JP2005080118A - Method and apparatus for eliminating multipath noise - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and apparatus for eliminating multipath noise wherein a processing process for FM detection detects and eliminates a multipath noise signal so as to eliminate the noise at a high speed and downsize the circuit configuration thereby minimizing demerits from the standpoint of design / manufacture. <P>SOLUTION: A Tan<SP>-1</SP>value is calculated in units of sampling time of an IF signal to obtain a momentary angular velocity, an angular velocity being a phase change in units of sampling time is obtained, the angular velocity is compared with a prescribed threshold value, and the angular velocity in excess of the threshold value is corrected to a particular correction value. The multipath noise is eliminated by using a value of ((center frequency of IF signal ± maximum frequency shift of FM modulation) / sampling frequency) for the threshold value of noise detection, fixing the threshold value to the threshold value or below in units of the sampling time and employing moving average or straight line interpolation to implement noise elimination. Further, the FM detection is realized by merely eliminating an offset (center frequency of IF signal / sampling frequency) from the angular velocity. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to FM reception, and more particularly to improvement of a multipath noise removal method and removal apparatus.

  In recent years, FM broadcast receivers, particularly in-vehicle radios, have become very demanding to remove annoying pulse noise and multipath noise, or to eliminate the annoyance, and to improve sound quality.

  In an FM broadcast receiver, a signal output from a front end constituted by an antenna and a high-frequency amplifier circuit is often used as an intermediate frequency (generally 10.7 MHz in Japan). The signal is converted into a signal (IF signal) and then amplified, and FM demodulation processing is performed by an FM detection circuit. The IF signal amplified here has a constant amplitude, and the FM component included in the IF signal is a signal having a frequency fluctuation sufficiently small with respect to the IF. However, it is known that the level and frequency of the IF signal rapidly change when multipath noise occurs.

  FIG. 15 is a diagram showing an example of level fluctuation of the IF signal when multipath noise occurs, and FIG. 16 is an enlarged view of a part of the multipath noise occurrence portion in the output of the FM detection circuit. As can be seen from FIG. 15, when multipath noise occurs, the level of the IF signal is abruptly reduced, such as between T1 and T2, and the IF signal is lost.

  In the conventional FM detection circuit, a signal for multipath noise removal is generated from the signal after FM detection, and the multipath noise is removed from the FM detection output using the signal (see Patent Documents 1 to 3). .

JP-A-2-283129 JP 2001-36422 A JP-A-6-177786

  FIG. 13 is a diagram showing a multipath noise removing device (first prior art) described in Patent Document 2. In FIG. The first prior art includes a front end 12, an IF amplification means 13, an FM detection means 14, a multipath noise removal means 15, a noise removal means 16, a stereo demodulation means 17, an amplification means 18, and a speaker as a configuration of the FM receiver. 19, 20, SP means 21 and HC means 22.

  Here, the multipath noise removing means 15 includes an HPF 15a for detecting pulsed high frequency noise from the FM detection signal from the FM detection means 14, and an absolute value converting means (ABS) 15b for outputting a signal corresponding to the absolute value of the input signal. A comparator 15c that compares the output of the signal level from the ABS 15b with a threshold to determine multipath noise, a threshold generator 15d that generates a threshold to determine multipath noise, and a detection processing period of the FM detection signal for the multipath noise. The delay means 15e for delaying and the holding means 15f for holding the output of the delay means 15e by the signal generated by the comparison means 15c are configured. The noise removing means 16 performs a process for removing pulsed noise typified by ignition noise.

  In the first conventional technique, the HPF 15a passes a frequency component having high multipath noise from the FM detection signal and outputs a waveform corresponding to each spike noise. The ABS 15b converts this output into an absolute value to obtain a multipath noise detection signal, which is output to the comparison means 15c and the threshold value generation means 15d, respectively. The comparison unit 15c compares the multipath noise detection signal with the threshold value output by the threshold value generation unit 15d, and outputs a gate signal obtained by binarizing the multipath noise detection signal. The threshold generation means 15d generates the threshold based on the smoothed value of the input multipath detection signal for a certain period. The delay means 15e adjusts the timing by giving a delay time corresponding to the time required for outputting the gate signal by the HPF 15a, the ABS 15b, the comparison means 15c, and the threshold value generation means 15d to the FM detection signal from the FM detection means 14. Is to do.

  The holding unit 15f holds the value of the FM detection signal immediately before the occurrence of multipath noise by the delay of the FM detection signal by the delay unit 15e during the period of the gate signal output by the comparison unit 15c. That is, noise is removed by substantially replacing individual spike noises constituting the multipath noise with values holding the FM detection signal immediately before the occurrence of the multipath noise.

  In the first prior art, a multipath component (spike-like signal component) represented by, for example, T1-T2 in FIG. 16 is extracted from the FM detection signal, and a signal for removing multipath noise is created. The output of the multipath portion is suppressed from the FM detection signal delayed by the time of the signal generation processing, and the multipath is removed by outputting the signal immediately before the multipath is detected as an interpolation signal during that period. Is.

FIG. 14 is a diagram showing an impulsive noise removing device (second prior art) described in Patent Document 3. The second prior art includes an A / D conversion circuit 33, a TAN- ¹ type demodulation circuit 34, and a multipath noise removal means 35 for A / D converting the IF signal of the FM signal as a configuration of an impulsive noise noise removal device. Composed. The multipath noise removing unit 35 includes a shift register 35a as a delay circuit, subtractors 35c and 35e, threshold circuits 35d and 35f, an OR circuit 35g, a shift register 35h, an interpolation value calculation unit 35i, an interpolation interval determination unit 35j, and an interpolation circuit. 35b.

In the second prior art, the input IF signal is sampled at a predetermined sampling frequency to be a digital signal, FM demodulated by the TAN- ¹ type demodulation circuit 34, and the demodulated digital demodulated output value is delayed by the shift register 35a. Subtractors 35c and 35e generate subtraction values between the input value and the plurality of delayed output values, respectively. The threshold circuits 35d and 35f determine whether or not the output values of the subtractors 35c and 35e are equal to or greater than a predetermined value, and the interpolation section determination unit 35j interpolates based on the OR circuit 35g output of the outputs of the threshold circuits 35d and 35f. An interval is determined, and an interpolation value for the interpolation interval is generated by the interpolation value calculator 35i. Then, the interpolation circuit 35b performs noise removal by interpolating the digital demodulated output value including the noise portion with the interpolation value by linear interpolation. Note that the shift register 35h matches the timing at which the noise portion is detected with the interpolation timing.

  In the first prior art, the signal level of FM detection output is determined based on a predetermined threshold for detecting multipath noise. For example, it is realized by performing full-wave rectification processing on the FM detection output, passing through a low-pass filter, and performing level determination. In such a method, as shown in FIG. 16, when the influence of multipath periodically appears in the detection output, multipath noise may not be detected due to filter characteristics or the like, and the filter processing delay or the like after the interpolation processing may occur. In some cases, noise may be perceived in the sense of hearing from before the interpolation due to a detection timing shift due to. In addition, multiple detection circuits are required to detect and remove highly accurate multipath noise according to the detection conditions, such as the ability to follow very short pulse noises, which increases system size and complexity of control. There is also the problem of producing.

In the second prior art, a TAN ^-1 type demodulator is used. However, noise removal is not performed in the FM detection process as in the TAN ^-1 value calculation stage, and the voice band after the FM detection is also used. Noise is detected by comparing a subtracted value between an input value as an output and a plurality of delayed output values with a predetermined threshold value. This prior art performs processing for each sampling, but basically employs a configuration for detecting noise by logical sum of outputs of a plurality of subtractors, so that noise detection and removal processing is basically performed in sampling units. This is based on the principle of performing noise detection and removal processing in a predetermined range including noise.

Any conventional technique is a method for removing noise by generating a noise removal signal or the like from the FM demodulated output, and a delay means having a large delay time is required for synchronization between the FM demodulated output and the noise removal signal. There are also many demerits in design and manufacturing, such as complicated design such as processing delay time adjustment and increase in system scale.
The above problems are also caused by performing detection and removal of multipath noise on a signal in the voice band after FM detection.

(the purpose)
SUMMARY OF THE INVENTION A main object of the present invention is to provide a multipath noise removal method and a removal apparatus that detect and remove noise in units of sampling time by performing detection and removal of a multipath noise signal in the process of FM detection. There is.

  Another object of the present invention is to provide a multipath noise removal method and a removal apparatus that can realize detection and removal of a multipath noise signal with a small circuit configuration and minimize design and manufacturing disadvantages. There is.

  Another object of the present invention is to provide a multipath noise removal method and a removal apparatus that can detect and remove a multipath noise signal without a detection timing shift by simple control in units of sampling time.

The present invention is configured to detect and remove a multipath noise signal in the process of Tan ^-1 type detection, calculate a Tan ^-1 value for each IF signal sampling time, obtain an instantaneous angular velocity, An angular velocity that is a phase change is obtained, the angular velocity is compared with a predetermined threshold value, and an angular velocity that exceeds the threshold value is corrected to a specific correction value. For the noise detection threshold, (the center frequency of the IF signal ± the maximum frequency deviation of the FM modulation) / sampling frequency is used, and for noise removal, the sampling time unit is fixed to the threshold value or below the threshold value, Alternatively, multipath noise is removed by moving average or linear interpolation. FM detection is realized by simply removing the offset value (IF signal center frequency / sampling frequency) from the angular velocity. That means
The multipath noise removing method of the present invention includes a step of converting an IF signal into a complex number, a step of obtaining an instantaneous angular velocity by calculating a Tan ⁻¹ value in a system sampling time unit from the complexed IF signal, and a sampling time unit. The step of obtaining the angular velocity that is the phase change in sampling time unit from the instantaneous angular velocity of the sample, the step of comparing the angular velocity of the sampling time unit with a predetermined threshold, and correcting the angular velocity of the sampling time unit exceeding the threshold to a specific value And the threshold is (the center frequency of the IF signal ± the maximum frequency deviation of the FM modulation) / sampling frequency, and the step of correcting includes the step of correcting the angular velocity (the center frequency of the IF signal). + Maximum frequency deviation of FM modulation) / In case of value larger than sampling frequency The angular velocity is corrected to the first specific value not exceeding the (center frequency of IF signal + maximum frequency deviation of FM modulation) / sampling frequency, and the angular velocity is (center frequency of IF signal−maximum frequency deviation of FM modulation). Shift) / when the value is smaller than the sampling frequency, the angular velocity is corrected to the second specific value not lower than the (center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency. It is characterized by.

Multipath noise elimination device of the present invention, the I / Q separation means for complex the IF signal, the complex number of the Tan ^-1 value for calculating the instantaneous angular rate is Tan ^-1 values from the IF signal at a sampling time units of the system A calculating means; an angular velocity calculating means for obtaining an angular velocity that is a phase change in sampling time units from an instantaneous angular velocity in sampling time units; and a comparing means for comparing the angular velocity in sampling time units obtained by the angular velocity calculating means with a predetermined threshold value. Noise removing means for correcting the angular velocity of the sampling time unit exceeding the threshold value to a specific correction value, wherein the threshold value is (center frequency of IF signal ± maximum frequency deviation of FM modulation) / sampling The frequency of the noise removal means is such that the angular velocity is (the center frequency of the IF signal + the maximum frequency deviation of the FM modulation). When the value is greater than the sampling frequency, the angular velocity is corrected to the first specific value that does not exceed the (center frequency of the IF signal ± maximum frequency deviation of the FM modulation) / sampling frequency. (The center frequency of the signal−maximum frequency deviation of the FM modulation) / a value smaller than the sampling frequency, the angular velocity is less than (the center frequency of the IF signal−the maximum frequency deviation of the FM modulation) / the sampling frequency. It is characterized by correcting to a specific value.

  Further, when the angular velocity is a value larger than (the center frequency of the IF signal + the maximum frequency deviation of the FM modulation) / sampling frequency, the correction value is (the center frequency of the IF signal + the maximum frequency deviation of the FM modulation) / sampling. If the angular velocity is a value smaller than (center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency, then (center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency. It is characterized by being.

  Furthermore, in the above invention, the correction value is a value calculated by a moving average of angular velocities before the latest sampling or a value calculated by linear interpolation of angular velocities from the latest sampling before and after the detection of the multipath noise signal.

According to the present invention, an instantaneous angular velocity is obtained by calculating a Tan ⁻¹ value in a sampling time unit of the IF signal, an angular velocity that is a phase change amount in a sampling time unit is obtained, the angular velocity is compared with a predetermined threshold, Since the angular velocity exceeding the threshold value is corrected to a specific correction value, high-speed noise detection and removal processing in sampling units is possible.

In addition, by calculating the angular velocity in Tan- ¹ type FM detection, and performing multipath noise detection and noise removal processing at the processing stage, all processing circuits for multipath noise detection, removal, and FM detection are made Tan. ^-1 type detection circuit can be configured, and the FM demodulator can be downsized.

  Furthermore, the present invention simplifies the circuit configuration for delay adjustment with respect to the processing delay without causing a large processing delay, as compared with the conventional technique in which noise detection and removal is performed from the FM detection output or its processing output by processing in units of sampling time. It is possible.

In particular, the noise detection is realized by using the calculated angular velocity as a threshold value (the center frequency of the IF signal ± the maximum frequency deviation of the FM modulation) / sampling frequency, and the noise is removed by the sampling time unit. It is possible to eliminate multipath noise by fixing and correcting within the range or the threshold thereof, or correcting by moving average or linear interpolation.
Further, according to the present invention, FM detection can be easily realized by simply removing the offset value (center frequency of IF signal / sampling frequency) from the angular velocity corresponding to the change in instantaneous angular velocity.

(Embodiment 1)
FIG. 1 is a diagram showing a first embodiment of a multipath noise removal method and removal apparatus in FM reception according to the present invention.
The FM receiver comprises a front end 2, IF amplification means 3, FM detection means 4, stereo demodulation means 7, amplification means 8, speakers 9 and 10, SP means 11, and HC means 12.

The FM detection means 4 includes an I / Q separation means 4a for complexizing the IF signal, a Tan- ¹ value calculation means 4b for performing processing by digital signal processing, and an output (Tan- ¹ value) of the Tan- ¹ value calculation means 4b. Angular velocity calculating means 4c for calculating an angular velocity corresponding to the phase change by delay comparison, and detecting noise when detecting the multipath noise signal compared to the first and second specific threshold values in order to detect the multipath noise signal from the angular velocity. A comparison means 4d for outputting a signal; a threshold value means 4e for generating and outputting the threshold; a noise removal means 4f for correcting the angular velocity by a noise detection signal output by the comparison means 4d and removing noise; and an output from the noise removal means 4f. The offset removing unit 4g is configured to remove the offset value and output the FM demodulated signal. As can be seen from the above configuration, the multipath noise detection process is executed in the FM detection means 4.

In the present embodiment, a Tan- ¹ type detection method performed by digital signal processing is used for FM detection. In Tan- ¹ type detection, phase angle information (hereinafter referred to as “instantaneous angular velocity”) in sampling time units is obtained, and a difference (angular velocity) between instantaneous angular velocities before and after the sampling time unit is calculated (hereinafter referred to as “delay comparison”) (Nishimura). Yoshiichi “Know-how on DSP processing” CQ publisher ISBN 4-7898-3352-6, Chapter 5 5-1-2). Next, the center frequency component as an offset value is removed by subtracting the IF center frequency / sampling frequency from the calculated angular velocity to output an FM detection signal. The FM detection is completed by processing the control signal combined with the FM detection signal and the combined signal of the stereo signal as necessary.

FIG. 2 is a diagram illustrating the principle of FM detection (demodulation) by the Tan- ¹ type detection method. The principle of Tan ^-1 type detection is that the IF signal is converted into two signal values of the orthogonal coordinate system of the real axis and the imaginary axis by using a Hilbert filter or the like, and the function of Tan ^-1 is used to obtain phase angle information. Thus, the instantaneous angular velocity is obtained, and since the differential of the phase is the frequency, the phase change (angular velocity) between the samples is calculated, and this angular velocity is output as a signal component subjected to FM detection. Specifically, the following processing is performed in the I / Q separation unit 4a to the offset removal unit 4g.

  The I / Q separation means 4a receives the IF signal (sin (x)) 3a and converts it to a digital signal IF signal sin (x) at a predetermined sampling frequency. The IF signal sin (x) and a Hilbert filter or the like The IF signal cos (x) that has passed through the 90-degree phase shifter is generated and output to perform I / Q separation, and the IF signal is converted into a complex number.

The Tan ⁻¹ value calculating means 4b receives complex-valued sampling time unit signals sin (x) and cos (x), calculates sin (x) / cos (x) in sampling time units, and Tan ^-1 Determine the value of (x). Next, referring to a table storing the relationship between Tan ⁻¹ (x) and (x), each sample of Tan ⁻¹ (x) is converted to (x) to obtain an instantaneous angular velocity.

The angular velocity calculation means 4c compares the instantaneous angular velocity at the current sampling time (time point) with the instantaneous angular velocity at the previous sampling time point to obtain a change (angular velocity) of each sample (x).
The output of the angular velocity calculation means 4c is generated through a noise removal means 4f, which will be described later, and then an FM demodulation signal is generated by removing the offset value (for the IF center frequency) generated at each sampling time in the offset removal means 4g. Output.

  Next, multipath noise detection and noise removal related to the FM detection signal by the comparison means 4d, threshold value means 4e, and noise removal means 4f shown in FIG. 1 will be described.

In Tan- ¹ type detection, processing is performed by a digital signal obtained by sampling the IF signal. The instantaneous angular velocity obtained as a Tan ⁻¹ value from the digital signal of each sampling time unit is obtained when the maximum frequency deviation of the center frequency (IF frequency) of the carrier wave is ± 75 kHz.
The value is in the range of (IF ± 75 kHz) / sampling frequency.

  The threshold means 4e generates an upper limit value ((IF + 75 kHz) / sampling frequency) and a lower limit value ((IF−75 kHz) / sampling frequency) of the instantaneous angular velocity in sampling time units as threshold values and outputs them to the comparison means 4d. The comparison unit 4d compares the threshold value input from the threshold unit 4e with the angular velocity from the angular velocity calculation unit 4c, and the comparison result is out of the threshold value (angular velocity <(IF−75 kHz) / sampling frequency, (IF + 75 kHz) / sampling frequency). If <instantaneous angular velocity>, a noise detection signal indicating detection of multipath noise is output to the noise detection means 4f. When the noise detection means 4f receives the noise detection signal, the noise detection means 4f performs noise removal processing on the output of the angular velocity calculation means 4c.

FIG. 3 is a diagram illustrating a flowchart of multipath noise removal processing according to the first embodiment. In step 1, the IF signal of the digital signal converted into a complex number is input. In step 2, the above-mentioned Tan- ¹ value is calculated for the input IF signal to calculate the instantaneous angular velocity. In step 3, a delay comparison is performed on the instantaneous angular velocity calculated in step 2, and a phase change (angular velocity) in sampling time units is calculated. In step 4, it is determined whether or not the angular velocity calculated in step 3 exceeds the upper limit value (IF + 75 kHz) / sampling frequency). If the angular velocity exceeds the upper limit value, the angular velocity is fixed (pasted) at the upper limit value in step 5. If the angular velocity does not exceed the upper limit value, it is determined in step 6 whether or not the angular velocity is below the lower limit value ((IF-75 kHz) / sampling frequency). If the angular velocity falls below the lower limit value, the angular velocity is pasted on the lower limit value in step 7. By performing the above processing for each IF signal of the digital signal, noise removal processing is performed.

  Specifically, for example, assuming that IF is 10.7 MHz, the angular velocity of the sampling time unit during normal operation is a value in the range of (10.7 MHz ± 75 kHz) / sampling frequency. The angular velocity is calculated by comparing the instantaneous angular velocities of When the calculated angular velocity is out of the above range, noise is removed by fixing it to the upper limit value or the lower limit value of the above range. By removing the offset value of 10.7 MHz / sampling frequency from the angular velocity from which noise has been removed, a signal component of ± 75 kHz is extracted and FM detection (demodulation) is performed.

  As described above, the present embodiment is characterized in that the multipath noise is removed by correcting the angular velocity to a value within the range of normal operation in the process of FM detection, and performing FM detection processing with the corrected value. And

  FIG. 4 is a diagram illustrating a signal example when the present embodiment is used for FM detection of an IF signal including multipath noise. FIG. 4A shows an IF signal including multipath noise, and FIG. 4B shows a portion of the angular velocity calculated from the IF signal that exceeds the above upper limit and lower limit due to the influence of multipath noise ( For T0′-T0, T1-T2,..., T11-T12), the waveforms pasted to the upper limit value and the lower limit value are shown. Note that the output waveform of the angular velocity in FIG. 4B is a waveform example when the control signal, the stereo signal, and the like are synthesized.

  FIG. 5 is a diagram showing a waveform of an audio band output obtained by FM detection (demodulation) according to the present embodiment. FIG. 5A shows the waveform of the voice band output from which the multipath noise is not removed, and FIG. 5B shows the waveform of the voice band output when the multipath noise is removed using the present embodiment. ing. Spike-like noise amplitudes appearing in the demodulated signals in FIGS. 5A and 5B during the period where the influence of multipath noise occurs, between T1-T2, between T3-T4,..., Between T7-T8. When FIG. 5 is compared, it can be confirmed that the noise component is reduced in FIG. 5B using the present embodiment.

  FIG. 6 is a diagram illustrating a result (frequency characteristic) of performing FFT processing on each audio band output. When attention is paid to the portion surrounded by the frequency band F1-F2 and the magnitude L1FFT-L2FFT in each frequency characteristic, the frequency band F1-F2 is a voice band of 0 Hz to 19 kHz, and the frequency components of L1FFT-L2FFT are compared. When this embodiment is used, it can be confirmed that the multipath noise component is sufficiently removed because the noise component appearing in the removed speech band is about 12 dB less than when the multipath noise is not removed. .

  By adding the processing to suppress the noise part as in the prior art and output the signal immediately before the noise part as an interpolated signal during the suppressed period to the audio band output of the demodulated output of the present embodiment as described above Further, an improvement in performance of 12 dB can be expected.

(Embodiment 2)
FIG. 7 is a diagram showing a configuration of the second exemplary embodiment of the present invention. The overall configuration of the receiver of the present embodiment is the same as that of the embodiment shown in FIG. 1, and Tan- ¹ type detection and multipath noise signal detection by FM detection means 4 are also the same as those of the first embodiment. Process in a similar manner.

The FM detection means 4 of the second embodiment includes an I / Q separation means 4a for converting an IF signal into a complex number, a Tan ⁻¹ value calculation means 4b for calculating an instantaneous angular velocity by digital signal processing, and a Tan ⁻¹ value calculation means 4b. Angular velocity calculation means 4c for calculating the angular velocity by delay comparison from the output of the output, buffer memory 4h for storing the signal sequence of the angular velocity in sampling units, comparing the angular velocity with a predetermined threshold value, and detecting a noise detection signal when detecting a multipath noise signal 4d, a threshold value unit 4e that generates and outputs the threshold value, and a noise removal unit 4f that uses the contents of the buffer memory 4f based on a noise detection signal output from the comparison unit 4d. Interpolation value generation means 4i for generating and outputting an interpolation value for noise, and interpolating the interpolation value for noise removal during a multipath noise period of angular velocity. The noise removal unit 4f for performing noise cancellation, to remove the offset value from the output of the noise removing unit 4f, composed of an offset removing means 4g for outputting an FM demodulated signal.

  8 and 9 are flowcharts illustrating the processing according to the second embodiment. In the present embodiment, multipath noise is removed by interpolating the angular velocity in the multipath noise period using short-term movement history information of the angular velocity until immediately before the occurrence of multipath noise. Assuming that the interpolation period is long, hold the positive and negative peak values of angular velocity immediately before the occurrence of multipath noise, and if the interpolation value exceeds both peak values due to continued interpolation processing, the interpolation value The occurrence of an abnormal value of angular velocity is prevented by switching the further increase or decrease of the angle to decrease or increase, respectively.

In FIG. 8, in step 10, multipath noise detection processing is performed. Multipath noise detection processing is executed in the processing flowchart shown in FIG. The IF signal converted into a complex number in Step 1 of FIG. 9 is input, Tan- ¹ value calculation is performed on the IF signal input in Step 2, and instantaneous angular velocity is calculated. A delay comparison is performed on the instantaneous angular velocity calculated in step 3 to calculate an angular velocity corresponding to the phase change in sampling time units. Whether the angular velocity calculated in step 4 and step 5 exceeds the upper limit value (IF + 75 kHz) / sampling frequency) regarding the frequency of FM modulation, or is lower than the lower limit value (IF-75 kHz) / sampling frequency) If the angular velocity exceeds the upper limit (IF + 75 kHz) / sampling frequency) or falls below the lower limit (IF-75 kHz) / sampling frequency), it is determined that multipath noise has occurred, In step 7, the multipath noise detection flag, which is a detection flag indicating the detection of multipath noise, is set to “1” (noise detection), the angular velocity is equal to or higher than the lower limit (IF-75 kHz) / sampling frequency), and the upper limit. (IF + 75kHz) / sampling frequency) Multipath noise occurs when It determines that non performs processing to "0" (noise not detected) multipath noise detection flag in step 6.

  Next, when the multipath noise detection flag is “0” (noise non-detection) in step 10 shown in FIG. 8, the angular velocity is stored in the buffer in the system sampling unit in step 11. In step 12, the moving average value is calculated from the angular velocity stored in the buffer. For example, in order to calculate an interpolation value using a moving average for interpolation instead of the sample value when multipath noise is detected immediately after a certain sampling time and thereafter, the moving average is related to the moving average. Using the angular velocity stored in the buffer as data, an average value (referred to as a “moving average value”) of changes in a predetermined number of angular velocities (sampling values) before the sampling is obtained at each sampling time point. Note that the moving average value is a positive or negative value depending on the increase or decrease of the angular velocity over time. Further, in step 13, the latest two peak values on the positive side and negative side of the amplitude of the angular velocity of the predetermined number of sampling values before each sampling time are detected from the angular velocity stored in the buffer, and the peak hold value (P1) is detected. , P2) is held as a peak hold process. In step 14, the calculation result of step 12 and the measurement result of step 13 are stored.

  If the multipath noise detection flag is “1” (noise detection) in step 10, a determination process is performed in step 15 to determine whether multipath noise has been detected even at the previous sampling time. If multipath noise is not detected as a result of the determination in step 15 (NO), the moving average calculated and stored in step 12 is used as the interpolation processing for the first detection of multipath noise in step 16 to the angular velocity at the immediately preceding sampling time. The values are added and interpolated as an interpolation value (correction sample value) at the sampling time. If multipath noise is detected as a result of the determination in step 15 (YES), the movement calculated and saved in step 12 to the previous interpolation value as interpolation processing for the detection of multipath noise at least second time in step 17 The average value is added and interpolated as an interpolation value (correction sample value) at the sampling time.

  Next, in step 18, the interpolated value of the processing result in step 16 or step 17 is compared with the two peak hold values (P1, P2) on the positive side and negative side held in step 14, and 2 on the positive side and negative side are compared. If it is within the range of the two peak hold values (P1, P2), return (RETURN), and if the interpolation value of the processing result is two peak hold values on the positive side and negative side or outside the range, step 16 or After inverting the polarity of the moving average value used in step 17, return (RETURN).

  FIG. 10 is a diagram illustrating an angular velocity interpolation result according to the second embodiment. FIG. 10A shows an IF signal affected by multipath noise, and FIG. 10B shows an angular velocity interpolated using a moving average value for a part affected by multipath noise. The angular velocity of the interpolation section is controlled to change within the range of the two peak hold values (P1, P2) by the reversal control of the polarity of the moving average value, and appropriate interpolation is realized.

  In this embodiment, as a method for generating a peak hold value and correcting an angular velocity, instead of using two peak values on the positive side and negative side of the signal amplitude of a predetermined number of angular velocity sampling values, With respect to the angular velocity, the peak hold value is held by holding the positive and negative peak values of the angular velocity at a constant interval before that, for example, in units of 0.1 seconds, and is set to a specific value that does not exceed the value. It can be configured to correct the angular velocity.

  In the first embodiment, interpolation is performed such that the angular velocity is pasted to a predetermined number of the upper limit value or the lower limit value of the normal range as a method of removing multipath noise. Since the average value is used to interpolate an angular velocity that does not include a multipath noise signal from an angular velocity that includes a multipath noise signal, the first and second embodiments and In comparison, multipath noise can be removed, angular velocity distortion can be suppressed, and the noise removal effect can be enhanced. The moving average value is desirably calculated by the sampling frequency of the system / the number of samplings of 58 kHz or more.

(Embodiment 3)
FIG. 11 is a diagram illustrating a processing flowchart of the operation according to the third embodiment. The configuration of this embodiment is the same as that of the second embodiment shown in FIG. The third embodiment uses another processing method for correcting the angular velocity to a value in a predetermined range, and the angular velocity in the period during which the multipath noise is detected is the sampling time immediately before the detection of the multipath noise. The effect of multipath noise is removed by linear interpolation between the angular sample values at the sampling time immediately after the detection.

  In step 10, multipath noise detection processing from step 1 to step 8 shown in FIG. 9 is performed. That is, as described above, the multipath noise detection flag is set to “1” when the occurrence of multipath noise is detected at each sampling time, and the multipath noise detection flag is set to “0” when the occurrence of multipath noise is not detected. Is performed.

  If the multipath noise detection flag is “1” (noise detection) in step 10, a determination process is performed in step 22 to determine whether multipath noise is detected even at the previous sampling time. When the determination in step 22 is NO, that is, when multipath noise is first detected, the number of samplings (multipath noise detection sampling number) in which occurrence of multipath noise is detected in step 23 is counted by the counter, and the count value 1 is performed and a return is made (RETURN). If the determination in step 22 is YES, that is, if the occurrence of multipath noise is detected continuously, in step 24, a process of adding +1 to the multipath noise detection sampling number of the counter is performed, and the process returns (RETURN). That is, in steps 10, 22, 23, and 24, an operation is performed in which the counter counts the number of samples at each sampling point in the period in which the occurrence of multipath noise is detected.

  In step 10, when the multipath noise detection flag is “0” (multipath noise is not detected), the FM output signal is stored in the buffer in sampling units in step 18. Only the sampling value at the time when multipath noise is not generated is stored in the buffer. In step 19, it is determined whether or not the multipath noise detection sampling number of the counter is 0 when the angular velocity sampling value is stored, that is, whether or not multipath noise has been generated up to that point. If the result of determination is that the number of samplings is 0 (YES), interpolation is not required due to multipath noise, so return (RETURN). If the number of samplings is 1 or more (NO), that is, multipath noise is present. If it has occurred, the angular velocity during the period in which the multipath noise is detected in step 20 is set to the previous sampling value stored in the buffer (sample value at the sampling time immediately before the detection of the multipath noise) and the buffer. Are interpolated between the current sampling values (sample values at the time of sampling immediately after the end of the multi-pass noise) stored in (5) by the number of linear interpolation values. In step 21, the sampling number count value of the counter is cleared to 0 upon completion of the interpolation process, and the process returns (RETURN).

  FIG. 12 is a diagram illustrating an interpolation result of angular velocities according to the third embodiment. FIG. 12A shows an IF signal affected by multipath noise, and FIG. 12B shows an angular velocity obtained by linearly interpolating a portion subjected to multipath noise.

  As described above, in the third embodiment, when the angular velocity from the angular velocity calculation means 4c is substantially compared with the threshold value 4d and a section outside the threshold value is generated in the angular speed, the section is the multipath noise signal. A section determined to be affected and a normal value existing before and after the section determined to be a multipath noise signal is read from the buffer memory 4h, and the noise removal interpolation signal generation means 4i performs normal operation. Using the angular velocity in the reception state and the angular velocity after the influence of the multipath noise signal is eliminated, when the influence of the multipath noise signal is detected, a linear interpolation value between the signals is calculated. Using the calculation result of the linear interpolation, the noise removing unit 4f performs an interpolation process on a portion affected by the multipath noise signal.

(Other embodiments)
In the first embodiment, as a method of correcting the angular velocity to a value in a normal range, the comparison result between the angular velocity and a predetermined upper limit value and a lower limit threshold value. Although an example of pasting to the lower limit value has been described, a method of controlling by using a plurality of threshold values to be compared and pasting values used for correction is conceivable. Specifically, in addition to the threshold value used for detecting and correcting multipath noise in addition to the value corresponding to the maximum frequency shift width of FM broadcast, the bandwidth of stereo broadcast in FM broadcast (± 56 kHz) and the bandwidth of monaural broadcast By using a value corresponding to the width (± 16 kHz), the influence of multipath noise can be removed better.

  As another embodiment, when multipath noise frequently occurs, multipath detection using a threshold value based on the signal bandwidth of FM monaural broadcasting and pasting to an upper limit value and a lower limit value of angular velocity are performed, By making the stereo separation conventionally used monaural, it can be configured to achieve a good noise reduction in terms of hearing.

  In the above embodiment, the threshold ((IF signal center frequency + FM modulation maximum frequency deviation) / sampling frequency) / ((IF signal center frequency−FM modulation maximum frequency deviation) for detecting multipath noise from the angular velocity. In this example, the first and second specific values for noise removal have been described. However, each value is equal to or less than (center frequency of IF signal + maximum frequency deviation of FM modulation) / sampling frequency) And a second specific value equal to or greater than ((center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency) (provided that the first specific value> the second specific It is obvious that the first and second specific values can be realized by ((IF signal center frequency + FM modulation maximum frequency deviation) / sampling frequency) and It is preferable that a value in the vicinity of ((the maximum frequency deviation of the center frequency -FM modulation of the IF signal) / sampling frequency).

It is a figure which shows the 1st Embodiment of this invention. It is a figure which shows the principle of FM detection of a Tan- ¹ type | mold detection method. It is a figure which shows the flowchart of the multipass noise removal process of 1st Embodiment. It is a figure which shows the signal example at the time of using this Embodiment for FM detection of IF signal containing multipath noise. It is a figure which shows the waveform of the audio | voice band output which FM-detected (demodulated) by 1st Embodiment. It is a figure which shows the result (frequency characteristic) which carried out the FFT process of the audio | voice band output. It is a figure which shows the structure of 2nd Embodiment. It is a figure which shows the process flowchart of 2nd Embodiment. It is a figure which shows the process flowchart of 2nd Embodiment. It is a figure which shows the removal of the multipath noise of 2nd Embodiment. It is a figure which shows the process flowchart of 3rd Embodiment. It is a figure which shows the interpolation result of the angular velocity of 3rd Embodiment. It is a figure which shows the structure of a 1st prior art. It is a figure which shows the structure of a 2nd prior art. It is a figure which shows the influence of the multipath noise with respect to IF signal and FM detection output. FIG. 16 is an enlarged view of a part of the multipath noise generation portion shown in FIG. 15.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Antenna 2 Receiver front end 3 IF amplification means 3a IF output 3b S meter output 4 FM detection means 7 Stereo demodulation means 8 Low frequency amplification means 9, 10 Speaker 11 Stereo separation control means (SP means)
12 High cut control means (HC means)
4a I / Q signal separation means 4b Tan ^-1 value calculation means 4c Angular velocity calculation means 4d Comparison means 4e Threshold means 4f Noise removal means 4g Offset removal means 4h Buffer memory 4i Noise removal interpolation value generation means

Claims (10)

A step of converting the IF signal into a complex number, a step of obtaining an instantaneous angular velocity by calculating a Tan ⁻¹ value in a system sampling time unit from the complexed IF signal, and a phase change in a sampling time unit from the instantaneous angular velocity of the sampling time unit Obtaining an angular velocity that is a minute, comparing an angular velocity in sampling time units with a predetermined threshold value, and correcting an angular velocity in sampling time units exceeding the threshold range to a specific value. A characteristic multipath noise removal method.   The threshold value is (center frequency of IF signal ± maximum frequency deviation of FM modulation) / sampling frequency, and in the correcting step, the angular velocity is (center frequency of IF signal + maximum frequency deviation of FM modulation) / sampling frequency. In the case of a larger value, the angular velocity is corrected to the first specific value not exceeding the (center frequency of the IF signal + maximum frequency deviation of FM modulation) / sampling frequency, and the angular velocity is corrected to (the center frequency of the IF signal). (Maximum frequency deviation of FM modulation) / second specific value that does not fall below the angular velocity (center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency when the value is smaller than the sampling frequency. The multipath noise removing method according to claim 1, wherein the method is a step of correcting to a multipath noise.   When the angular velocity is greater than (center frequency of IF signal + maximum frequency deviation of FM modulation) / sampling frequency, the correction value is (center frequency of IF signal + maximum frequency deviation of FM modulation) / sampling frequency. Yes, when the angular velocity is a value smaller than (center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency, it is (center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency. The multipath noise removing method according to claim 2, wherein:   3. The multipath noise elimination method according to claim 1, wherein the correction value is a value calculated by a moving average of angular velocities before the latest sampling.   3. The multipath noise removal method according to claim 1, wherein the correction value is calculated by linear interpolation of angular velocity based on the latest sampling before and after the detection of the multipath noise signal. I / Q separation means for converting the IF signal into complex numbers, Tan- ¹ value calculation means for calculating an instantaneous angular velocity that is a Tan- ¹ value in the system sampling time unit from the complexized IF signal, and an instantaneous angular velocity in the sampling time unit An angular velocity calculating means for obtaining an angular velocity that is a phase change amount in sampling time units, a comparing means for comparing an angular velocity in sampling time units obtained by the angular velocity calculating means with a predetermined threshold, and an angular velocity in sampling time units exceeding the threshold value. And a noise removing means for correcting the signal to a specific correction value.   The threshold is (the center frequency of the IF signal ± maximum frequency deviation of the FM modulation) / sampling frequency, and the noise removing means has the angular velocity of (the center frequency of the IF signal + the maximum frequency deviation of the FM modulation) / sampling frequency. In the case of a larger value, the angular velocity is corrected to the first specific value that does not exceed the (center frequency of the IF signal ± maximum frequency deviation of the FM modulation) / sampling frequency, and the angular velocity is (the center of the IF signal). Frequency-maximum frequency deviation of FM modulation) / second specific value that does not fall below the angular velocity (center frequency of IF signal-maximum frequency deviation of FM modulation) / sampling frequency when the value is smaller than the sampling frequency. The multipath noise removing apparatus according to claim 6, wherein the multipath noise removing apparatus corrects to   When the angular velocity is greater than (center frequency of IF signal + maximum frequency deviation of FM modulation) / sampling frequency, the correction value is (center frequency of IF signal + maximum frequency deviation of FM modulation) / sampling frequency. Yes, when the angular velocity is a value smaller than (center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency, it is (center frequency of IF signal−maximum frequency deviation of FM modulation) / sampling frequency. The multipath noise removing apparatus according to claim 7.   The multipath noise removing device according to claim 6 or 7, wherein the correction value is a value calculated by a moving average of angular velocities before the latest sampling.   The multipath noise removing apparatus according to claim 6 or 7, wherein the correction value is calculated by linear interpolation of angular velocity based on the latest sampling before and after the detection of the multipath noise signal.

Images