JP5057840B2 - Spectrum analyzer - Google Patents

Description

  The present invention relates to a spectrum analysis apparatus that obtains a feature quantity of a measurement target included in an input signal based on a spectrum analysis result of the input signal.

  In general, the target object (observation target) is measured by receiving a reflected signal obtained by oneself or another person radiating electromagnetic waves, laser waves, light waves, etc. into the space and reflecting them at the target, and processing the signal. There is a radar device that obtains information about a target by performing the above, or a sensor device that receives information on a target by receiving a radiation signal of the target and performing signal processing.

  As radar systems for measuring the distance from a radar device to a target, for example, pulse radar, pulse compression radar, FMCW (Frequency Modulated Continuous Wave) radar, multi-frequency CW (Continuous Wave) radar, FMCW radar and multi-frequency CW radar Various systems such as FMICW (Frequency Modulated Integrated Wave) radar and multi-frequency ICW (Interrupted Continuous Wave) radar have been proposed.

  Moreover, in the conventional target detection apparatus, in order to measure the target distance, a super resolution method such as MUSIC (Multiple Signal Classification) or ESPRIT (Estimation of Signal Parameters via Rotational Intelligence Techniques) is used (for example, a super resolution method). 1).

  When measuring the distance with MUSIC or ESPRIT, correlation processing is performed between the target reflected signals for each transmission frequency, and the number of target reflected signals is first estimated from the eigenvalues obtained as a result of eigenvalue analysis. Measure the target distance.

  As estimation methods, methods using AIC (Akaike's Information Criteria) and MDL (Minimum Description Length) have been proposed (for example, see Non-Patent Document 1). In general, a certain threshold value is set in advance, and the number of eigenvalues exceeding the threshold value is often used as an estimated value of the number of signals.

WO 2006/085352 A1 M.Wax and T.Kailth, “Detection of Signals by Information Theoretic Criteria”, IEEE Trans. Acoustic, Speech, and Signal Processing, ASSP-33, pp. 387-392, 1985.

However, the prior art has the following problems.
These conventional methods need to first estimate the number of targets before obtaining the target measurement value, and are effective when the S / N ratio (signal / noise ratio) is large. Otherwise, there is a possibility of misjudging the signal number estimation. In particular, when a smaller number than the actual number of signals has been estimated, there is a problem that the target measurement value has a large error in the subsequent processing.

  Furthermore, in the method of setting a threshold value in advance, subjective determination based on an empirical rule is required in determining this threshold value. Therefore, there is a possibility that the signal number estimation may be erroneously determined depending on the subjective determination, and there is a problem that the target measurement value has a large error in the subsequent processing.

  The present invention has been made to solve the above-described problems, and obtains a spectrum analyzer capable of measuring a target while suppressing an error factor, regardless of estimation of the number of signals or setting of a threshold value. For the purpose.

The spectrum analysis apparatus according to the present invention converts the analog signal obtained by mixing the transmission wave and the input signal observed in a predetermined period and frequency-converting it into a digital signal by performing a sampling process. An A / D conversion unit that generates a digital baseband signal for each, a correlation processing unit that performs correlation processing between the digital baseband signals generated by the A / D conversion unit for each predetermined period, and calculates a covariance matrix; The eigenvalue analysis of the covariance matrix calculated by the correlation processing unit is performed to calculate eigenvalues and eigenvectors, and the number of noise subspace matrices to be calculated is calculated based on the eigenvalue pair having the largest intensity ratio of adjacent eigenvalues. type (T is an integer of 2 or more) and the eigenvalue analysis unit that specifies, T type noise subspace from the calculated eigenvalues and eigenvectors eigenvalue analysis unit A subspace calculator for calculating a matrix, and evaluation spectrum calculating unit for calculating an evaluation spectrum for each noise subspace of T type calculated by the subspace calculator, T type from the evaluation spectrum calculated by the evaluation spectrum calculating unit A peak detection unit for obtaining a peak detection value for each noise subspace , a spectrum calculation unit for performing spectrum analysis on an input signal observed in a predetermined period, and a spectrum analysis result obtained by the spectrum calculation unit A statistical processing unit that obtains a characteristic amount of the measurement target included in the input signal by performing statistical processing on the peak detection values for each of the T types of noise subspaces , and the spectrum calculation unit performs sampling processing defining a range for spectral analysis from the sampling interval, and outputs a spectrum analysis of the range, the peak detector An assumed distance corresponding to the distance of the peak detection value obtained for each of the T types of noise subspaces is calculated, and the statistical processing unit performs statistical processing on the assumed distance calculated for each of the T types of noise subspaces by the peak detection unit. As a first step, a cluster gate having a predetermined width is set around each point of the assumed distance calculated for the first noise subspace of the T types, and as a second step, T If the assumed distance calculated for the second noise subspace of the type is within the cluster gate set in the first stage, the cluster gate center is reset to the average value of the two points in the cluster gate. If it is outside the cluster gate set in the first stage, a new cluster gate is set around the assumed distance point calculated for the second noise subspace. Our third noise From the assumed distance calculated for the subspace to the assumed distance calculated for the Tth noise subspace of the T types, cluster processing similar to the second stage is sequentially performed, and the fourth stage is performed. As a result of a series of processes from the first stage to the third stage, when there are a plurality of assumed distances included in each cluster gate, it is recognized as a target, and the gate center value is output as the target distance. As a stage, the number of signals as a feature amount is obtained from the number of target distances output in the fourth stage .

  According to the present invention, a threshold value is not set, the number of signals is not estimated from the result of eigenvalue analysis, a plurality of evaluation spectra are analyzed in a state where ambiguity is maintained, and the result is statistically processed. By doing so, it is possible to obtain a spectrum analyzing apparatus capable of measuring a target while suppressing an error factor without estimating the number of signals or setting a threshold value.

  Hereinafter, preferred embodiments of a spectrum analyzing apparatus of the present invention will be described with reference to the drawings.

Embodiment 1 FIG.
FIG. 1 is a configuration diagram of a spectrum analysis apparatus according to Embodiment 1 of the present invention. This spectrum analysis apparatus includes a signal generator 1, a divider 2, a pulse generator 3, a transmission antenna 4, a reception antenna 5, a mixer 6, an A / D conversion unit 7, a correlation processing unit 8, an eigenvalue analysis unit 9, a subspace. The calculation unit 10, the evaluation spectrum calculation unit 11, the peak detection unit 12, and the statistical processing unit 13 are configured.

  Here, the mixer 6, the A / D converter 7, the correlation processor 8, the eigenvalue analyzer 9, the subspace calculator 10, the evaluation spectrum calculator 11, and the peak detector 12 are captured from space by the reception antenna 5. This corresponds to a spectrum calculation unit that performs spectrum analysis using the received wave as an input signal.

  The signal generator 1 is a generator that generates a transmission signal having a plurality of transmission frequencies. The distributor 2 divides the transmission signal, and divides it into a pulse generator 3 and a mixer 6 for transmission. The pulse generator 3 pulsates one of the divided transmission signals to generate a transmission wave. The transmission antenna 4 radiates a transmission wave to space.

  The reception antenna 5 takes a received wave from space as an input signal. The mixer 6 mixes the transmission wave divided by the distributor 2 and the reception wave taken in from the reception antenna 5 and converts the frequency to generate a baseband signal. The A / D converter 7 converts the baseband signal from an analog signal to a digital signal at a predetermined sampling interval, and generates a digital baseband signal.

  The correlation processing unit 8 performs correlation processing between digital baseband signals received from transmission waves for each transmission frequency, and calculates a covariance matrix. The eigenvalue analysis unit 9 performs eigenvalue analysis of the covariance matrix to obtain eigenvalues and eigenvectors. The subspace calculation unit 10 calculates a plurality of noise subspace matrices from the eigenvalues and eigenvectors.

  The evaluation spectrum calculation unit 11 obtains an evaluation spectrum for each noise subspace. The peak detector 12 detects a peak from the evaluation spectrum obtained by the evaluation spectrum calculator. Further, the statistical processing unit 13 performs a statistical process on the obtained peak detection value to output a distance measurement value as a feature amount of the input signal.

  Although not shown in FIG. 1, a gain control circuit such as an amplifier or an AGC (Automatic Gain Control) circuit may be used as necessary.

  Next, a series of operations of the spectrum analyzer of the present invention will be described. The case where a multi-frequency ICW signal obtained by pulsing a continuous wave (multi-frequency CW) whose frequency increases stepwise will be described as a transmission signal. FIG. 2 is a diagram schematically showing a multi-frequency ICW transmission signal according to Embodiment 1 of the present invention.

Signal generator 1 has a frequency (the N 2 or more integer) stepwise N at regular intervals Δf type increases (refer to this period modulation period represents a period in T _P) multifrequency CW signal repeating the Is generated. If the duration of one frequency is T _PRI , the modulation period is T _P = N · T _PRI .

As shown in FIG. 2, the frequencies of the multi-frequency CW signal are f ₁ , f ₂ ,..., F _N. The frequency f _n (n = 1, 2,..., N) is expressed as f _n = f ₁ + (n−1) Δf. f ₁ is the minimum transmission wave frequency. The interval corresponding to the signal of a certain frequency f _n to be referred to as n-th step. The m-th modulation cycle (m is an integer equal to or greater than 1) is referred to as an m-th modulation process.

  Next, the multi-frequency CW signal is pulsed by the pulsator 3 via the distributor 2 and becomes a multi-frequency ICW signal. FIG. 2A schematically shows a transmitted pulse, and the vertical axis indicates frequency. The pulsed multi-frequency ICW signal is radiated into the air as a transmission wave via the transmission antenna 4.

The reflected wave from the target is received as a received wave via the reception antenna 5 and sent to the mixer 6. In the mixer 6, the reception wave from the reception antenna 5 and the multi-frequency CW signal distributed by the distributor 2 are mixed and subjected to frequency conversion, and then output to the A / D conversion unit 7 as a baseband signal. Baseband signal, by the A / D converter 7 is sampled at a sampling interval T _s, and is converted into a digital signal.

Here, the sampling interval T _s is less time pulse width. Further, the output signal of the A / D converter 7 is referred to as a digital baseband received _signal, represented by u p (m, n, k ). The digital baseband received signal u _p (m, n, k) is input to the correlation processing unit 8. Here, m represents the m-th modulation process, n represents the transmission frequency is f _n , and k is the range bin number. As shown in FIG. 2B, the range bin number is a sample point number when the received signal is sampled at the sampling interval T _s within the duration T _PRI of one frequency equal to the pulse repetition period.

FIG. 2B illustrates a case where the sampling interval T _s is equal to the pulse width and the number of range bins is K. The number of modulation periods required to output one set of target detection results is M (M is an integer equal to or greater than 1). The time corresponding to this is a detection data (hereinafter referred to as observation value) output cycle of the target distance which is an observation value. The multi-frequency ICW signal is continuously transmitted beyond the observation value output period.

In the configuration of FIG. 1, the digital baseband received signal u _p (m, n, k) is a real number, but the high-frequency signal received by the receiving antenna 5 is frequency-converted to obtain a digital in-phase (I) signal, quadrature ( Q) It may be configured to obtain a signal. In this case, the digital baseband received signal u _p (m, n, k) is handled as a complex digital signal having a real part as an in-phase signal and an imaginary part as a quadrature signal.

  Further, the transmission antenna 4 and the reception antenna 5 may be the same, and a circulator that separates the transmission path and the reception path may be provided. Further, a plurality of reception antennas 5 may be configured as array antennas.

Next, the operation after the correlation processing unit 8 will be described as a process for one observation value output period.
The correlation processing unit 8 performs correlation processing on the digital baseband received signal u (m, n, k) as shown in the following equation (1).

  The subscript P in the above formula (1) indicates the number of correlation average processes of the digital baseband received signal u (m, n, k), and 1 <P <N. The lower case p is an arbitrary integer from 1 to P. The subscript H represents conjugate transposition, and * represents conjugate. Furthermore, J is a matrix of (N−P + 1) rows (N−P + 1) columns.

Next, the eigenvalue analysis unit 9 performs eigenvalue analysis of the correlation value R (k) obtained by the above equation (1), and calculates an eigenvalue λ _q and an eigenvector e _q corresponding to the following equation (2).

The subscript q in the formula (2) represents the eigenvalue number, subscript q is a larger value the smaller eigenvalue _{λ q (λ 1 ≧ λ 2} ≧ ··· ≧ λ N-P + 1). Moreover, _eq is a Q column vector. FIG. 3 is a diagram schematically showing eigenvalues calculated by the eigenvalue analysis unit 9 according to Embodiment 1 of the present invention. The subspace calculation unit 10 calculates a noise subspace matrix E (t) from the eigenvalue λ _q and the eigenvector e _q obtained by the eigenvalue analysis unit 9 by the following equation (3).

  FIG. 4 is a diagram showing the relationship between the intensity ratios of adjacent eigenvalues with respect to the eigenvalues calculated by the eigenvalue analysis unit 9 according to Embodiment 1 of the present invention. As shown in FIG. 4, T in the above equation (3) is expressed as T = Q−qmax + 1, where qmax is an eigenvalue number corresponding to the denominator of the eigenvalue pair having the largest intensity ratio between adjacent eigenvalues. Therefore, the illustration according to FIG. 4 shows a case where Q = 7 and qmax = 3, and thus T = 5.

  Although it is possible to estimate the number of signals from the value of T, here, the number of signals is not estimated, but only T types of noise subspace matrices are calculated.

  Next, the evaluation spectrum calculation unit 11 calculates an evaluation spectrum P (r, t) of the following expression (4) for each noise subspace matrix E (t) obtained by the subspace calculation unit 10.

  Here, r is a variable representing distance, and c is the speed of light. The maximum distance of r is c / Δf / 2.

  The peak detection unit 12 takes the peak value of the evaluation spectrum P (r, t) and sets the distance corresponding to the distance r at this time as the assumed distance R (t, v). Here, v takes a value from 1 to the maximum (Qt), and represents the number of the peak value to be acquired in the evaluation spectrum P (r, t).

  FIG. 5 is a diagram illustrating a detection result by the peak detection unit 12 of the spectrum analysis apparatus according to Embodiment 1 of the present invention. More specifically, in the case of Q = 7 and T = 5, the result of peak detection by the peak detection unit 12 based on the evaluation spectrum obtained by the evaluation spectrum calculation unit 11 is represented.

  5A to 5F show the relationship between the evaluation spectrum and the assumed distance R (t, v), respectively, where (a) is t = 1, (b) is t = 2, ( c) shows the evaluation spectrum and assumed distance R (t, v) when t = 3, (d) shows t = 4, (e) shows t = 5 (= T), and (f) shows , (A) to (e) are shown superimposed.

  Next, the statistical processing unit 13 statistically processes the assumed distance R (t, v) obtained by the peak detection unit 12 to determine a target and obtain a distance R (w) to the determined target. Here, w takes any value from 1 to (Q−t), which is the target number.

  As a specific method, a method disclosed in Patent Document 2 (Japanese Patent Laid-Open No. 9-33628) can be used. The assumed distance R (t, v) is classified into a distance data group called a cluster, and a plurality of assumed distances R (( Let the average value of t, v) be the distance Rt (w). If there is only a single assumed distance R (t, v) in the cluster, it is determined as a spurious and is not regarded as a target.

  FIG. 6 is a diagram showing an outline of processing by the statistical processing unit 13 of the spectrum analyzing apparatus according to Embodiment 1 of the present invention, and is divided into seven stages (1) to (7). First, in (1), a cluster gate having a width of ± D is set around each assumed distance R (1, v) as a result of FIG. 5A. The assumed distance R (1, v) is represented by a black dot. Note that D is determined from the distance resolution desired to be obtained.

  Next, in (2), if the assumed distance R (2, v), which is the result of FIG. 5B, is within the cluster gate set in (1), the two points in the gate are averaged, Reset the cluster gate center to the average value. If it is outside the cluster gate, a new cluster gate is set around that point. The assumed distance R (2, v) is represented by a black point, and the assumed distance already shown in the previous flow is represented by a white point.

  Hereinafter, similarly, in (3), the assumed distance R (3, v) that is the result of FIG. 5C, and in (4), the assumed distance R (4, that is the result of FIG. 5D). As for v), in (5), cluster processing is performed for the assumed distance R (5, v), which is the result of FIG. 5 (e).

Next, in (6), if a plurality of assumed distances R (t, v) are included in each cluster gate as a result of the cluster processing, it is recognized as a target, and the gate center value is set as the target distance R (w). Output. If the assumed distance R (t, v) is single, it is determined as spurious and is not recognized as a target.
Finally, in (7), the number of signals is determined from the number of target distances R (w) output in (6).

  As described above, according to the first embodiment, a plurality of evaluation spectra are analyzed in a state where ambiguity is maintained, and the result is statistically processed to finally obtain a target distance that is a feature amount. ing. As a result, it is possible to measure a stable distance without causing a large error in the target distance by measuring the distance by erroneous signal number estimation. Furthermore, the number of signals (target number) can be determined from the result.

  Note that the feature amount obtained by statistically processing the spectrum analysis result is not limited to the distance to the measurement target, and the angle of the measurement target can also be obtained.

Embodiment 2. FIG.
FIG. 7 is a configuration diagram of the spectrum analyzing apparatus according to the second embodiment of the present invention. Compared to the configuration diagram of FIG. 1 in the first embodiment, the configuration of FIG. 7 in the second embodiment is different in that the distance gate calculation unit 14 is further provided. The distance gate calculation unit 14 corresponds to a part of the spectrum calculation unit.

  In the first embodiment, the statistical processing unit 13 performs the statistical processing for the entire measurement distance range regardless of the range bin number. However, the range bin received signal sampled by the A / D converter 7 is actually only the signal from the target included in the distance gate width represented by the following equation (5).

  Therefore, in the spectrum analysis apparatus according to the second embodiment, the distance gate calculation unit 14 calculates the distance gate width from the above equation (5), and the peak detection unit 12 uses the calculation result to peak only the necessary distance range. Perform detection.

  FIG. 8 is a diagram illustrating a detection result by the peak detection unit 12 of the spectrum analysis apparatus according to the second embodiment of the present invention. More specifically, the peak detection is performed by reflecting the result of the distance gate width calculated by the distance gate calculation unit 14 with respect to FIG. 5 describing the processing of the peak detection unit 12 in the first embodiment. The situation is illustrated.

  In this way, by defining the distance gate width, the number of assumed distances R (t, v) to be subjected to statistical processing is reduced, leading to a reduction in processing load on the statistical processing unit 13 at the subsequent stage.

  As described above, according to the second embodiment, by using the distance gate obtained from the range bin number, the target distance deviating from the distance gate is removed and the peak detection process is performed, thereby reducing the load on the statistical processing unit. In addition, the same effect as in the first embodiment can be obtained.

Embodiment 3 FIG.
FIG. 9 is a configuration diagram of the spectrum analyzing apparatus according to the third embodiment of the present invention. Compared to the configuration diagram of FIG. 1 in the first embodiment, the configuration of FIG. 9 in the third embodiment is different in that it further includes a speed separation processing unit 15. The speed separation processing unit 15 corresponds to a part of the spectrum calculation unit.

  In the previous embodiment 1 or 2, the number of signals that can be processed must be less than the rank Q of R (k) in the above equation (1).

  Therefore, when the target target is expected to have various speed components, if the distance is measured after separating the target for each speed, the number of signals of the same speed component is less than Q. Overall, more signals can be handled.

  The speed separation processing unit 15 performs Fourier transform on the digital baseband received signal u (m, n, k) calculated by the A / D conversion unit by the following equation (6), and the signal is divided into speed components. To separate. Here, f is a variable of the Doppler frequency indicating the speed.

Then, the velocity separation processing unit 15 obtains a peak value f _pk from the Fourier transformed signal x (f, n, k), and inputs the signal x (f _pk , n, k) at that time to the correlation processing unit 8. To do. Note that there may be a plurality of _fpk . At this time, the above equation (1) is transformed into the following equation (7).

  Subsequent processing is the same as in the first or second embodiment, and the spectrum analysis apparatus in the third embodiment can perform independent processing for each speed.

  As described above, according to the third embodiment, target targets can be separated for each speed, and independent processing can be performed for each speed, and the number of signals having various speed components is Q or more as a whole. However, if the number of signals of the same speed component can be reduced to less than Q by separating the target for each speed, the total number of signals can handle a target of Q or more, and the same effect as in the first embodiment. Is obtained.

  It is possible to combine the distance gate calculation unit described in the second embodiment with the configuration of the third embodiment, and the same effect as in the second embodiment can be obtained.

  In addition, it is possible to improve the estimation accuracy of the feature amount by performing spectrum analysis after correcting the phase amplitude error caused by hardware for the input signal.

  In addition, when an input signal is input, reception beam forming means that forms a reception beam by combining a plurality of reception antennas and reflected waves received by a plurality of reception antennas can be used.

  Furthermore, it is conceivable that the above-described spectrum analysis apparatus of the present invention is mounted on a mobile platform (car, ship, airplane) as an application example.

It is a block diagram of the spectrum analyzer by Embodiment 1 of this invention. It is the figure which represented typically the transmission signal of the multifrequency ICW system in Embodiment 1 of this invention. It is the figure which showed typically the eigenvalue calculated in the eigenvalue analysis part in Embodiment 1 of this invention. It is the figure which showed the relationship of the intensity ratio of an adjacent eigenvalue regarding the eigenvalue calculated in the eigenvalue analysis part in Embodiment 1 of this invention. It is the figure which showed the detection result by the peak detection part of the spectrum analyzer in Embodiment 1 of this invention. It is the figure which showed the process outline | summary by the statistical processing part of the spectrum analyzer in Embodiment 1 of this invention. It is a block diagram of the spectrum analyzer in Embodiment 2 of this invention. It is the figure which showed the detection result by the peak detection part of the spectrum analyzer in Embodiment 2 of this invention. It is a block diagram of the spectrum analyzer in Embodiment 3 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Signal generator, 2 divider | distributor, 3 pulse converter, 4 transmitting antenna, 5 receiving antenna, 6 mixer, 7 A / D conversion part, 8 correlation processing part, 9 eigenvalue analysis part, 10 subspace calculation part, 11 evaluation Spectrum calculation unit, 12 peak detection unit, 13 statistical processing unit, 14 distance gate calculation unit, 15 speed separation processing unit.

Claims (6)

The analog signal that has been frequency-converted by mixing the transmission wave and the input signal observed in a predetermined period is converted into a digital signal by sampling processing, and a digital baseband signal is generated for each predetermined period. An A / D converter to perform,
A correlation processing unit that performs correlation processing between the digital baseband signals for each predetermined period generated by the A / D conversion unit and calculates a covariance matrix;
Eigenvalue analysis of the covariance matrix calculated by the correlation processing unit is performed to calculate eigenvalues and eigenvectors, and the number of noise subspace matrices to be calculated is calculated based on an eigenvalue pair having the largest intensity ratio of adjacent eigenvalues. An eigenvalue analysis unit that identifies T types (T is an integer of 2 or more) ;
A subspace calculation unit that calculates the T types of noise subspace matrices from the eigenvalue and the eigenvector calculated by the eigenvalue analysis unit;
An evaluation spectrum calculation unit that calculates an evaluation spectrum for each of the T types of noise subspaces calculated by the subspace calculation unit;
A peak detection unit that obtains a peak detection value for each of the T types of noise subspaces from the evaluation spectrum calculated by the evaluation spectrum calculation unit, and a spectrum analysis is performed on the input signal observed in the predetermined period. A spectrum calculation unit for performing
A statistical processing unit that obtains a feature quantity of a measurement target included in the input signal by performing statistical processing on the peak detection value for each of the T types of noise subspaces obtained from the spectrum analysis result by the spectrum calculation unit And
The spectrum calculation unit defines a range for performing spectrum analysis from a sampling interval when performing sampling processing, and outputs a spectrum analysis result within the range ,
The peak detector calculates an assumed distance corresponding to the distance of the peak detection value obtained for each of the T types of noise subspaces,
The statistical processing unit, as the statistical processing for the assumed distance calculated for each of the T types of noise subspace in the peak detection unit,
As a first step, a cluster gate having a predetermined width is set around each point of the assumed distance calculated for the first noise subspace of the T types,
As a second stage, if the assumed distance calculated for the second noise subspace of the T types is within the cluster gate set in the first stage, two points in the cluster gate The cluster gate center is reset to the average value, and if it is outside the cluster gate set in the first stage, a new centered on the point of the assumed distance calculated for the second noise subspace Set the cluster gate,
As a third step, from the assumed distance calculated for the third noise subspace of the T types to the assumed distance calculated for the Tth noise subspace of the T types, Sequentially perform the same cluster processing as in the second stage,
As a fourth stage, when there are a plurality of assumed distances included in each cluster gate as a result of the series of processes from the first stage to the third stage, it is recognized as a target and the gate center value is set as the target distance. Output as
As a fifth stage, a spectrum analyzing apparatus characterized in that the number of signals as the feature amount is obtained from the number of target distances output in the fourth stage . The spectrum analysis apparatus according to claim 1,
The spectrum calculation unit performs a Fourier transform on the digital baseband signal for each predetermined period generated by the A / D conversion unit, thereby converting an input signal observed in the predetermined period to a predetermined frequency component Each further has a separation processing unit for separating each
The spectrum analysis apparatus, wherein the correlation processing unit performs a correlation process between digital baseband signals for each frequency component separated by the separation processing unit, and calculates a covariance matrix. The spectrum analyzer according to claim 1 or 2,
The spectrum calculation unit uses a MUSIC method or an ESPRIT method as a spectrum analysis method. In the spectrum analysis device according to any one of claims 1 to 3,
The spectrum analysis apparatus according to claim 1, wherein the input signal subjected to spectrum analysis by the spectrum calculation unit is a radiated wave from a measurement target existing in space or a reflected wave reflected by the measurement target. In the spectrum analysis device according to any one of claims 1 to 3,
The spectrum analysis apparatus according to claim 1, wherein the input signal subjected to spectrum analysis by the spectrum calculation unit is a reflected wave from a measurement target with respect to a transmission pulse signal radiated into space with a stepwise change in frequency. In the spectrum analysis device according to any one of claims 1 to 5,
The peak detection unit calculates an assumed angle corresponding to the angle of the peak detection value obtained for each of the T types of noise subspaces instead of the assumed distance,
Instead of obtaining the number of target distances to the measurement target existing in the space by the statistical processing for the assumed distance , the statistical processing unit performs the statistical processing for the assumed angle of the measurement target existing in the space. A spectrum analysis apparatus characterized in that the number of target angles is obtained, and the number of signals as the feature amount is obtained from the number of target angles .

Images