JPH0774731A - Ciphering voice signal generator and ciphering voice signal receiver - Google Patents

Abstract

PURPOSE:To improve the degree of ciphering by using a QMF for division of a signal into a band of a predetermined width and injecting a synchronizing signal in place of a base band signal of a band of a high frequency with comparatively less information quantity among base band signals at divided bands. CONSTITUTION:A ciphering key converter 1 decides how voice signals on a frequency axis are to be rearranged and scrambled based on a key number designating a rule of the rearrangement. A QMF analyzer 2 divides a band of the voice signals and uses a QMF to obtain a base band signal of each band. A replacement device 3 injects a frequency-modulated synchronizing signal in place of a signal being a high frequency component in the base band signal being an output signal of the analyzer 2 and replaces the base band signals based on an output signal of the converter 1. A QMF synthesizer 4 receives a signal from the replacement device 3 and synthesizes the signals based on the QMF to provide an output of a ciphered voice signal. Thus, number of divisions of a band having lots of information quantity is increased and number of divisions for a band with less information quantity is decreased to improve the ciphering degree even when number of divisions is the same.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a secret voice signal generator and a secret voice signal receiver for voice communication,
In particular, a secret voice signal generator and a secret voice signal receiver that scramble on the frequency axis, and a secret voice signal generator that transmits and receives a synchronization signal necessary for descrambling on the time axis as one segment of scramble on the frequency axis, This invention relates to a secret voice signal receiving device.

[0002]

2. Description of the Related Art In recent years, the protection of information confidentiality in communication has become a social problem, and in particular, there is an increasing demand for an effective confidential communication method for preventing information leakage in voice communication. The secret methods can be roughly classified into analog methods and digital methods. As the analog method, scrambling on the time axis and scrambling on the frequency axis are generally used. In the scrambling on the time axis, a signal to be transmitted is cut at predetermined time intervals on the time axis (decomposed into segments).
Then, the segments are rearranged according to the rearrangement rules arranged between the transmitting side and the receiving side. In addition, scrambling on the frequency axis divides the signal to be transmitted into frequency spectra, cuts it into predetermined frequency widths on the frequency axis (decomposes into segments), and follows the rearrangement rules arranged between the transmitting side and the receiving side. , To sort the segments.

FIG. 14 shows an execution process of scrambling on the frequency axis in the conventional confidential voice signal generator and confidential voice signal receiver. On the side of the secret voice signal generator, the A / D-converted voice sample value sequence is sequentially divided into frames using a rectangular window, and the N sample values (for one frame) are fast Fourier transformed (FFT: Fast Fourier Transform).
Transform), the frequency spectrum array obtained by this is rearranged according to the encryption key, and then the inverse fast Fourier transform (IFFT: Inverse) is performed.
It is converted into a time-series signal by the Fast Fourier Transform) and is sent as a scrambled signal. On the side of the confidential voice signal receiving device, the scrambled signal is received, frame synchronization is performed, the same frame as that on the side of the confidential voice signal generating device is divided and FFT is performed, and the spectrum obtained by this is reversed. further,
A reproduced voice signal is obtained by performing IFFT on the restored spectrum.

[0004]

FIG. 15 is a diagram for explaining frequency spectrum division and scrambling. In the conventional confidential voice signal generator and confidential voice signal receiver, FIG.
As shown in, the frequency spectrum could be divided only at equal intervals. To increase the degree of confidentiality, it is necessary to increase the number of divisions, but if the number of divisions of the band with a large amount of information of the audio signal is increased, the number of divisions of the band with a small amount of information also increases, and the processing time increases. . Further, even if the signal is divided by a predetermined bandwidth, it cannot be divided perpendicularly to the frequency axis, so that a signal in a certain band leaks to an adjacent band. In the conventional technique, it is difficult to perform processing on both ends of the divided band, so that the sound quality is deteriorated due to the signals leaking to the adjacent divided bands. Furthermore, it takes time to create the program. In the confidential communication method, scrambling on the time axis and scrambling on the frequency axis may be used together, but the synchronization timing is easily known. An object of the present invention is to change a frequency spectrum division interval between a band having a large amount of information of a voice signal and a band having a small amount of information, and to easily perform processing on both ends of the divided band,
It is an object of the present invention to provide a confidential voice signal generating device and a confidential voice signal receiving device in which a program can be easily created and a synchronization timing is hard to determine.

[0005]

In order to solve this problem, the present invention converts a voice signal into a frequency spectrum and divides it into bands of a predetermined width by using a QMF (Quadratu).
re MirrorFilter). In addition, the synchronization signal is injected instead of the baseband signal of the high frequency band in which the amount of audio information is relatively small among the baseband signals of the band divided by QMF. The sync signal is FM modulated for injection instead of the baseband signal. Then, on the receiving side, phase estimation is performed using a differential method to establish synchronization.

[0006]

FIG. 5 shows a block diagram of the components of the QMF.
Here, the sampling rate is 2 of the Nyquist frequency.
X0, h0 (n) is a low-pass filter (LPF) with a cutoff frequency of π / 2, and h1 (n) is a cutoff frequency of π /
Suppose it is a high pass filter (HPF) of 2. If the spectrum of the input signal x (n) is as shown in Fig. 6 (a), the LPF side of the analyzer extracts only the low frequency side of the spectrum of x (n) and undersamples it. The theorem is satisfied, and a baseband signal having a spectrum as shown in FIG. 6B is obtained. On the other hand, on the HPF side, since only the high frequency side of the x (n) spectrum is taken out and under-sampled, the spectrum is shown in FIG. 6 (c).
The signal in the base region is obtained by folding back as shown in. In the synthesizer, when the HPF and LPF are ideal filters, x (n)
And x '(n) are exactly the same.

For example, as shown in FIG. 7, if the QMFs are cascade-connected in several stages and the decomposed baseband signals are exchanged, the voice signal is converted into a frequency spectrum and divided into bands of a predetermined width, and on the frequency axis. Can be scrambled. The upper side of the branch branched in FIG. 7 corresponds to the LPF side in FIG. 5, and the lower side corresponds to the HPF side. From FIG. 5, the number of filters in the n-th stage is 2 to the n-th power, so it seems that the processing time is longer than that in the first-stage part, but in reality, undersampling is halved at each stage. Since it is performed, the calculation amount of each stage is almost the same. Therefore, the calculation amount for the number of divisions increases only logarithmically, which is efficient. Also,
If the number of cascade stages is not the same, for example, as shown in FIG. 8, the division intervals can be changed as shown in FIG.
By changing the characteristics of F and HPF (shifting the cutoff frequency), mutual signals can be prevented from leaking into the adjacent bands that are divided.

[0008]

EXAMPLES Examples will be described with reference to FIGS. 1 and 2. FIG. 1 is a block diagram of a secret voice signal generator according to the present invention. How to sort the audio signals on the frequency axis with respect to the key number that specifies the sorting rule,
Cryptographic key converter for deciding whether to perform scrambling 1
And a QMF analyzer 2 that divides an input signal (voice signal) into bands and obtains a baseband signal of each band by QMF,
An FM-modulated synchronizing signal x (t) is injected instead of the signal which is the high frequency component of the baseband signal which is the output signal of the QMF analyzer 2, and the replacement of these baseband signals is performed by the cryptographic key conversion. Exchanger 3 which is performed based on the output signal of the device 1
And the signal of the switching device 3 is received and combined by QMF,
It is composed of a QMF synthesizer 4 which outputs a secret voice signal.

FIG. 2 is a block diagram of a confidential voice signal receiving apparatus. An encryption key converter 5 for deciding how to rearrange and scramble the audio signals on the frequency axis with respect to the key number, and to receive the confidential voice signal and band-divide the confidential voice signal. QM the baseband signal of the band
The QMF analyzer 6 obtained by F and the signal that is the highest frequency component of the baseband signal that is the output signal of the QMF analyzer 6 are F
A mutator 7 which outputs the M-modulated sync signal x (t), and scrambles the baseband audio signal, which is the other frequency component, based on the output signal of the encryption key converter 5 and outputs the same. , Q which receives the signal of the interchanger 7 and synthesizes it by QMF and outputs it as an audio signal
It is composed of an MF synthesizer 8. In this embodiment, QM
Instead of the signal that is the high frequency component of the baseband signal that is the output signal of the F analyzer 2, the FM modulated synchronization signal x (t)
, And the sync signal is scrambled with the audio signal, but the sync signal may be transmitted and received by a conventional method.

An embodiment of generation and phase estimation of the FM-modulated synchronizing signal x (t) will be described below. For modulation / demodulation and phase estimation using the FM of the synchronization signal, the application “reference signal generator” by the same inventor and the same applicant (1993)
Application dated September 2, 2012). FIG. 3 shows a block diagram of the transmitting side. The output terminal of the integrator 26 that receives the synchronization signal is connected to the input terminal of the modulation index multiplier 27, and the output terminal of the modulation index multiplier 27 is connected to one input terminal of the carrier wave adder 28. A carrier signal is input to the other input terminal of the carrier adder 28. The output end of the carrier wave adder 28 is connected to the input end of the cosine calculator 29.
The 29 output terminals are the output terminals for transmitting the FM-modulated synchronizing signal.

FIG. 4 shows a block diagram of the receiving side. F for input
The first quadrature modulator 11 which receives the M-modulated synchronizing signal is composed of multipliers 11a and 11b and a phase delayer 11c,
A local signal is input to the other input of the first quadrature modulator 11. The two output ends of the first quadrature modulator 11 are connected to the two input ends of the first low-pass filter 12, respectively, and the two output ends of the first low-pass filter 12 are respectively connected. It is connected to the two input terminals of the first phase angle calculator 13. The first quadrature modulator 11, the first low-pass filter 12 and the first phase angle calculator 13 constitute a calculating means 22. The output end of the first phase angle calculator 13 is connected to the input end of a differentiator 14, and the output end of the differentiator 14 is a multiplier.
The second quadrature modulator 15 is connected to one input end of the second quadrature modulator 15 and includes a comparison signal generator 21 to be described later. The output end of is connected. The two output terminals of the second quadrature modulation section 15 are connected to the two input terminals of the second low-pass filter section 16, and the two output terminals of the second low-pass filter section 16 are respectively connected. It is connected to the two inputs of the integrator 17. The second quadrature modulator 15 and the second low-pass filter 16 are quadrature demodulators.
Make up 24. A control signal for changing a time constant, which will be described later, is also input from the controller 19 to the integrator 17. The two output terminals of the integrator 17 are respectively connected to the two input terminals of the second phase angle calculator 18, and the output terminal of the second phase angle calculator 18 is the input of the controller 19 and the loop filter 20. It is connected to the. The controller 19 and the loop filter 20 form a controller 25. The controller
The two outputs of 19 are connected to the integrator 17 and the inputs of the loop filter 20, respectively. The output terminal of the loop filter 20 is connected to one input terminal of the comparison signal generator 21 composed of an adder 21a and a cosine calculator 21b. The comparison signal generator
A signal of a product of the carrier frequency ωc and time is input to the other input terminal of 21. The comparison signal generator
One of the two output ends of 21 is an output end that outputs the desired reference signal, and the other is the second quadrature modulator 15 as described above.
Is connected to the input end of. The second quadrature modulator 15,
Second low-pass filter 16, integrator 17, second phase angle calculator
18, controller 19, loop filter 20, and comparison signal generator 21
Constitutes a signal generator 23.

Next, the operation of the embodiment will be described. On the transmission side, the synchronization signal cosωm t is input to the integrator 26, integrated, and output to the modulation index multiplier 27. The modulation index multiplier 27 multiplies the integrated signal by the modulation index and sends it to the carrier adder 28. The carrier adder 28 adds the product of the carrier frequency ωc and the time t to the signal multiplied by the modulation index. A cosine calculator 29 performs a cosine calculation using the output signal of the carrier adder 28 as a phase signal, and the sync signal cosωm t is F
It is sent to the receiving side as an M-modulated synchronizing signal x (t).

On the receiving side, the FM modulated sync signal x
In response to (t), the first quadrature modulator 11 performs quadrature modulation with the signal cos ωct of the local frequency ωc to obtain two signals of the in-phase and quadrature components of the base region of the synchronization signal. The in-phase component and the quadrature component of the two signals are the first low-pass filter 12
Is low-pass filtered by and is input to the first phase angle calculator 13. The first phase angle calculator 13 calculates the phase angle from both signals. The phase angle is time-differentiated by the differentiator 14 and sent to the second quadrature modulator 15.

The second quadrature modulator 15 includes a comparison signal generator 21.
The signal from which the phase angle is time-differentiated is quadrature-modulated as a local signal, and two signals of the in-phase component and the quadrature component are obtained.
Outputs two signals. The second low-pass filter 16 receives the two signals, performs low-pass filtering, and outputs the respective signals.
Both signals are input to the integrator 17. The integrator 17 increases the time constant with time when synchronization is achieved by the control signal from the controller 19, and performs integration. The two signals of the in-phase component and the quadrature component that have been integrated are input to the second phase angle calculator 18, and the phase angle thereof is calculated. The signal of the phase angle is inputted to the controller 19 and the loop filter 20, and the controller 19
Outputs a control signal when the phase angle becomes substantially zero, which is regarded as synchronized. The control signal is sent to the loop filter 20 and the integrator 17. As described above, the control signal increases the time constant of the integrator 17, and also increases the time constant of the loop filter 20.

The signal of the phase angle input to the loop filter 20 is filtered by the loop filter 20. The signal of the filtered phase angle, that is, the signal representing the phase error is input to the comparison signal generator 21. In the comparison signal generator 21, the signal representing the phase error is added with a value obtained by multiplying the same frequency ωm as the synchronizing signal by time, and the reference signal to be obtained is output, while the signal obtained by performing the cosine calculation on the reference signal is output as the second signal. Quadrature modulator 15
Output to. The signal is transmitted to the second quadrature modulator 15 as described above.
It becomes a local signal of.

The talk is returned to the secret voice signal generator and the secret voice signal receiver. In this embodiment, a QMF structure using a polyphase filter is used to reduce the amount of calculation. FIG. 9 (a) shows a block diagram of thinning using an ordinary FIR filter. In this case, one input sample needs to be multiplied by N + 1 times, and only an even-numbered sample is needed for output decimation, and an odd-numbered sample is not needed. Therefore, if the output is performed only when the switch falls to the polyphase filter 1 using a polyphase filter in which the filter coefficients are divided into even-order and odd-order as shown in FIG. Can be halved in FIG. 9 (a). LPF impulse response is hL
(n) and the impulse response of HPF is hH (n), QM
The components of the F analysis unit can be represented as shown in FIG. 10 (a). Here, the cutoff frequency of LPF and HPF is π.
Since it is / 2, the relationship of Equation 1 is established.

[0017]

[Equation 1]

The impulse response of the polyphase filter 1 when using the LPF is P1 (n), the impulse response of the polyphase filter 2 is P2 (n), and the impulse response of each when using the HPF is Q1 ( If the relationship of Equation 1 is used as n) and Q2 (n), it can be rewritten as shown in FIG. 10 (b). In this figure, the polyphase filter is P1 (n) and P2.
Since there are two types, (n), these can be summarized as shown in Fig. 10 (c), and the number of filters can be two. Therefore, by using the polyphase filter in the component of the analysis unit of the QMF, the number of multiplications is halved, and the number of filters is also halved, so that it is possible to realize with a calculation amount of 1/4 as a whole.

Similarly, the efficiency improvement of the components of the QMF synthesis section will be described below. FIG. 11 shows a block diagram of interpolation using an ordinary FIR filter. In Fig. 11 (a), zeros are inserted between samples in the part where oversampling is performed twice, and in subsequent filters, zeros are stored in the buffer every other sample, and the filter coefficient and this buffer are stored. It is very inefficient because it requires multiplication by zero. Here, using the same polyphase filter used for thinning out, the above problem can be solved by doing as shown in Fig. 11 (b), the calculation of multiplication is halved, and the buffer is also reduced. You only need one. The constituent elements of the synthesizing part of the QMF can be expressed as shown in FIG. 12 (a), and can be expressed as shown in FIG. 12 (b) by using a polyphase filter. The output of FIG. 12 (b) is the polyphase filter P1.
(n) and -P1 (n), and P2 (n) and -P2 (n) are added together. Since the system in this figure is linear, instead of adding at the output of the polyphase filter,
As shown in (c), the result will be the same even if addition is performed at the input end of the polyphase filter. Therefore, the total calculation amount in the combining unit as well as the analyzing unit can be reduced to 1/4.

As described above, when the QMF is used, a general DSP has a powerful instruction for shifting and summing products in one step for realizing an FIR filter, and it is possible to reduce the number of instructions and easily make a macro. is there. Further, by performing segment replacement including the synchronization signal on the frequency axis using QMF, it is possible to reduce voice intonation and further increase the degree of confidentiality. Then, when performing group communication, it is possible to prevent the synchronization signal from colliding with other groups.

[0021]

According to the present invention, a QMF is used to convert a voice signal into a frequency spectrum and divide it into bands of a predetermined width, and a sync signal is used as a baseband signal of the divided bands. , It is decided to inject in place of the baseband signal of the high frequency band with a relatively small amount of information, so that the frequency spectrum division interval can be changed and divided into a band with a large amount of information of the audio signal and a band with a small amount of information. Thus, a secret voice signal generator and a secret voice signal receiver which can easily be processed at both ends of the band are obtained. Since the frequency spectrum division interval can be changed between a band with a large amount of information of a voice signal and a band with a small amount of information, if the number of divisions of a band with a large amount of information is increased and the number of divisions of a small band is reduced, the same number of divisions The degree of confidentiality is improved compared to the past.

Further, as shown in FIG. 13, since the folded back baseband signal is obtained, the unfolded baseband signal and the folded back baseband signal are mixed together to improve the confidentiality level. . Furthermore, it is also a secret story that the sync signal is injected instead of the baseband signal of the high frequency band, which has a relatively small amount of information, among the baseband signals of the divided bands, and scrambled together with the audio signals of other bands. It contributes to the improvement of the degree. Since only the characteristics of the LPF and HPF can be changed, the processing can be performed on both ends of the divided band, so that there is no possibility that the sound quality will be deteriorated by the signals leaking to the adjacent two adjacent bands. .

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of a secret voice signal generator according to the present invention.

FIG. 2 is a block diagram of an embodiment of a confidential voice signal receiving apparatus of the present invention.

FIG. 3 is a block diagram of a transmitter side of an FM-modulated synchronizing signal used in the device of the present invention.

FIG. 4 is a block diagram of a receiving side of an FM-modulated synchronizing signal used in the device of the present invention.

FIG. 5 is a block diagram of components of a QMF.

6 is a diagram showing spectra of each part of QMF in FIG. 5, where (a) is the input spectrum, (b) is the S0 (m) spectrum, (c) is the S1 (m) spectrum, and (d). Is a diagram showing a spectrum of Q0 (n), (e) is a spectrum of Q1 (n), and (f) is a diagram showing an output spectrum.

FIG. 7 is a diagram for explaining scrambling on the frequency axis using QMF.

FIG. 8 is a diagram illustrating a method of varying the bandwidth divided by scrambling on the frequency axis using QMF depending on the bandwidth.

9A and 9B are diagrams illustrating efficiency improvement of thinning using an FIR filter, FIG. 9A is a block diagram of thinning using an ordinary FIR filter, and FIG. 9B is a block diagram of improved efficiency.

FIG. 10 is a diagram for explaining efficiency improvement of the QMF analysis unit, (a) is a block diagram of components of the QMF analysis unit, and (b) is a diagram.
Is a block diagram obtained by modifying (a) based on Equation 1, and (c) is a streamlined block diagram.

11A and 11B are diagrams illustrating efficiency improvement of interpolation using a FIR filter, FIG. 11A is a normal block diagram, and FIG. 11B is a block diagram with improved efficiency.

[Fig. 12] Fig. 12 is a diagram for explaining efficiency improvement of the QMF synthesizing unit, (a) is a block diagram of components of the QMF synthesizing unit, and (b).
Is a modified block diagram of (a), and (c) is a streamlined block diagram.

13A and 13B are diagrams illustrating frequency spectrum division and scrambling in the device of the present invention, where FIG. 13A is a diagram showing before scrambling and FIG. 13B is a diagram showing after scrambling.

FIG. 14 is a diagram showing an execution process of scrambling on a frequency axis in a conventional confidential voice signal generation device and a confidential voice signal reception device, (a) is a diagram showing an execution process of scrambling,
(b) is a diagram showing a descrambling execution process.

15A and 15B are diagrams illustrating division and scrambling of a frequency spectrum in a conventional device, FIG. 15A is a diagram showing before scrambling, and FIG. 15B is a diagram showing after scrambling.

[Explanation of symbols]

 1 encryption key converter 2 QMF analyzer 3 shuffler 4 QMF combiner 5 encryption key converter 6 QMF analyzer 7 shuffler 8 QMF combiner 11 first quadrature modulator 12 first low-pass filter 13 13th 1 phase angle calculator 14 differentiator 15 second quadrature modulator 16 second low-pass filter 17 integrator 18 second phase angle calculator 19 controller 20 loop filter 21 comparison signal generator 22 calculating means 23 Signal generator 24 Quadrature demodulator 25 Control unit 26 Integrator 27 Modulation index multiplier 28 Carrier adder 29 Cosine calculator

Claims (3)

[Claims] 1. A secret voice signal generator for scrambling a voice signal on a frequency axis using a QMF, wherein the scrambling method on the frequency axis is determined based on input key number data. Cryptographic key converter (1)
And dividing the voice signal into a predetermined band and obtaining a base band signal of each of the divided bands by QMF.
An MF analyzer (2), an interchanger (3) that interchanges and outputs baseband signals of respective bands output from the QMF analyzer based on the signal from the encryption key converter, and the interchanger And a QMF synthesizer (4) for receiving the signal from the QMF and synthesizing and outputting it by a QMF. 2. The audio signal is an audio signal scrambled on the time axis, and the interchanger (3) is FM-modulated synchronization serving as a reference for applying the scramble on the time axis. Upon receiving the signal, the synchronization signal is injected instead of the baseband signal of the high frequency component in the baseband signal of each band input from the QMF analyzer (2), and the encryption key converter (1 2. The secret voice signal generator according to claim 1, wherein the respective signals are exchanged based on the signals from the above. 3. A confidential voice signal receiving apparatus for receiving a confidential voice signal obtained by scrambling a voice signal on a frequency axis using QMF, which corresponds to a key number used when scrambling the voice signal. An encryption key converter (1) that determines how to descramble the frequency axis based on the data of the input key number, and receives the secret voice signal and divides the signal into predetermined bands. At the same time, a QMF analyzer (2) that obtains the divided baseband signals of respective bands by QMF, and the baseband signals of the respective bands output from the QMF analyzer are converted into signals from the encryption key converter. And a QMF synthesizer (3) for receiving the signal from the interchanger and outputting the audio signal substantially by combining with the interchanger (3). )
And a secret voice signal receiving device.