v for the subroutine for the identification of interference-signal spectral lines is specified as at least the maximum expected number P_max of interference-signal spectral lines per frequency band v with the addition of the value 2.

The next procedural stage S130 of the subroutine according to the invention for the identification of interference-signal spectral lines comprises the determination of the eigenvalues λ., . . . ,λ. and the associated eigenvectors y., ...,v. of each of the total of Nfft estimates R_v of the autocorrelation matrix associated with each frequency band v, which is implemented according to known mathematical methods for eigenvalue analysis of matrices and the determination of corresponding eigenvectors based upon these methods.

In the next procedural stage S140, for every frequency band v and accordingly for the estimate R_v of the autocorrelation matrix associated with each frequency band v, a subdivision of all respectively determined eigenvalues λ>, ...,λ» into a first group of eigenvalues λ., ...,λ^_(ν> associated with the noise components and into a second group of eigenvalues λ»,· · · ,λ» associated with the sinusoidal interference signal components is implemented. With the number of noise eigenvalues λ., . . . ,λ»τ(_ν> and the number of interference-signal eigenvalues λ*,_(ν>·ι, . . .,λ«, the number p(v) of sinusoidal interference signals per frequency band v is determined from procedural stage S140.

In the next procedural stage S150, the noise power σ.._ν of each frequency band v is calculated according to equation (14) with the determined noise eigenvalues.

In the next procedural stage S160, the scaled angular frequencies ω»™» of the sinusoidal interference signals disposed in the respective frequency band v are determined by determining an estimation function P_vassociated with the respective frequency band v, which is based upon the eigenvalues and eigenvectors of the estimate R_v of the autocorrelation matrix associated with the respective frequency band v, and by evaluating this estimation function P_v. The MUSIC (Multiple Signal Classification) estimation function P-,_v presented in equation (17) can be used, by way of example, as the estimation function P_v. Alternatively, other estimation methods based upon the eigenvalue analysis of autocorrelation matrices can also be used. According to equation (17) , the MUSIC estimation function P«,._v provides maxima at those angular frequencies ω»»>, at which respectively an eigenvector Vi associated with a noise component is orthogonal to an arbitrary column vector e» of the signal autocorrelation matrix R_s and at which, accordingly, the scalar product in the denominator expression comprising respectively an eigenvector v. associated with a noise component and an arbitrary column vector e» of the signal autocorrelation matrix R_s is zero. The scaled angular frequencies co»™» of the sought sinusoidal interference signals are derived from the scaled angular frequencies ω— of the largest maxima of the estimation function P«,_v corresponding to the number of interference-signal eigenvalues . . . ,λ..

In procedural stage S160, the determination of the powerlevel values P«,_v.> associated with the individual sinusoidal interference signals for each frequency band v is additionally implemented by resolving the linear equation systems (20) . For this purpose, the respective noise power σ.,_ν, all of the interference-signal eigenvalues and the Z-transform Vfe'⁰’·”'·*) of the eigenvectors A-h>_iv,.i, . . . ,λ* obtained from the individual interference-signal eigenvalues ν^,νι.ι, . . . ,y. are required for each frequency band v, wherein the individual Ztransforms K.(e^7a,’°™*) calculated at the individual scaled angular frequencies ω»— of the sinusoidal interference signals are obtained through the estimation function P»,_v,».

In the next procedural stage S170, all of the spectral lines with their respective power-level values P-.v.», which were identified in the previous procedural stage S160 in the individual frequency bands v at an approximately identical, scaled angular frequency ω»™», are combined to form a combined power-level curve according to Figure 7. The difference Δω»™» between the two scaled angular frequencies, which, according to condition (28), must be smaller than a specified maximum angular-frequency difference Δω»»—, is used as the criterion for approximately-equal angular frequencies ω»™» of spectral lines. For every scaled angular frequency ω»™», the respective power-level curve is derived from the powerlevel values Pm»,_v.» localised at the individual FFT bin frequencies f₀, respectively assigned to the individual frequency bands v.

The spectral lines respectively identified at a given scaled angular frequency ω»™» in the individual frequency bands v in procedural stage S160 can be derived respectively from a superimposition of several spectral lines. This superimposition can result from main lines and subsidiary lines of one or more sinusoidal interference signals, of which the frequency is disposed

2006200119 12 Jan 2006 in a frequency band v, and at least one subsidiary line of at least one further sinusoidal interference signal, of which the frequencies are disposed in frequency bands v ±i adjacent to the frequency band v and which come to be disposed in the frequency band v as a result of the leakage effect.

In procedural stage S180, the power-level values A of the individual spectral lines, which result from sinusoidal interference signals with frequencies in different frequency bands v+i, are determined by resolving the linear equation systems (30) and (33) . The power-level values disposed in the preceding procedural stage S170 at the individual FFT bin frequencies vf. of the spectral lines of the respective sinusoidal interference signals, determined in procedural stage S160, for example, using the MUSIC method, and the modulus-squared window transmission functions |H(f)p frequency-displaced by the individual FFT bin frequencies v-fo of the FFT filter bank 1 are entered in the linear equation systems (30) and (33), applicable for a respectively scaled angular frequency Ignoring an error vector integrated in the equation systems (30) and (33), which models the process error achieved with the

MUSIC method, the individual power-level values A of the individual spectral lines are calculated by inversion of the matrix H formed with the individual windowtransmission functions |H(f)p and subsequent multiplication by the vector P.,_v from the power-level values determined by means of the MUSIC method in the individual frequency bands v according to equation (32).

Alternatively, the individual power-level values P_k of the individual spectral lines can also be determined by minimising the error vector e within the context of a minimisation method according to equation (33).

In the next procedural stage S190, the respective frequency f_k is determined for every spectral line according to equation (34).

In the final procedural stage S195, a list is prepared with all of the spectral lines at the frequencies f_k, which are to be removed.

In the following section, the subroutine according to the invention for the removal of sinusoidal interference signals from a noise signal according to Figure 11 is described:

In the first procedural stage S210, the number p (v) of interference—signal spectral lines to be removed is determined for every frequency band v, starting from the list of all sinusoidal interference signals identified in the frequency range as prepared in procedural stage S195 of the subroutine according to the invention for the identification of sinusoidal interference signals. In this context, all spectral lines, of which the frequencies according to equation (35) do not satisfy the Nyquist condition and which are not useful for further processing are discarded. The frequency band ν&lt;^«„, in which the main line of the sinusoidal interference signal comes to be disposed, is calculated according to equation (36), in order to determine the number p(v) of interference-signal spectral lines per frequency band v to be removed, starting from the frequency f_k of the identified sinusoidal interference signal. All frequency bands according to equation (37), in which respectively a main line or one of the subsidiary lines of the sinusoidal interference signal are disposed, are incremented on this basis.

In the next procedural stage S220, by analogy with procedural stage S10 in the subroutine according to the invention for the measurement of sinusoidal interference signals in a noise signal or respectively procedural stage S110 of the subroutine for the identification of sinusoidal interference signals in a noise signal, a total of N_fft measured signals, of which the frequency spectra are bandpass-filtered respectively with reference to one of the frequency bands v, is determined. The total of N_fft measured signals, of which the frequency spectra are bandpass-filtered respectively with regard to one of the frequency bands v, is determined according to equation (1) via an FFT filter bank 1.

In the next procedural stage S230 of the subroutine according to the invention for the removal of sinusoidal interference signals in a noise signal, by analogy with procedural stage S20 of the subroutine for the measurement of the noise-power spectrum or respectively procedural stage S120 of the subroutine according to the invention for the identification of sinusoidal interference signals in a noise signal, a respective estimate R_v of an autocorrelation matrix associated with each frequency band v is calculated starting from the total of N_fft measured signals, of which the frequency spectra are bandpass-filtered respectively with regard to one of the frequency bands v according to equation (10). In this context, the matrix dimension M(v) of the autocorrelation matrix Rv associatad with the frequency band v is, however, adjusted at least to the number p(v) of interference-signal spectral lines with the addition of the value 1. As a result of the stochastic characteristic of the noise signal, the unbiased quality of the estimate R_v of the respective autocorrelation matrix associated with each frequency band v is increased by multiple averaging.

The next procedural stage S240 contains the eigenvalue analysis, using a known mathematical method, of each of the autocorrelation matrices, which are assigned to each frequency band v.

With reference to the number p(v) of noise eigenvalues determined in procedural stage S210 for each frequency band v, in procedural stage S250, the respective noise power σ.,_ν is calculated for each frequency band v from the sum of the M(v)-p(v) smallest eigenvalues, wherein M(v) is the dimension of the estimate R_v of the autocorrelation matrix associated with the frequency band v.

In procedural stage S260, the total discrete noise-power spectrum is determined from the noise powers σ.,_ν associated with each of the individual frequency bands v, which were determined in the previous procedural stage S250.

In the following section, the first embodiment of the method according to the invention for noise measurement with the combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal is described with reference to Figure 12:

In the first embodiment of the method according to the invention for noise measurement with the combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal according to Figure 12, the two procedural stages of analysis of the measured signal x(t) or x(v-At) into a total of N_FFT measured signals, bandpass-filtered respectively to one of the N_FFT frequency bands and calculation of the autocorrelation matrix R_v respectively associated with a frequency band v are calculated only once in procedural stages S10 and S20 for all three subroutines for the measurement, identification and removal of interference-signal spectral lines in a noisepower spectrum. In this context, the matrix dimension M of the respective autocorrelation matrix R_v is specified according to equation (11) for all three subroutines together at the maximum required value of the maximum expected number P_max of interference-signal spectral lines per frequency band v with the addition of the value 2.

For the calculation of the noise-power spectrum S(v fj) in procedural stage S20 of the subroutine for the measurement of the noise-power spectrum with sinusoidal interferers, only the mean value of the elements R_v (k,k) (k=l, ..., M) of the respective autocorrelation matrix R_vis taken.

After the calculation of the noise-power spectrum with sinusoidal interferers in procedural stages S10 and S20, the identification of sinusoidal interferers in the noise-power spectrum is implemented in the remaining required procedural stages S130 to S195 of the first embodiment. (Procedural stages S110 and S120 of the subroutine according to the invention for the identification of sinusoidal interferers in a noise-power spectrum have already been processed in procedural stages S10 and S20 of the subroutine for the measurement of the noise-power spectrum with sinusoidal interferers).

In the case of a high noise variance of the measured signal x(t) or x(v-At) and accordingly a problematic identification of the sinusoidal interferers in the noise-power spectrum, in procedural stage S200, the number N_avg of the averagings of the estimate R_v of the repsective autocorrelation matrix to be implemented is increased within the framework of an initialisation of the respective autocorrelation matrices and accordingly, the stochastic noise component in the measured signal x(t) or x(v-At) is minimised by comparison with the deterministic interference-signal spectral lines, and the 25 procedure for measuring the noise-power spectrum is restarted in procedural stage S10.

Finally, after the identification of the sinusoidal interferers in the noise-power spectrum according to procedural stages S130 to S195, the removal of the identified sinusoidal interferers from the noise-power spectrum is implemented in the remaining required stages S210, S240 to S260 of the first embodiment. The procedural stages S220 and S230 of the subroutine according to the invention for the removal of sinusoidal interferers from a noise-power spectrum have already been processed in procedural stages S10 and S20 of the subroutine for the measurement of the noise-power spectrum with sinusoidal interferers.

After the removal of the sinusoidal interferers from the noise-power spectrum, the measurement of the noise-power spectrum is re-started cyclically in procedural stage S10.

The following section describes the second embodiment of the method according to the invention for noise measurement with the combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal with reference to Figure 13.

In the second embodiment of the method according to the invention for noise measurement with the combined subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal with reference to Figure 13, subroutine-specific matrix dimensions M for the autocorrelation matrices R_vassociated with the respective frequency bands v are used for each of the three subroutines of measurement, identification and removal.

In the first procedural stages S10 and S20, which are identical to those of the subroutine for the measurement of the noise-power density in figure 9, the autocorrelation matrices R_v each have the dimension M=1

2006200119 12 Jan 2006 and are therefore calculated in an efficient manner with regard to calculation time. In this manner, it is possible to calculate the noise-power density in realtime guasi-continuously; this is indicated by the feedback branch to the start of the flow diagram adjoining procedural stage S20.

If the user of the noise measurement system intends to identify the occuring interference-signal spectral lines, the autocorrelation matrices R_v associated with the respective frequency bands v are re-initialised with a dimension M, which corresponds to the maximum expected number P_raax of interference-signal spectral lines per frequency band vwith the addition of the value 2. The subroutine according to the invention for the identification of sinusoidal interferers in a noise-power spectrum is implemented with the autocorrelation matrices R_v initialised with the new matrix dimension M in procedural stages S110 to S195.

If the identification of the sinusoidal interferers in the noise-power spectrum with the maximum matrix dimension M=P_max+2 for the individual autocorrelation matrices proves difficult because of the high noise variance, in procedural stage S200, by analogy with the first embodiment in Figure 12, the user can achieve a stronger averaging out of the stochastic noise-signal components in the measured signal and therefore a more accurate identification of the interference-signal spectral lines by increasing the averaging number N_avgwithin the framework of an initialisation of the individual autocorrelation mactrices R_v . The subroutine for the identification of sinusoidal interferers in the noise-power spectrum is repeated in procedural stage S110 with the new averaging number N_avg.

If the subroutine for the identification of sinusoidal interferers in a noise-power spectrum is not repeated which is the normal case — the individual autocorrelation matrices R_v are initialised in procedural stage S30 with the matrix dimension M=1 for the measurement of the noise-power spectrum in procedural stages S10 and S20.

Following the implementation of the subroutine for the identification of sinusoidal interferers in a noise—power spectrum, if the user intends to remove the identified sinusoidal interferers from the noise-power spectrum, the autocorrelation matrix R_v associated with the respective frequency band v is initialised with the respective matrix dimension M(v), which corresponds to the number p(v) of sinusoidal interferers identified in the frequency band v with the addition of the value 1.

The subroutine for the removal of the sinusoidal interferers in a noise-power spectrum is then implemented with the re-initialised autocorrelation matrices R_v in procedural stages S210 to S260. This subroutine can be implemented by the user several times - which is indicated by the feedback branch from the end of procedural stage S260 to the beginning of procedural stage S210 - because the calculations are less calculation—intensive by comparison with the subroutine for the identification of sinusoidal interferers as a result of the matrix dimension.

If the user does not wish to repeat the removal of identified sinusoidal interferers, the individual autocorrelation matrices R_v are initialised in procedural stage S30 with the matrix dimension M=1 for the measurement of the noise-power spectrum in procedural stages S10 and S20.

The following section describes the system according to the invention for noise measurement with the combinable subroutines for the measurement, identification and removal of sinusoidal interference signals in a noise signal with reference to Figure 14.

The continuous or time-discrete measured signal x(t) or x(v-At) consists of a continuous or time-discrete noise signal w(t) or w(v-At) and several continuous or timediscrete sinusoidal interference signals or . This continuous or time-discrete measured signal x(t) or x(v-At) is processed in an FFT filter bank 1, of which the structure is presented in detail in Figure 3, in a total of N_FFT overlapping FFTs. The respective frequency spectra of the continuous and timediscrete measured signal x(t) or x(v-At) are bandpassfiltered with regard to a given frequency band v in the windowings 2,,2,, ..., 2_m assigned to the individual FFTs and in the subsequent down mixings 3,,3,,...,3^. The individual measured signals, of which the frequency spectrum is bandpass-filtered with regard to a frequency band v, are then restored in subsequent undersamplings 4i, 4,, . ., 4wrr, with regard to the oversampling caused by the overlapping, to the original sampling rate according to Figure 3.

The measured signals x(v ov-At), of which the respective frequency spectrum is bandpass-filtered with regard to a frequency band v, are supplied at the total of N_FFToutputs of the FFT filter bank 1, to the units 5,,5,,..,5for the estimation of the autocorrelation matrix R_v . The individual eigenvalues and eigenvectors of the estimate R_v of the autocorrelation matrix associated with each frequency band v are additionally determined in the respective units 5,,5,,..,5-^ for the estimation of the autocorrelation matrix R_v, from the estimate R_vassociated with each frequency band v of the autocorrelation matrix.

In a subsequent unit 6,, 6,, . . ., 6™ for determining the number p(v) of interference signals per frequency band v, a separation of all eigenvalues λ,,..., λ. determined at the estimated value R_v of the autocorrelation matrix associated with the frequency band v into noise eigenvalues λ,, . . . ,λ» and interference eigenvalues λ,, . . .,λ«,&lt;vi takes place. The number p(v) of interference signals per frequency band v corresponds to the number of determined interference-signal eigenvalues λ-,_(ν,.», . . ., λ. of the estimate Rv associated with the frequency band v of the autocorrelation matrix.

The subsequent units 7,,7,,..., determine respectively the noise power σ.,_ν of the frequency band v from the noise eigenvalues λ>, . . .,λ^_ινι, which corresponds respectively to a given number of smallest eigenvalues determined in the units 5i, 5a, . ., 5«rra for the estimation of the autocorrelation matrix R_v , wherein the given number represents the dimension M reduced by the number p(v) of interference signals per frequency band v of the estimate R_v associated with the frequency band v of the autocorrelation matrix.

The subsequent units 8i, 8a, ..., 8«m for frequency estimation determine, for example, via the MUSIC method using an estimation function by means of a maximum value observation, the scaled angular frequencies (iL— of the interference-signal spectral lines occurring in the respective frequency band v. In this context, the maxima occur in those cases, in which the eigenvectors Vi, ...,vp<v> associated with the respective noise eigenvalues λι, . . ., λ»„<ν> are orthogonal to an arbitrary column vector e. of the signal autocorrelation matrix R_s of the respective frequency band v.

Finally, the power-level values P»._v,_k associated with the individual interference-signal spectral lines in the individual frequency bands v are determined in subsequent units 91,94,...,9^ for power-level measurement, starting from the noise powers σ.,_ν, the interference-signal eigenvalues . . ., λ« and the Z-transforms ) of the interference-signal eigenvectors ...,v. at the individual determined, scaled angular frequencies ω— of the individual interference-signal spectral lines.

2006200119 12 Jan 2006,

In a subsequent unit 10 for the measurement of powerlevel curves per determined, scaled angular frequency ca»™», all of the power-level values P»._v.» of spectral lines, 5 which provide approximately identical scaled angular frequencies ω»™» in different frequency bands v, are combined at the individual FFT bin frequencies ν-f» of the FFT filter bank 1 to form a power-level curve.

The power-level values Pk of individual spectral lines, which result from different sinusoidal interference signals with frequencies f^ in different frequency bands v and which are superimposed at an identical scaled angular frequency ox.™» to form a single spectral line, are determined in a unit 11 for determining individual spectral lines from a superimposed spectral line. This superimposition of individual spectral lines, which is based upon the occurrence of subsidiary lines of sinusoidal interference signals because of leakage effects caused by windowing in respectively adjacent frequency bands v and the superposition, accurate with regard to frequency, of main lines of sinusoidal interference signals, is reversed by a deconvolution. For this purpose, a linear equation system with a matrix H comprising the modulus-squared window transmission functions |H(f-v f.) |> frequency-displaced by the respective FFT bin frequency vf., the vector P»._v with the power-level values P«._v.» determined at the respective scaled angular frequency ax.nJi in the individual adjacent frequency bands in the units 9.,9.,...,9^ for power-level measurement and the vector P of the sought power-level values Pt of the associated individual spectral lines, is resolved.

After a determination of the frequencies f_k associated with the power-level values A of the determined spectral lines in the unit 11 for determining individual spectral lines from a superimposed spectral line, the individual, identified spectral lines with their respective powerlevel values A and frequencies f_k are entered in respective lists in a unit 14. Alternatively, in unit 14, the power-level values P_k and frequencies f_k of spectral lines in a noise spectrum to be removed can be entered in the respective lists from an external source.

In order to remove the identified or specified spectral lines of sinusoidal interference signals, the eigenvalues λι, ...,λ- determined in the total of N_FFT estimation units

5i, 5₂, ..., 5_nfft of the autocorrelation matrices R_vassociated with the respective frequency bands v are supplied to the total of N_FFT units 7/, 7»', . . ., l_m' for the determination of the noise power σ.,_ν associated with the respective frequency band v. These N_FFT units

7i', Ί,', . . ., 7™' for the determination of the noise power o..v associated with the respective frequency band v are supplied by a unit 12 for determining the number of sinusoidal interference signals per frequency band v respectively with the number p(v) of sinusoidal interference signals per frequency band v. The unit 12 for determining the number of sinusoidal interference signals per frequency band v evaluates the list prepared in unit 14 with all of the identified or specified spectral lines of sinusoidal interference signals.

The individual units 7.^z, 7.^z, . . ., for determining the noise powers σ.,_ν associated with the individual frequency bands v are connected to a unit 13 for generating the noise-power spectrum S(v), in which the noise-power spectrum 5(v) of the entire frequency range to be measured is determined without the undesired spectral lines of the sinusoidal interference signals.

Finally, the unit 15 generates the noise-power spectrum extending over the entire frequency range including the spectral lines of sinusoidal interference signals from noise spectra including the spectral lines of sinusoidal interference signals determined in the units 5», 5,,..., 5™ for the estimation of the autocorrelation matrix R_v for the respective frequency band v. In order to determine the noise-power spectrum in the units 5., 5.,..., 5»» for the estimation of the autocorrelation matrix R_v , the dimension M of the respective autocorrelation matrix R_vis set to the value 1, so that the noise-power spectrum associated with the respective frequency band v including the spectral lines of sinusoidal interference signals is derived in the sense of equation (12), from the calculation of an autocorrelation matrix R_v of this kind.

Figure 15 presents a phase-noise spectrum with superimposed spectral lines of sinusoidal interference signals. The display also contains the frequencies at which spectral lines of sinusoidal interference signals

2006200119 12 Jan 2006 have been identified by the subroutine according to the invention for the identification of sinusoidal interference signals in a noise signal.

Figure 16 shows the identical phase-noise spectrum to that shown in Figure 15, from which spectral lines associated with sinusoidal interference signals have been removed by the subroutine according to the invention for the removal of sinusoidal interference signals in a noise 10 signal.

The determination of the estimated value R_v for the autocorrelation matrix associated with the frequency band v in the case of the method according to the invention 15 for identification is generally completed after a total of N_avg averagings, and is implemented continuously in the case of the method according to the invention for the removal of interference-signal spectral lines, wherein the removal of the spectral lines can begin after the 20 provision of a single estimate R_v of the autocorrelation matrix associated with the frequency band v, the accuracy of which can be improved with increasing averaging lengths N_aVg of the estimate R_v of the autocorrelation matrix associated with the frequency band v.

The invention is not restricted to the embodiments presented. In particular, instead of the MUSIC method, other frequency estimation methods, which are based upon an eigenvalue analysis of autocorrelation matrices, such 30 as the Pisareko method can also be used.