JPH09304500A - Signal detection method - Google Patents

Abstract

(57) [Abstract] [Purpose] To detect weaker signals than before in a signal detection method that detects a narrowband signal in water using a sonobuoy having a directional receiver and obtains the direction of arrival. [Arrangement] An azimuth dispersion calculation unit 13 calculates a dispersion of azimuth values in a certain time interval stored in the azimuth storage unit 12 for a frequency component which is primarily determined to be a signal by the level detection in the signal detection unit 10. . The signal determination unit 14 secondarily determines the signal and noise depending on whether the dispersion value of the direction is larger or smaller than the threshold. Since noise is isotropic, the variance is large, and since the signal is directional, the variance is small. [Effect] A low S / N signal can be detected without deteriorating the false alarm rate.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal detection method for detecting underwater sound, and more particularly to a signal detection method for a narrowband signal of underwater sound by a sonobuoy having a directional receiver.

[0002]

2. Description of the Related Art A conventional signal detection method comprises a sonobuoy having a directional receiver and a system for performing signal processing thereof in order to detect a narrow band signal and obtain its frequency and direction. This will be described with reference to the drawings.

FIG. 2 shows a conventional system, which is composed of a directional sonobuoy 1 on the transmitting side and a receiving side 2. The sonobuoy 1 is an omnidirectional receiver 11, a north-south dipole directional receiver 12 having dipole directivity north and south with respect to magnetic north, and an east-west dipole directional receiver 13 having dipole directivity east and west. The transmitting unit 14 transmits the underwater sound received by those wave receivers to an observation device or the like by radio waves.

The receiving side 2 is an observing device and the like, and a receiving section 3 for receiving radio waves from the Sonobui, frequency analyzing sections 4, 5, 6 for frequency-analyzing the signals of the three Sonobui receivers, Integrators 7, 8 and 9 that temporally integrate the analysis result for each frequency component, and a signal detector 10 that detects a signal from the integrated frequency analysis result of the underwater sound received by the omnidirectional receiver. And an azimuth calculation unit 11 that calculates the arrival azimuth of the received sound from the relationship between the reception levels of the respective receivers.

Next, the operation will be described. The narrowband signal emitted by the underwater source is directional sonobuoy 1,
Omnidirectional receiver 11, north-south dipole directional receiver 1
2. Received by the east-west dipole directional receiver 13. Each received sound is sent to the transmission unit 14, frequency-multiplexed, modulated to a radio frequency, amplified in power, and then sent out via the antenna. On the other hand, on the receiving side 2, this radio wave is sent to the receiving section 3 via the antenna and demodulated to receive the received sound of the omnidirectional receiver 11 to the frequency analyzing section 4 of the north-south dipole directional receiver 12. Frequency analysis unit 5
Then, the received sound of the east-west dipole directional receiver 13 is output to the frequency analysis unit 6, respectively.

In the frequency analysis unit 4, the underwater sound received by the omnidirectional receiver 11 is subjected to frequency analysis to obtain a frequency spectrum. When a narrow band signal exists in water, it appears as a narrow band spectrum in the frequency spectrum. Underwater sound generally contains a lot of ambient noise, so noise appears in the frequency spectrum as well, and the noise spectrum fluctuates at each observation time, making it difficult to detect narrowband signals. Integrate for the given time length. Next, the integration result is sent to the signal detection unit 10. In the signal detection unit 10, a threshold is set with the average value of the noise spectrum as a reference, and a spectrum exceeding the threshold is detected as a signal component. The method is as follows.

First, the noise average value Mp of the frequency component data of the underwater sound is obtained by the equation 1. It is assumed that there are N cell frequency components and the frequency cells are arranged in ascending order.

[0008]

[Equation 1]

However, Xi: the level of the i-th cell (i
= 1 to N), Mp: noise average value (p = 1 to N) for the p-th cell, W: number of cells used for noise average value calculation, p <= W / 2 or p> = N- In the case of W / 2, noise average values of p = W / 2 + 1 and p = N−W / 2−1 are used, respectively. Also, the cell for calculating the noise average is excluded from the calculation target.

Next, the threshold Ti is calculated from the equation 2 based on the noise average value.

Ti = K · Mi (2) where K: constant Then, the threshold is compared with the frequency component, and a signal exceeding the threshold is detected as a signal.

On the other hand, the received sound of the north-south dipole directional receiver 12 is sent from the receiving section 3 to the frequency analysis section 5, and the received sound of the east-west dipole directional receiver 13 is sent to the frequency analysis section 6. After frequency analysis, it is sent to each of the integrators 8 and 9 to perform integration of a predetermined time length for each frequency component. The outputs of the integrators 8 and 9 are sent to the azimuth calculator 11. The azimuth calculation unit 11 obtains the arrival azimuth of the signal as described below. FIG. 3 shows the incoming signal S at each receiver.
It shows how (t) is received. If the output of the north-south dipole directional receiver is NS and the output of the east-west dipole directional receiver is EW,

[0013]

[Equation 2]

Here, K is a constant such as sound pressure sensitivity. In the above formula

[0015]

(Equation 3)

Therefore, EW / NS = sinα / cosα = tanα Therefore, α = arctan (EW / NS) By the way, although the quadrant of the azimuth cannot be determined only by the above equation, the north-south dipole is the output characteristic of the receiver. In the directional receiver, when the sound enters from the north side, it becomes in-phase with the output of the omnidirectional receiver, and when the sound enters from the south side, it becomes the opposite phase of the output of the omnidirectional receiver. Similarly, the east-west dipole directional receiver has the same phase when entering from the east, and the opposite phase when entering from the west. By examining the phase relationship of each receiver with the table in Fig. 4, the quadrant Make a decision.
The azimuth calculation is performed for each frequency component subjected to the frequency analysis to determine the azimuth.

As described above, in the conventional method, whether the signal is a narrow band signal or not is determined by the threshold based on the level magnitude.

[0018]

In this conventional signal detection method, whether or not a narrow band signal is present is determined only by the magnitude of the level. Therefore, when a low S / N narrow band signal is detected, the threshold value is detected. Need to lower. As a result, the probability of noise components being detected also increases, resulting in a large number of false detections, which is not practical.

[0019]

The signal detection system of the present invention frequency-analyzes the received sound of each of the Sonobuoy omnidirectional receiver, the north-south dipole directional receiver, and the east-west dipole directional receiver. Means, a means for time-integrating the result for each frequency component, a signal detecting means for making a primary discrimination between a signal and noise by providing a lower threshold level for the frequency component of the integration result, and the above three types of receiving means. Means for calculating the azimuth for each frequency component from the level ratio and phase determination of the integration result of the frequency components of the wave sound, means for storing this in time series, and frequency for which the signal is primarily discriminated by the signal detecting means. A means for calculating a time-series azimuth variance of components and a signal deciding means for secondarily deciding that, of the frequency components that have been primarily discriminated, those whose azimuth variance is below a predetermined threshold are signals. It is equipped with a.

[0020]

The present invention will be described below with reference to the drawings. FIG. 1 shows an embodiment of the present invention. The transmitting side sonobuoy used is conventional and is the same as the directional sonobuoy 1 in FIG. 2 and is omitted in FIG.

In FIG. 1, the radio waves transmitted from the sonobuoy are sent to the receiving unit 3 to cause omnidirectionality, north-south dipole directivity,
It is demodulated into the received sound of each receiver of the east-west dipole directivity, frequency-analyzed by the frequency analyzers 4, 5, 6 and then time-integrated for each frequency component by the integrators 7, 8, 9. Is also the same as before. Further, the azimuth calculation unit 11 calculates the azimuth for each frequency component from the level ratio of the integrated value of the frequency components of the received sound of these three types of wave receivers and the phase determination, which is also the same as the conventional method.

On the other hand, in the signal detecting section 10, the level threshold is set by the conventional formulas 1 and 2, but the constant K in the formula 2 is set to a value smaller than the conventional one. FIG. 5 shows the level threshold. Since the signal component is superimposed on the underwater noise, the distribution of the noise component and the distribution of the signal component having a peak at a higher level than that can be obtained by taking the occurrence frequency for each level. As shown in the figure, when the signal is discriminated by the threshold TH of the level, the black portion erroneously detects noise as a signal, that is, an erroneous alarm. Generally, the threshold is set to a level determined by a necessary detection rate (hatched portion in FIG. 5) and a false alarm. In the present invention, the threshold is lowered as compared with the conventional method. As a result, it becomes possible to detect a signal having a lower S / N than the conventional one. However, when the threshold is lowered, false alarms increase, but in the present invention, the noise component is removed by checking the consistency of the azimuth as a secondary discrimination as described below.

Returning to FIG. 1, the bearing storage unit 12 stores the bearing values for each frequency component calculated by the bearing calculation unit 11 in a time series. In the azimuth dispersion calculator 13, the signal detector 1
At 0, the azimuth value in the fixed time interval is read from the azimuth storage unit 12 for the frequency component that is primarily discriminated as a signal based on the level, and the variance thereof is calculated. The signal determination unit 14 secondarily determines the signal and the noise depending on whether the variance value of the direction is larger or smaller than the threshold.
Since noise can be regarded as isotropic, the variance is large. On the other hand, the direction of the signal varies because the noise is superimposed on the signal, but since the signal itself has directionality, the variance of the direction is smaller than when only noise is used. . By utilizing this point, the noise component erroneously included in the primary discrimination can be removed.

The calculation of the variance will be described here. Considering _{N 1} , azimuth value data obtained for each frequency analysis cycle for a certain frequency as x ₁ , x ₂ , ..., X _N , the variance is obtained by the following equation.

[0025]

(Equation 4)

The threshold level with respect to the obtained dispersion has a deviation (square root of dispersion) of about 1 when it is assumed that the azimuth values are uniformly distributed between 0 ° and 360 °.
Since it is 04, a value obtained by multiplying this by 1 / K 'is set.
The value of K'may be varied in response to various situations on the actual sea level, for example.

[0027]

As described above, according to the present invention, the threshold of the detection level is lowered to lower the S / N ratio than the conventional one.
Signals can be processed, and false detections that increase due to this can be suppressed by focusing on the magnitude of the azimuth dispersion, so
This has an effect that a target signal can be detected even for an acoustic signal having a poor S / N.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a block diagram of a conventional system.

FIG. 3 is an explanatory diagram of a directivity pattern and reception of a directivity sonobuoy.

FIG. 4 is a table showing the determination of the orientation quadrant of the detection signal.

FIG. 5 is an explanatory diagram of frequency distributions of signal and noise levels and thresholds.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Directivity Sonobui 2 Reception side 3 Reception part 4-6 Frequency analysis part 7-9 Integration part 10 Signal detection part 11 Direction calculation part 12 Direction storage part 13 Direction dispersion calculation part 14 Signal determination part

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yasushi Sasaki 2-4-16-203 Oyamaguchi, Shirai-machi, Inba-gun, Chiba (72) Inventor Taku Yamamoto 5-7-1, Shiba, Minato-ku, Tokyo NEC Corporation Inside the company

Claims (3)

[Claims] 1. A first discrimination for firstly discriminating between a signal and noise by comparing an output from an omnidirectional receiver, a directional receiver, and an output from the omnidirectional receiver with a first threshold. Means, direction detecting means for receiving the output from the directional receiver and determining the arrival direction of the received signal when the signal is discriminated by the first discriminating means, and the time series of the detection result from the direction detecting means. Signal detection method comprising: a variance calculation unit that obtains a typical variance; and a second determination unit that compares the variance with a second threshold to make a secondary determination as to whether it is a signal or noise. 2. The received signal is divided into a plurality of frequency bands,
The signal detection system according to claim 1, wherein it is determined whether the signal is a signal or noise for each frequency band. 3. A means for frequency-analyzing the received sound of each of the Sonobui omnidirectional receiver, the north-south dipole directional receiver, and the east-west dipole directional receiver, and the frequency analysis result for each frequency component. Means for time-integrating, a signal detecting means for making a primary discrimination between a signal and noise by providing a first threshold for the frequency component of the integration result, and a level ratio and phase of the integration result of the frequency components of the three types of received sounds. Means for calculating the azimuth for each frequency component from the judgment, and 1 if it is a signal
A means for calculating the time-series azimuth variance of the next discriminated frequency component, and a quadratic discrimination that the signal whose azimuth variance is less than or equal to the second threshold among the first-discriminated frequency components is a signal. A signal detection method having a signal determination means for performing.