JP2006054533A - Periodicity detection method and periodicity detecting apparatus for periodic signal - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problems of conventional periodicity detection methods for a periodic signal, employing the partial correlation method that takes a long time to detect the period because there are many numbers of parameter combinations and that it is difficult to obtain the period of header information, because peaks are hard to be detected when the length of the header information is short. <P>SOLUTION: In a receiver for receiving a periodic signal whose frame period is unknown, a frame length detector employs the AR model estimate method to detect the frame length. That is, the frame length detector segments data of a detection object, calculates coefficients äai} in the AR model and thereafter calculates a summed average AVsi (=si/N) (steps S2 to S5). Finally, the summed average results AVsi are arranged for detecting the peak and the peak period is obtained (step S6). The peak period obtained in this way can be detected as a period of a UW, that is a header, and this corresponds to the frame period. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a period detection method and a period detection apparatus for a periodic signal, and more particularly to a period detection method and a period detection apparatus for detecting an appearance period of a periodic signal such as specific header information included in a frame.

  When the header information is unknown regarding a 0 or 1 bit string that periodically includes header information whose length / pattern is unknown, such as a communication frame, it is very significant to obtain the period of the header information. The length and pattern of the header information can be obtained by detecting the appearance period of the header information.

  Conventionally, a partial correlation method is used as a period detection method of a periodic signal for obtaining an appearance period of specific header information from 0 and 1 bit strings periodically including specific header information whose length and pattern are unknown. . This partial correlation method will be described together with FIG. 7. As shown in FIG. 7A, the partial correlation method is to detect a header appearance period of a frame including a periodic signal such as specific header information. First, from a bit string of a given frame, a bit string (partial reference string) of a certain length Y is extracted from the position X, and correlation is obtained while shifting the given bit string by one bit at a time.

  Until the periodic peak appears, the extraction position of the reference string and the number of bits are changed, and the correlation values are arranged as shown in FIG. 7B to detect the peak. In the partial correlation method, the period between the peaks is the period of the header information to be obtained.

  Further, as another conventional period detection method for detecting the appearance period of a periodic signal, there is known a method for detecting a weak signal with time periodicity buried in noise by using a stochastic resonance phenomenon (for example, , See Patent Document 1). In this conventional period detection method, a noise signal in which a weak signal with time periodicity is buried is input to a non-linear system, the power spectrum of the output time series of the non-linear system is analyzed, and a distribution representing statistical properties is obtained. While calculating gradually, increasing the input noise intensity of a noise source prepared separately for input to the nonlinear system, calculate a distribution that similarly represents statistical properties, and calculate the power spectrum for each noise intensity. The peak of each power spectrum is detected, its peak value is determined, the maximum of these peak values is determined, and the period of the maximum peak is set as the period of the weak signal.

JP 2000-221546 A

  However, in the conventional periodic signal periodic detection method using the partial correlation method, the parameters for taking the partial correlation depend on two parameters of the position and length of the bit string, and these two If the parameter combinations are not correct, peaks cannot be detected when the correlation values are arranged, so that there are many combinations and it takes time to detect the period. Further, when the length of the header information is short, it is difficult to detect a peak, and it is difficult to obtain the period of the header information.

  Further, in the conventional periodic signal periodic detection method described in Patent Document 1, the distribution representing the statistical properties is similarly obtained while gradually increasing the input noise intensity of a noise source separately prepared for input to the nonlinear system. Since the power spectrum is calculated for each noise intensity and the peak of each power spectrum is detected, there is a problem that it takes time to detect the period of the weak signal buried in the noise.

  The present invention has been made in view of the above points, and an object of the present invention is to provide a period detection method and a period detection apparatus for a periodic signal that can detect a signal period (frame length) with high accuracy in a relatively short time. To do.

  In order to achieve the above object, the periodic signal period detection method according to the first aspect of the present invention is such that a frame composed of a fixed pattern header and fixed length data is synthesized in time series, and the frame period is unknown. A first step of receiving a periodic signal and applying an autoregressive (AR) model estimation method to the received periodic signal to detect a peak period among a plurality of coefficients of the AR model estimation method A second step of obtaining a header period, and a third step of detecting a frame period of the received periodic signal based on the header period obtained in the second step. And

  In order to achieve the above object, the periodic signal period detection method according to the second aspect of the present invention is the above-described second step, wherein the periodic signal data is cut out to obtain a plurality of coefficients in the AR model estimation method. A fourth step of repeatedly calculating a value for each separately and adding them repeatedly for a predetermined number of times to calculate an addition average, and a fifth step of obtaining a peak value by arranging the respective average addition values of a plurality of coefficients. The frame period is detected from the period of the peak value.

  In the periodic signal period detection method of the first and second inventions described above, the period of the header can be obtained by applying the AR model estimation method to detect the peak periods of the values of the plurality of coefficients. .

  In order to achieve the above object, the periodic signal according to the third aspect of the present invention is a periodic signal in which frames composed of a fixed pattern header and fixed length data are synthesized in time series, and the frame period is unknown. Based on the first step of receiving the signal, the second step of applying the autocorrelation method to the received periodic signal to determine the period of the header, and the period of the header obtained in the second step, And a third step of detecting a frame period of the received periodic signal. In the present invention, the autocorrelation method can be applied to detect the frame period of a periodic signal whose frame period is unknown.

  In order to achieve the above object, the period detection device according to the first aspect of the present invention is a periodicity in which frames composed of a fixed pattern header and fixed length data are synthesized in time series, and the frame period is unknown. A receiving means for receiving a signal, and applying an autoregressive (AR) model estimation method to the received periodic signal to detect a peak period among a plurality of coefficients of the AR model estimation method, It is characterized by having period detection means for obtaining the period of the header and extraction means for extracting data of the received periodic signal based on the header period obtained by the period detection means.

  In order to achieve the above object, a period detection device according to a second aspect of the present invention includes a period detection unit according to the first aspect of the present invention, comprising a plurality of cascade-connected adders to which a periodic signal is supplied, and a periodic signal. And multiplying each output signal of the plurality of delay circuits separately from each of a plurality of coefficients of the AR model estimation method, and multiplying the multiplication results by a plurality of adders A plurality of multipliers to be supplied to the corresponding one adder, and a plurality of coefficients of the plurality of coefficients so that the prediction error output from the adder at the final stage among the plurality of adders is minimized. The determination of each value is repeated a preset number of times, and the cycle of the header is detected from the cycle of the peak value of each of the obtained average addition values of the plurality of coefficients.

  In the period detection devices of the first and second inventions described above, the period of the header can be obtained by applying the AR model estimation method and detecting the periods of the peaks of the plurality of coefficient values.

  Furthermore, in order to achieve the above object, the period detection device according to the third aspect of the present invention is a periodicity in which frames including a fixed pattern header and fixed length data are synthesized in time series, and the frame period is unknown. A receiving means for receiving a signal, a period detecting means for obtaining a header period by applying an autocorrelation method to the received periodic signal, and a received period based on a header period obtained by the period detecting means And extraction means for extracting sex signal data. In this invention, the autocorrelation method is applied to detect the frame period of a periodic signal whose frame period is unknown.

  According to the present invention, the peak value among a plurality of coefficients of the AR model estimation method is obtained, the header period is obtained based on the period of the peak value, and the frame period and data of the periodic signal are further determined from the header period. Because it depends on the two parameters of the position and length of the bit string and the combination of the two parameters is not correct, the peak cannot be detected when the correlation values are arranged Compared to the correlation method, it is less susceptible to noise, and can accurately detect the header period and the frame period, and the frame period detection accuracy can be improved as compared with the conventional method.

  Next, the best mode for carrying out the present invention will be described. FIG. 1 is a block diagram showing an embodiment of a periodic signal period detection apparatus according to the present invention. This embodiment is a period detection device used in an illegal radio wave monitoring system to control illegal radio waves and the like, and has a function of listening to illegal radio waves as a function of the device. This is a function for playing back the collected radio waves up to the original voice. When communication is performed using a digital wireless device, the system configuration shown in FIG. 1 is used to implement this function.

  In this embodiment, a signal having a format as shown in FIG. 2 used in a digital radio is received. In the figure, “UW” is a unique word having a fixed bit string pattern, which is a periodic signal transmitted from a bit string having a value of 0 or 1 and periodically transmitted. “DATA” is fixed-length data to be transmitted, and consists of a bit string having a value of 0 or 1. A value obtained by adding the length (time length) of these unique words and the data length (time length) is the frame period.

  Returning to FIG. 1, the antenna 11 receives illegal radio waves and other various radio waves, and supplies the received signals to the modulation scheme estimation / decoder 12. The modulation scheme estimator / decoder 12 estimates the modulation scheme of the input received signal, then demodulates the scheme based on the estimated modulation scheme, and supplies the decoded signal to the decoding unit 13. The decoding unit 13 includes a scrambler estimator / descrambler 15, a frame length detector 16, and a UW / data extractor 17.

  The scrambler estimator / descrambler 15 receives the decoded signal output from the modulation scheme estimator / decoder 12 as an input, estimates the scrambler used in the digital radio with respect to the input signal, After that, descrambling is performed to release the estimated scrambler.

  The frame length detector 16 is a circuit constituting the main part of the present embodiment, and uses an AR (autoregressive) model estimation method to be described later for a signal supplied from the scrambler estimator / descrambler 15. Thus, the UW appearance period of FIG. 2 corresponding to the header information, that is, the frame length is detected. The UW / data extractor 17 receives the detection result of the frame length detected by the frame length detector 16 as an input, extracts data based on the detected frame length, and decodes it into an audio signal.

  Next, the operation of the present embodiment will be described. The illegal radio wave received by the antenna 11 is supplied as a received signal to the modulation scheme estimator / decoder 12, where the modulation scheme is estimated, demodulated in a scheme corresponding to the estimated modulation scheme, and 0 or This is a decoded signal that is a data string of 1 values.

  The received signal decoded by the modulation scheme estimator / decoder 12 is descrambled by a scrambler estimator / descrambler 15 in which the scramble scheme applied to the received signal is estimated by a known method and the scramble is released. Is given. The descrambled data Xt is supplied to the frame length detector 16, and the appearance period of UW (corresponding to header information in the present invention) shown in FIG. 2, that is, the frame period is detected by the AR model estimation method described later. The

  The frame period information detected by the frame length detector 16 is supplied to the UW / data extractor 17 together with the input data Xt, where data excluding UW in the frame (DATA in FIG. 2) is extracted. The extracted data is reproduced in the UW / data extractor 17 or in the reproduction circuit of the next stage (in the case of compression, it may be necessary to estimate the compression method), for example, the original audio signal, or Restored to FAX data or the like.

Next, the AR model estimation method used by the frame length detector 16 for frame length detection will be described in detail with reference to the configuration of FIG. 3 and the flowchart of FIG. In FIG. 3, a part of the frame length detector 16 includes p adders 21 _{1 to} 21 _p connected in cascade and p delay units 23 ₁ connected in cascade to delay the input signal Xt by unit time. to 23 and _p, delayer _{23 1,} 23 2, ..., and the output delay signal of 23 _p, the coefficient _{a 1,} a 2, ..., and _{a p} multiplied separately, the multiplication result .., 21 _p and corresponding multipliers 22 _{1 to} 22 _p that output to the corresponding adders 21 ₁ , 21 ₂ ,.

In the AR model estimation method used by the frame length detector 16 for frame length detection, first, the value si in the arithmetic mean calculation memory is cleared to 0 beforehand (step S1 in FIG. 4). Subsequently, the detection target data is cut out, and the coefficient {ai} in the AR model shown in FIG. 3 is calculated (step S2 in FIG. 4). Here, i is a value from 1 to p, and the coefficient {ai} is a coefficient a ₁ to a coefficient that minimizes a prediction error that is the output of the adder 21p at the final stage in FIG. a _p .

  Subsequently, the value of the coefficient ai obtained this time is added to the value si of each arithmetic mean calculation memory to update the value of si (step S3 in FIG. 4). Subsequently, it is determined whether or not the current calculation result in step S3 is a preset Nth calculation result (step S4 in FIG. 4). Otherwise, the coefficient ai is calculated again in step S2, and the averaging in step S3 is performed again. The value si of the calculation memory is updated.

  In this way, the calculations in steps S2 and S3 are repeated N times set in advance. When iterating, the data cut out in step S2 is always cut out from disjoint parts. In addition, the number of repetitions N is determined experimentally in advance.

  When it is determined in step S4 that the calculation has been performed N times, the value si of each addition average calculation memory is divided by N to calculate the addition average AVsi (= si / N) (step S5 in FIG. 4). Finally, peak detection is performed by arranging the averaging results AVsi to obtain a peak period (step S6 in FIG. 4). That is, for each addition average AVsi, p addition average values from coefficients a1 to ap are obtained, and a peak value among them is obtained. In this way, the peak values are sequentially obtained with different times for the received signal. The peak period obtained in this way can be detected as the UW period as the header, and this corresponds to the frame period.

  Next, the results of detection of the frame length according to this embodiment and the experimental results of the conventional frame length detection result will be described with reference to FIG. 5A and 5B and FIGS. 5C and 5D to be described later, the vertical axis indicates the correlation value or the AR filter coefficient value, and the horizontal axis indicates the position for calculating the correlation (in the AR model estimation method, the coefficient ai, where i = p is 1025).

  When the UW length is sufficient (for example, 24 bits), the frame length detection result by the conventional partial correlation method is shown in FIG. 5A, and the frame by the AR model estimation method (linear prediction method) according to this embodiment is shown in FIG. The long detection result is shown in FIG. When the UW length is sufficiently long, noise is conspicuous in the conventional partial correlation method shown in FIG. 5A, but the peak position can be identified by both the partial correlation method and the AR model estimation method (linear prediction method). Therefore, the peak period, that is, the frame period can be determined.

  However, when the UW length is not sufficient (for example, 8 bits), the frame length detection result by the conventional partial correlation method is shown in FIG. 5C, and the AR model estimation method (linear prediction method) according to the present embodiment. As shown in FIG. 4D, the conventional partial correlation method has a peak correlation value that is not so different from the correlation values of other positions, and it is difficult to identify the peak position. On the other hand, according to the present embodiment, as shown in FIG. 5D, there is a sufficient difference in the value between the coefficient with the peak and the coefficient with no peak, and there is a peak at the position of ai = 512. I understand. Therefore, according to the present embodiment, it is possible to accurately detect the frame period of the periodic signal of a weak signal buried in noise, which has been difficult in the past, even in a short UW length.

  Next, another embodiment of the present invention will be described. In the above embodiment, the frame length is detected by the AR model estimation method, but in the other embodiments, the AR model estimation method is replaced with an autocorrelation method. This autocorrelation method uses an input periodic signal, which is a bit string of data 0 and 1, shown in FIG. 6 (A) with the signal shown in FIG. 6 (B) in which the data is shifted with respect to the periodic signal. This is a method of obtaining a correlation.

  In this method, if the data length is very long, the amount of calculation increases, so the result is actually output after being cut off to some extent, but this autocorrelation method has substantially the same effect as the above embodiment. Is obtained.

  Although not directly related to the present invention, in the above embodiment, the modulation scheme is estimated, but it is not necessary if the modulation scheme is known, and if the scramble method is also known. The scramble estimation means is unnecessary. On the other hand, when receiving a radio wave whose error correction code method is unknown, it is necessary to provide an error correction code estimation / decoder at the first stage of the decoding unit 13.

It is a block diagram of one embodiment of the device of the present invention. It is a figure which shows an example of the format of the signal received with the apparatus of FIG. It is a block diagram of an example of the principal part of the apparatus which implement | achieves AR model estimation method. It is an explanatory flowchart of an example of an AR model estimation method. It is a figure which shows the frame length detection result of the conventional apparatus and one embodiment of this invention apparatus by comparing when the UW length is long and when it is short. It is explanatory drawing of an example of the autocorrelation method used for frame length detection in other embodiment of this invention apparatus. It is explanatory drawing of an example of the partial correlation method used for frame length detection with the conventional apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Antenna 12 Modulation system estimation / decoder 13 Decoding part 15 Scrambler estimator / descrambler 16 Frame length detector 17 UW / data extractor

Claims (6)

A first step of receiving a periodic signal in which a frame composed of a fixed pattern header and fixed length data is synthesized in time series, and the frame period is unknown;
A process for obtaining a period of the header by applying an autoregressive (AR) model estimation method to the received periodic signal and detecting a peak period among a plurality of coefficients of the AR model estimation method. Two steps,
And a third step of detecting a frame period of the received periodic signal based on the period of the header obtained in the second step.   In the second step, the data of the periodic signal is cut out and the values are separately calculated for each of the plurality of coefficients in the AR model estimation method. 4. A fourth step of calculating, and a fifth step of obtaining a peak value by arranging the respective average addition values of the plurality of coefficients, and detecting the frame period from the period of the peak value. Item 6. A periodic signal period detection method according to Item 1. A first step of receiving a periodic signal in which a frame composed of a fixed pattern header and fixed length data is synthesized in time series, and the frame period is unknown;
A second step of applying an autocorrelation method to the received periodic signal to determine the period of the header;
And a third step of detecting a frame period of the received periodic signal based on the period of the header obtained in the second step. A receiving means for receiving a periodic signal in which a frame composed of a fixed pattern header and fixed length data is synthesized in time series, and the frame period is unknown;
A period for obtaining the period of the header by applying an autoregressive (AR) model estimation method to the received periodic signal and detecting a peak period among a plurality of coefficients of the AR model estimation method. Detection means;
A period detecting device comprising: extraction means for extracting the data of the received periodic signal based on the period of the header obtained by the period detecting means.   The period detection means includes a plurality of cascaded adders to which the periodic signal is supplied, a plurality of cascaded delay circuits that delay the periodic signal by unit time, and the AR model estimation method. A plurality of multipliers for multiplying each output signal of the plurality of delay circuits separately from each of the plurality of coefficients and supplying the multiplication result to a corresponding one of the plurality of adders; Determining a value of each of the plurality of coefficients so as to minimize a prediction error output from an adder at a final stage of the plurality of adders, and the circuit unit is obtained a predetermined number of times. 5. The period detection device according to claim 4, wherein the period of the header is detected from a period of a peak value of each addition average value of the plurality of coefficients. A receiving means for receiving a periodic signal in which a frame consisting of a fixed pattern header and fixed length data is synthesized in time series, and the frame period is unknown,
Period detection means for obtaining the period of the header by applying an autocorrelation method to the received periodic signal;
A period detecting device comprising: extraction means for extracting the data of the received periodic signal based on the period of the header obtained by the period detecting means.

Images