WO2012028938A1

WO2012028938A1 - Signal processor, signal processing method and radar apparatus

Info

Publication number: WO2012028938A1
Application number: PCT/IB2011/002005
Authority: WO
Inventors: Masaru Ogawa; Makoto Ohkado; Isahiko Tanaka; Koji Suzuki; Atsushi Kawakubo
Original assignee: Toyota Jidosha Kabushiki Kaisha
Priority date: 2010-09-02
Filing date: 2011-09-01
Publication date: 2012-03-08
Also published as: JP2012052958A

Abstract

At least one of a predetermined upper limit value (dmax) and a predetermined lower limit value (dmin) is set, and at least one of a transformation that makes a data value equal to the upper limit value (dmax) if the data value is larger than the upper limit value (dmax) and a transformation that makes a data value equal to the lower limit value (dmin) if the data value is smaller than the lower limit value is performed on a data set of obtained signals so as to obtain a harmonic component in a signals. Then, a frequency analysis is executed on the transformed data set to detect a signal using a harmonic component in the result of the frequency analysis.

Description

SIGNAL PROCESSOR, SIGNAL PROCESSING METHOD AND RADAR

APPARATUS

BACKGROUND OF THE INVENTION

1. Field of the Invention

[0001] The invention relates to a signal processor, signal processing method and radar apparatus that processes an obtained data set. 2. Description of Related Art

[0002] FM-CW radars have been widely used as radars for vehicles including motor vehicles. FM-CW radars including a homodyne signal-receiving system, however, tend to suffer from the interference by noises near direct-current (DC) components at a fundamental wave mixer used in the signal-receiving system, resulting in an undesired noise factor of the signal -receiving system and a reduction of the target detection distance performance of the FM-CW radar.

[0003] To counter this, it has been proposed to provide modulating means for transmission signals or local oscillator signals and obtain video signals by offsetting either the frequency of the received signals input to the mixer used in the signal-receiving system or the frequency of the local oscillator signals input to the same mixer, by an amount corresponding to an intermediate frequency (refer to Japanese Patent Application Publication No. 11-109026 (JP-A-11-109026)). According to this technology, it is possible to reduce the interference by noises near DC components, achieving an improved noise factor of the signal-receiving system and thus an enhanced target detection distance performance of the FM-CW radar.

[0004] The technology described in Japanese Patent Application Publication No. 11-109026, however, has a drawback that an oscillator for generating an intermediate frequency F p is required. Further, when the frequency of the beat signals produced in a conventional FM-CW radar that does not use the intermediate frequency Fn? is F_b, the frequency of the beat signals produced in an FM-CW radar using the technology described above is Fi_F + Fb or FIF - F_b. In view of the fact that frequency folding occurs unless FIF - F > 0 and such frequency folding may lead to detection errors, the intermediate frequency FIF is required to be much higher than the beat signal frequency F_b (FIF » Fb). Due to this, a high beat signal frequency and a high A/D sampling frequency are required.

SUMMARY OF THE INVENTION

[0005] The first aspect of the invention relates to a signal processor that processes a data set of obtained signals. The signal processor includes a signal processing portion that: sets at least one of a predetermined upper limit value and a predetermined lower limit value; modifies the data set to obtain a harmonic component in the signals by performing, on the data set, at least one of a transformation that makes a data value equal to the upper limit value if the data value is larger than the upper limit value and a transformation that makes a data value equal to the lower limit value if the data value is smaller than the lower limit value; and executes a frequency analysis on the transformed data set and detects a signal included in the data set using the harmonic component in a result of the frequency analysis.

[0006] In the first aspect, the data set may be obtained based on a time interval.

[0007] In the above aspect, the data set may be obtained based on a constant interval.

[0008] In the above aspect, the frequency analysis may use a Fast Fourier Transform (FFT).

[0009] In the above aspect, the upper limit value and the lower limit value may be values expressible by straight lines or curves that are symmetrical with respect to a reference value for the obtained signals.

[0010] In the above aspect, the upper limit value and the lower limit value may be values expressible by straight lines and are set, based on at least one of a maximum value, a minimum value, and an average of obtained data, to a value smaller than the maximum value of the data and a value larger than the minimum value of the data, respectively.

[0011] The second aspect of the invention relates a signal processor that processes time-series data that is obtained by obtaining, at a predetermined time interval, beat signals produced by mixing transmission signals and received signals with each other. The signal processor includes a signal processing portion that: sets at least one of a predetermined upper limit value and a predetermined lower limit value; performs, on a data set in the time-series data, a transformation that makes a data value equal to the upper limit value if the data value is larger than the upper limit value when the upper limit value is set; performs, on the data set, a transformation that makes a data value equal to the lower limit value if the data value is smaller , than the lower limit value when the lower limit value is set; and executes a frequency analysis on the transformed data set and detects the beat signals using a harmonic component in a result of the frequency analysis.

[0012] In the second aspect, a time interval of the time-series data may be constant, and the frequency analysis uses a Fast Fourier Transform (FFT).

[0013] In the above aspect, a target exists may be determined based on a difference between the result of the frequency analysis on the transformed time-series data and a result of a frequency analysis on the time-series data that has not yet been transformed.

[0014] The third aspect of the invention relates to a radar apparatus including: an oscillator that generates transmission signals; a directional coupler that distributes the transmission signals in two directions; a transmission antenna that externally transmits the transmission signals distributed from a first port of the directional coupler; a receiving antenna that externally receives reflected signals; a mixer that outputs beat signals by mixing the transmission signals distributed from a second port of the directional coupler and the reflected signals received by the receiving antenna with each other; a low-pass filter that filters the beat signals to remove high-frequency components of the beat signals; an A/D converter that executes an analogue-digital conversion on the filtered signals; and a signal processor that detects an external target by executing signal processing on digital signals obtained from the analogue-digital conversion. The signal processor: sets at least one of a predetermined upper limit value and a predetermined lower limit value for values of the digital signals; makes the values of the digital signals equal to the upper limit value if the values of the digital signals are larger than the upper limit value when the upper limit value is set; makes the values of the digital signals equal to the lower limit value if the values of the digital signals are smaller than the lower limit value when the lower limit value is set; and detects a signal included in the digital signals using a harmonic component obtained by executing a frequency analysis on the transformed digital signals.

[0015] The fourth aspect of the invention relates to a signal processing method for processing time-series data obtained by obtaining, at a predetermined time interval, beat signals by mixing transmission signals and received signals with each other. The method includes: setting at least one of a predetermined upper limit value and a predetermined lower limit value; performing, on a data set in the time-series data, a transformation that makes a data value equal to the upper limit value if the data value is larger than the upper limit value when the upper limit value is set; performing, on the data set, a transformation that makes a data value equal to the lower limit value if the data value is smaller than the lower limit value when the lower limit value is set; and executing a frequency analysis on the transformed data set and detecting the beat signals using a harmonic component in a result of the frequency analysis.

[0016] According to the invention, the signals that are hidden in noises near direct components and thus normally undetectable can be detected through simple signal processing.

BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Features, advantages, and technical and industrial significance of exemplary embodiments of the invention will be described below with reference to the accompanying drawings, in which like numerals denote like elements, and wherein:

FIG 1 is a view showing the entire configuration of a radar according to an example embodiment of the invention;

FIG. 2 is a graph illustrating the signal-processor-received signals, beat signals, and DC noises at the radar of the example embodiment of the invention;

FIG 3 is a graph illustrating the transformed signal-processor-received signals in the example embodiment of the invention;

FIG 4 is a graph illustrating the result of a frequency analysis (beat signal: 70 Hz) in the example embodiment of the invention;

FIG 5 is a graph illustrating the result of a frequency analysis (beat signal: 50 Hz) in the example embodiment of the invention;

FIG. 6 is a graph illustrating the result of a frequency analysis (beat signal: 40 Hz) in the example embodiment of the invention; and

FIG. 7 is a graph illustrating transformation of the signal-processor-received signals in the example embodiment of the invention.

DETAILED DESCRIPTION OF EMBODIMENTS

[0018] Hereinafter, an example embodiment of the invention will be described with reference to the drawings.

[0019] FIG. 1 shows the entire configuration of a radar according to the example embodiment, which is structured as an FM-CW homodyne radar by way of example.

[0020] The signals output from an oscillator 10 are sent to a transmission antenna 14 via a directional coupler 12, and the transmission antenna 14 transmits (emits) electric waves toward a target.

[0021] The electric waves reflected from the target (reflected waves) are received by a receiving antenna 20. The received signals, that is, the signals received by the receiving antenna 20 are mixed, at a mixer 22, with the transmission signals sent from the directional coupler 12, and beat signals are produced from the differences between the received signals and the transmission signals. The high-frequency components in the beat signals are removed at a low-pass filter (LPF) 24, and then the beat signals are sent to an A/D converter 26. The A/D converter 26 converts the output signals of the LPF 24 into digital signals, and then sends the digital signals to a signal processor 28. Receiving the digital signals obtained through the conversion, the signal processor 28 executes signal processing on the digital signals to detect the target.

[0022] This radar is an FM-CW radar that repeatedly performs signal transmissions in an ascending phase where the transmission wave frequency is increased at a predetermined gradient with respect to the time axis and a descending phase in which the transmission wave frequency is lowered at a predetermined gradient with respect to the time axis, as shown in FIG. 1. The transmission and received signals exhibit frequency differences caused by the time lags therebetween, and undergo frequency changes corresponding to the relative speed of the target.

[0023] In such a radar, when the distance to the target is R, a frequency Ft, of the beat signals obtained from the signals reflected from the target is expressed by the equation (1) below. Note that the equation (1), for simplification, does not factor in the speed relative to the target.

F_b(R) = (AF / T) x 2R /c ... (1) Note that AF I T represents the frequency change with respect to time and c represents the electric wave propagation velocity.

[0024] Meanwhile, DC noises occur due to not only the thermal noises in the mixer 22, but also the transmission waves directly propagating to the receiving antenna 20 from the transmission antenna 14, the transmission waves propagating to the receiving antenna 20 from the transmission antenna 14 through reflections at the radome, the transmission signals reflected from the transmission antenna 14 and then entering the mixer 22 through the directional coupler 12, etc. The frequency of such DC noises can be expressed as F_b(R₀), using the equivalent distance R₀, and it is normally lower than F_b(R). In a case where the time-series data of the signal-processor-received signals (DSP-received signals) (beat signals + DC noises) that are produced through the sampling at the A/D converter 26 is obtained at a constant time interval, it is possible to execute a frequency analysis on the obtained data using the FFT. In this case, two peaks occur at the DC noise frequency Fb(R₀) and the beat signal frequency F_b(R), respectively. More specifically, the peak at the DC noise frequency F_b(R₀) always occurs regardless of whether a target exists, while the peak at the beat signal frequency F_b(R) occurs only when a target exists, and the distance to the target can be estimated from this frequency.

[0025] FIGs. 4 to 6 illustrate the results of three frequency analysises in which the DC noise frequency F_b(R_o) was commonly set to 10 Hz while the beat signal frequency F_b(R) was set to 70 Hz, 50 Hz, and 40 Hz, respectively.

[0026] When the distance to the target is extremely short, even if the frequencies Fb(R₀) and Fb(R) are different, the peaks at them may sometime overlap each other as indicated by "thl N/A" in FIGs. 4 to 6. In such a case, they can not be distinguished from each other in the result of the frequency analysis on the received signals. This is because each peak covers not only the peak frequency but also the frequencies close to it. Further, if a window function process is added to the FFT to lower the sidelobe level, the width of the peak (main beam) further increases. In such a case, the peak frequency of the beat signals obtained from the target can not be identified, and thus the target can not be detected. In this embodiment, "thl" stands for "threshold level".

[0027] FIG. 2 illustrates the time-series data of the DSP-received signals (beat signals + DC noises), normal beat signals, and DC noises in the case illustrated in FIG. 6, which is a typical example case. The signal processor 28 sets a predetermined upper limit value dmax and a predetermined lower limit value dmin for the DSP-received signals, and executes a transformation process in which the values of the DSP-received signals are made equal to the upper limit value dmax when the values of the DSP-received signals are larger than the upper limit value dmax, and the values of the DPS-received signals are made equal to the lower limit value dmin when the values of the DPS-received signals are smaller than the lower limit value dmin, by cutting off the upper or lower part of the data. FIG 3 illustrates the signals obtained from this transformation. Note that, in the example illustrated in FIG. 3, the upper limit value dmax=0.5 was set to the value obtained by multiplying the amplitude, at that time, of the DSP-received signals by 0.5, and the lower limit value dmin=-0.5 was set to the value obtained by multiplying the same amplitude by -0.5 (these set values will hereinafter be referred to as "thl 0.5" where necessary).

[0028] The result of a frequency analysis on the transformed signals shown in FIG 3 is indicated by "thl 0.5" in FIG 6. As is known from FIG 6, only one peak occurred when the above-described data cutting was not performed using the upper and lower limit values thl, while peaks occurred at about 90, 140, 200, and 270 Hz, respectively, when the data cutting was performed using thl 0.5. These frequencies are harmonics of 40Hz, the frequency of the beat signals obtained from the target. Further, using thl 0.2 lessened the DC noise interference, resulting in occurrence of harmonics of about 90, 135, 180, 235, and 285 Hz, as shown in FIG. 6. The cause of the frequency of each harmonic not being a precise integral multiple of 40 Hz is considered to be the DC noise interference.

[0029] In the case illustrated in FIG. 4, the DC noise interference was further lessened using thl 0.2, and the frequencies of harmonics of 70 Hz, which is the frequency of the beat signals obtained from the target, were about 140 and 215 Hz, that is, nearly double and treble the beat signal frequency. This result shows that it is possible to estimate the frequency of the original (untransformed) beat signals by detecting harmonics, although some errors are included. In short, the frequency of the beat signals obtained from the target can be relatively accurately identified by examining the differences between the frequencies of multiple harmonics, even if the frequencies of the harmonics shift.

[0030] FIGs. 4 to 6 each illustrate the results of frequency analyses using thl 1.0, thl 0.5, thl 0.2, and thl N/A, respectively. From these results, it is found that harmonic components can be added by forming corners at the waveform of the DSP-received signals by applying the upper and lower limits to the waveform, and the frequency of the beat signals (i.e., 70, 50, or 40 Hz) can be identified based on the positions of the harmonic components.

[0031] Further, whether any harmonic exists can be determined through the comparison between the result of the frequency analysis using the transformed time-series data and the result of the frequency analysis using the original (untransformed) time-series data. That is, unless the data cutting is performed using upper limit dmax and lower limit dmin, basically, the signals do not include the above-stated harmonics, which can be obtained through the data cutting. Thus, through the above comparison, the frequency of the original beat signals can be distinguished from the harmonics, and therefore the target can be more reliably detected using the harmonics.

[0032] Meanwhile, in a case where the target has a relative speed, when the speed is v (m/s), a frequency Fb(R, v) of the beat signals is expressed by the equation (2) below. F_b(R, v) = (AF / T) x 2R / c + 2vf / c ... (2)

In this equation, f represents the center frequency of the transmission waves, and c represents the electric wave propagation velocity. In this case, the DC noise frequency Fb(R₀) remains unchanged.

[0033] As such, although the beat signal frequency changes according to the relative speed of the target, how the same frequency is influenced by the relative speed of the target depends solely on whether the beat signal frequency is approaching the DC noise frequency F_b(R_o). This fact is similar to or the same as, in terms of equivalency, the fact that it is influenced by the distance R to the target being small or large.

[0034] In a case where the interval of the time-series data of the DSP-received signals obtained through the sampling at the A/D converter 26 is not constant for some reason, the FFT can not be used for the frequency analysis, but a typical discrete Fourier transform method or a high-resolution analysis method can be used alternatively.

[0035] While this example embodiment postulates the use of an FM-CW radar, the processes in the example embodiment may be implemented in the same manner as above also in a case where the signals from a target are to be isolated from ever-present low-frequency noises, such as DC noises, and detected in a CW radar or a pulse radar that observes Doppler frequencies. Further, it is to be noted that data sets obtained by changing the frequency interval or antenna position may be used in the same manner, in place of or in addition to the data sets obtained based on the time interval.

[0036] The average of the DSP-received signal data shown in FIG 2 is offset to the positive side due to the DC noises. While dmax = -dmin in the example embodiment, a better frequency analysis result can possibly be obtained in terms of the Signal-to-Noise Ratio (SNR) if dmax and dmin are set individually, although depending upon the offset amount. Therefore, preferably, dmax and dmin are set based on the obtained data values, especially, the maximum value, minimum value, and average, so as to achieve a desired SNR. For example, dmax is set to the value obtained by multiplying the maximum value of the DSP-received signals by 0.5 while dmin is set to the value obtained by multiplying the minimum value by 0.2, or dmax is set to the value obtained by multiplying the amplitude centered at the average by 0.5 while dmin is set to the value obtained by multiplying the same amplitude by -0.5. In order to cut off the crests and troughs of the waveform of the DSP-received signals, the constant of the multiplication is preferably within approximately 0.2 to 0.7 (less than 1). Further, only one of dmax and dmin may be set.

[0037] Note that the observation durations and signal frequencies in the foregoing example embodiment are no more than examples, and thus they vary for each radar.

[0038] According to the foregoing example embodiment, thus, harmonics are generated in the signals included in the data set by limiting the values of the data set using the predetermined maximum and minimum values, and the harmonics in the signals can be detected through a frequency analysis on the data set transformed through this limitation. In particular, true signal frequencies can be estimated from the frequency intervals of harmonics. [0039] Meanwhile, signals including beat signals extremely low in frequency are obtained from the signals received by a radar if the radar is an FM-CW radar and the distance to the target is extremely short, or if the radar is a CW radar and the relative speed is extremely low. In a frequency analysis on such received signals, the peak of the beat signals obtained from the target is hidden in the peak of DC noises that are extremely low in frequency. Even in such a case, however, harmonics of the low-frequency beat signals can be generated by executing the above-described transformation process on the received signals and then executing a frequency analysis on the transformed signals. As such, by detecting the harmonics, it is possible to estimate true frequencies and therefore to detect even the signals that are hidden in DC noises (low-frequency signals) and thus are normally undetectable.

[0040] While the upper limit value dmax and the lower limit value dmin are set to fixed values in the foregoing example embodiment, they are not necessarily set to fixed values.

[0041] FIGs. 7 A and 7B each illustrate an example in which the upper limit value dmax and the lower limit value dmin are variables, not constants. In the example illustrated in FIG. 7A, the upper limit value dmax and the lower limit value dmin are variables expressible by oblique lines. In the example illustrated in FIG 7B, they are variables expressible by sine curves of which the frequencies are different from the signal frequency (while the sine curve frequencies in this example are high, the sine curve frequencies may be low). The upper and lower limit values are thus set and values larger than the upper limit value are replaced with the upper limit value and values smaller than the lower limit value are replaced with the lower limit value, the same effect as that obtained with the constant upper and lower limit values stated above can be obtained. In such a case, preferably, the upper and lower limit values are set to values symmetrical with respect to a given reference value. Further, preferably, the upper and lower limit values are values symmetrical with respect to a given point of the line of the reference value.

Claims

CLAIMS:

1. A signal processor that processes a data set of obtained signals, comprising:

a signal processing portion that:

sets at least one of a predetermined upper limit value and a predetermined lower limit value;

modifies the data set to obtain a harmonic component in the signals by performing, on the data set, at least one of a transformation that makes a data value equal to the upper limit value if the data value is larger than the upper limit value and a transformation that makes a data value equal to the lower limit value if the data value is smaller than the lower limit value; and

executes a frequency analysis on the transformed data set and detects a signal included in the data set using the harmonic component in a result of the frequency analysis.

2. The signal processor according to claim 1, wherein the data set is obtained based on a time interval.

3. The signal processor according to claim 1 or 2, wherein the data set is obtained based on a constant interval.

4. The signal processor according to claim 3, wherein the frequency analysis uses a Fast Fourier Transform (FFT).

5. The signal processor according to any one of claims 1 to 4, wherein the upper limit value and the lower limit value are values expressible by straight lines or curves that are symmetrical with respect to a reference value for the obtained signals.

6. The signal processor according to claim 5, wherein the upper limit value and the lower limit value are values expressible by straight lines and are set, based on at least one of a maximum value, a minimum value, and an average of obtained data, to a value smaller than the maximum value of the data and a value larger than the minimum value of the data, respectively.

7. A signal processor that processes time-series data that is obtained by obtaining, at a predetermined time interval, beat signals produced by mixing transmission signals and received signals with each other, comprising:

a signal processing portion that:

performs, on a data set in the time-series data, a transformation that makes a data value equal to the upper limit value if the data value is larger than the upper limit value when the upper limit value is set;

performs, on the data set, a transformation that makes a data value equal to the lower limit value if the data value is smaller than the lower limit value when the lower limit value is set; and

executes a frequency analysis on the transformed data set and detects the beat signals using a harmonic component in a result of the frequency analysis.

8. The signal processor according to claim 7, wherein a time interval of the time-series data is constant, and the frequency analysis uses a Fast Fourier Transform (FFT).

9. The signal processor according to claim 7 or 8, wherein whether a target exists is determined based on a difference between the result of the frequency analysis on the transformed time-series data and a result of a frequency analysis on the time-series data that has not yet transformed.

10. The signal processor according to any one of claims 7 to 9, wherein the upper limit value and the lower limit value are values expressible by straight lines or curves that are symmetrical with respect to a reference value for the obtained signals.

11. A radar apparatus, comprising:

an oscillator that generates transmission signals;

a directional coupler that distributes the transmission signals in two directions; a transmission antenna that externally transmits the transmission signals distributed from a first port of the directional coupler;

a receiving antenna that externally receives reflected signals;

a mixer that outputs beat signals by mixing the transmission signals distributed from a second port of the directional coupler and the reflected signals received by the receiving antenna with each other;

a low-pass filter that filters the beat signals to remove high-frequency components of the beat signals;

an A/D converter that executes an analogue-digital conversion on the filtered signals; and

a signal processor that detects an external target by executing signal processing on digital signals obtained from the analogue-digital conversion, wherein

the signal processor:

sets at least one of a predetermined upper limit value and a predetermined lower limit value for values of the digital signals;

makes the values of the digital signals equal to the upper limit value if the values of the digital signals are larger than the upper limit value when the upper limit value is set; makes the values of the digital signals equal to the lower limit value if the values of the digital signals are smaller than the lower limit value when the lower limit value is set; and

detects a signal included in the digital signals using a harmonic component obtained by executing a frequency analysis on the transformed digital signals.

12. A signal processing method for processing time-series data obtained by obtaining, at a predetermined time interval, beat signals by mixing transmission signals and received signals with each other, the method comprising:

setting at least one of a predetermined upper limit value and a predetermined lower limit value;

performing, on a data set in the time-series data, a transformation that makes a data value equal to the upper limit value if the data value is larger than the upper limit value when the upper limit value is set;

performing, on the data set, a transformation that makes a data value equal to the lower limit value if the data value is smaller than the lower limit value when the lower limit value is set; and

executing a frequency analysis on the transformed data set and detecting the beat signals using a harmonic component in a result of the frequency analysis.

13. The signal processing method according to claim 12, wherein a time interval of the time-series data is constant, and the frequency analysis uses a Fast Fourier Transform (FFT).

14. The signal processing method according to claim 12 or 13, wherein whether a target exists is determined based on a difference between the result of the frequency analysis on the transformed time-series data and a result of an frequency analysis on the time-series data that has not yet been transformed.

15. The signal processing method according to any one of claims 12 to 14, wherein the upper limit value and the lower limit value are values expressible by straight lines or curves that are symmetrical with respect to a reference value for the obtained signals.