JPH0529175B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a confidential communication device, and more particularly to a confidential communication device that performs confidential communication by transmitting and receiving scrambled output obtained by swapping the spectra of audio signals.

[Prior Art] There are two types of conventional secret communication devices of this type: those that perform scrambling in the time domain and those that perform scrambling in the frequency domain.

Devices that perform scrambling in the time domain include (1) a device that divides the audio signal into time blocks and rearranges the order of sample values within each block;
(2) Devices that swap the polarity of sample values are typical. Devices that perform scrambling in the frequency domain include (3) devices that use the frequency inversion method, and (4) devices that convert audio signals on the frequency axis. (5) A device that divides the frequency into multiple frequency slots and swaps these slots according to predetermined rules or inverts the frequency within the same slot; There are devices that can be replaced from time to time.

Among the various devices mentioned above, devices that perform scrambling in the time domain shown in (1) and (2) produce white noise in the scrambled output when the frequency of order shuffling and polarity reversal is increased to improve confidentiality. There is a problem in that clarity deteriorates due to the frequency characteristics of the transmission path as the number of transmission lines increases.

On the other hand, for devices that perform frequency domain scrambling shown in (3), (4), and (5), the frequency inversion method in (3) can be easily decoded, and the method in (4) can improve confidentiality. This requires a large number of filters with sharp frequency characteristics. Furthermore, although (5) provides a signal that is more difficult to decipher than (4), since the time response of the narrowband filter is slow, there is a problem in that the switching noise increases.

In order to eliminate the disadvantages of (1) to (5) above, an invention based on the scrambling concept described in (5) above was disclosed in a patent application published in 1986-4021, "Secret transmission device for audio signals." has been done.

Instead of using a narrowband filter, this invention performs spectrum swapping by applying Fast Fourier Transform (FFT) or Fast Hadamard Transform (FHT) to the audio signal, and then adds the inverse transform to the time domain. The receiving side performs FFT or FHT on this signal again to restore the permutation performed on the transmitting side, and then performs an inverse transformation to reproduce the audio signal.

According to the device disclosed in this invention,
Since the number of spectra can be increased simply by increasing the number of points in FFT or FHT, the key for encryption and decryption is It is easy to increase the number, making it difficult to decipher and increase confidentiality, and spectral swapping for scrambling can be easily performed by arithmetic processing. Since it is not necessary to use a narrowband filter with a blunt impulse response, it is said to have various advantages such as reducing noise caused by changing the order of spectrum switching.

[Problem that the invention seeks to solve]

The device of Publication No. 4021/1982, which was invented to eliminate the drawbacks of the devices (1) to (5) above that utilize scrambling processing in the time domain and frequency domain, also has the following problems. be.

In other words, this type of confidential communication device uses orthogonal transformation to form the spectral array of the audio signal to be transmitted, which can be FFT, DFT,
Or FHT. All of these utilize both the real axis component and the imaginary axis component, and therefore form a spectral array accompanied by phase information. As a result, the following two drawbacks are unavoidable.

The first drawback is that it is directly affected by the phase characteristics of the transmission path, resulting in an extremely large audible deterioration of the sound quality of the reconstructed sound on the receiving side.

This means that if there is no scrambling, even if it is affected by the phase characteristics of the transmission path, the effect will be a continuous change and will have almost no audible effect on the reconstructed audio, but if scrambling is performed, the This is because the influence of the phase characteristics of the path appears discontinuously.

Second, the possibility of selecting encryption and decryption keys is low.

This means, for example, that the processing unit of an audio signal is
Assuming 256 points, a complex spectrum of 256/2=128 points is obtained in the case of orthogonal transformation having equal real axis and imaginary axis components in DFT. Therefore, there are at most 128 key selection possibilities! It becomes nothing more than an individual.
Note that the processing unit of 256 points is when a frame of 32 m SEC is set with 8 kHz sampling, but since it is desirable that the temporal block length of this degree or less is the processing unit of the audio signal, the 8 kHz, 32 m SEC frame is set.
This is based on consideration of what can be obtained by mSEC.

The purpose of the present invention is to eliminate the above-mentioned drawbacks and fundamentally solve the problem of the phase characteristics of the transmission line by creating a spectral array using a transform means performed only by real axis components as an orthogonal transform. exclude,
An object of the present invention is to provide a confidential communication device that can significantly improve the possibility of key selection.

[Means for solving problems]

The secret speech device of the present invention includes a first orthogonal transformation means for creating a spectral array of the audio signal by subjecting the audio signal to be transmitted to an orthogonal transformation performed using only real axis components, and the first orthogonal transformation means. The first step is to obtain a scrambled output by permuting the spectral array created by according to a preset rule.
a first inverse orthogonal transform means for transmitting to a transmission path a signal obtained by subjecting the scrambled output to an inverse orthogonal transform performed using only real axis components; a second orthogonal transform means for creating a spectral array by performing an orthogonal transform performed only on real axis components; a second spectrum shuffling means for obtaining a scrambled output that has undergone reshuffling to restore the shuffling by the shuffling means; and a scrambled output by the second spectrum shuffling means that is executed using only real axis components. and second inverse orthogonal transform means that performs inverse orthogonal transform to obtain a decoded output of the audio signal.

〔Example〕

Next, the present invention will be explained in detail with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention. The configuration of the embodiment shown in FIG. 1 includes an input terminal 1, an orthogonal transform circuit 3, an inverse orthogonal transform circuit 4, a synthesis circuit 5, a spectrum switching control circuit 6, a timing generation circuit 7, a synchronization signal generation circuit 8, and an output terminal. 9
, input terminal 11, filter 1
2, synchronous signal extraction circuit 13, orthogonal transformation circuit 14,
spectrum switching circuit 15, inverse orthogonal transformation circuit 16,
It consists of a spectrum switching control circuit 17, a timing generation circuit 18, and a receiving side equipped with an output terminal 19, and these sending and receiving sides exchange confidential data via a transmission path 10.

Regarding the transmitting side, the orthogonal transform circuit 2
The spectrum switching circuit 3 and the spectrum switching control circuit 6 that control it are the first spectrum switching means, and the inverse orthogonal transformation circuit 4 is the first inverse orthogonal transformation means. In addition to these, a synthesis circuit 5 and a synchronization signal generation circuit 8 are used to synthesize a synchronization signal for frame identification as a processing block unit of a predetermined time length and the output of the inverse orthogonal transform circuit 4.
are arranged side by side, and the operation of the whole is controlled by a timing signal output from a timing generation circuit 7.

Also, regarding the receiving side, orthogonal transform circuit 1
4 is a second orthogonal transformation means, a spectrum switching circuit 15 controlled by a spectrum switching control circuit 17 is a second spectrum switching means, and an inverse orthogonal transformation circuit 16 is a second inverse orthogonal transformation means. They form conversion means whose operation is controlled by a timing generation circuit 18. The receiving side also includes a synchronizing signal extraction circuit 13 for extracting a synchronizing signal for performing frame synchronization between transmitting and receiving, and a filter 12 for removing the synchronizing signal.

In addition, the synthesis circuit 5 and the synchronization signal generation circuit 8 on the transmission side, and the synchronization signal extraction circuit 1 on the reception side
3 and filter 12 are well-known elements that are indispensable for this type of confidential communication device for frame synchronization between transmission and reception.

Next, the operation of the embodiment shown in FIG. 1 will be specifically explained.

The audio signal inputted from the input terminal 1 is subjected to orthogonal transformation performed only on the real axis components in the orthogonal transformation circuit 2. In this case, the applicable orthogonal transformation is
For example, DCT (Discrete Cosine Transform)
The target is KLT (Karhunen Loeve Transform), but in this embodiment, orthogonal transformation is performed using DCT.

The input audio signal is passed through an LPF (Low Pass Filter) with a predetermined high cutoff frequency and then filtered for 32 m.
Each SEC frame is extracted and sampled at a sampling frequency of 8kHz, which is subjected to DCT and expressed in the real axis domain using cosine. this
According to DCT, the audio waveforms of the cut-out sections every 32mSEC are set in such a way that the audio waveforms are successively arranged one after another in line symmetry with respect to the imaginary axis direction with respect to the target section, and these waveform samples are The waveform of the relevant interval is expressed by a set of arrays of Fourier cosine components, and this DCT, which is an orthogonal transformation, is closely related to human hearing itself and frequency analysis, and from this point of view, it is also effective in reducing quantization noise. Control in the frequency domain is easy, and optimal orthogonalization is performed in the sense of long-term averaging, and the result is supplied to the inverse orthogonal transform circuit 4.

To perform spectrum scrambling in the spectrum switching circuit 3, for example, the input orthogonal transform output is temporarily stored in the input memory, and then stored in a ROM etc. under the control of a switching key and a switching program. It can be easily implemented by storing it in the output memory while executing it and reading it out one after another in frame units.

The inverse orthogonal transform circuit 4 inversely transforms the input scrambled orthogonal transform spectrum, converts it back into time-domain time-series data, and supplies it to the synthesis circuit 5.

The synthesis circuit 5 obtains a synchronization signal for frame identification from the synchronization signal generation circuit 8, synthesizes this with the time domain data obtained from the inverse orthogonal transform circuit 4 in a predetermined format, and sends the synthesized signal to an output terminal as a scrambled output. Output to 9. Each of the above-described processes is controlled by a timing signal output from the timing generation circuit 7.

The scrambled output output to the output terminal 9 is transmitted to the receiving side via the transmission line 10.

In this case, the problem is the influence of the phase characteristics of the transmission line 10 on the scrambled output.

A conventional spectrum conversion process using Fourier transform has two cosine and sine components, that is, a real axis and an imaginary axis component, and therefore, a scrambled output with a phase component is sent to the transmission line 10.

In the case of audio signals, the influence of the phase characteristics of the transmission path 10 without such scrambling is continuous, and such continuous influence hardly causes any auditory problems. However, since this effect is due to scrambling processing, if it is added as a discontinuous effect to the frequency vs. phase characteristics of the audio signal to be transmitted, it will cause an extremely large auditory sound quality deterioration.

Figure 2 is a phase vs. frequency characteristic diagram for explaining the influence of the phase characteristics of the transmission line on a scrambled signal, and Figure 2a is a phase vs. frequency characteristic diagram showing the influence of the transmission line when scrambling is not performed. , FIG. 2b is a phase versus frequency characteristic diagram showing the influence of the transmission path when scrambling is performed.

As shown in Figure 2a, it is assumed that the power level of the audio signal in frame units is flat in this section, and this is expressed by eight spectrum samples from 0 to 7. f
Suppose that it is continuously influenced by the characteristics of .

If scrambling is not performed, the audio signal sent out onto the transmission path through inverse orthogonal transformation will transmit a spectrum with such phase-versus-frequency characteristics. There is no problem if it is received while maintaining continuity, but by scrambling, for example,
As shown in Figure 2, when the spectrum is swapped in sequence from sample 0 to sample 7, sample 1 to sample 3, and sent out to the transmission line, the phase versus frequency characteristic of the transmission line after transmission restoration is as shown in Figure 2b. If the sound is affected discontinuously in this way, the reconstructed sound will have an extremely significant auditory quality deterioration.

Paradoxically speaking, such problems are caused by the fact that audio information is transmitted including phase information through the transmission path during scrambled transmission and reception, and therefore, normal
Fourier transform using FFT or DFT includes real axis and imaginary axis components, which is an unavoidable problem when transmitting phase information as well.

Furthermore, given the characteristics of audio, the frame length for which the section is considered stationary is at most 30 m.
SEC or less is desirable, and taking this background into consideration, the frame length is set to 32mSEC in this example as well. However, in the case of 8kHz sampling, orthogonal transformation expressed by both real and imaginary axis components like DFT requires 1 of 256 samples. /2 128 complex spectra are found, and the key selection possibilities are correspondingly 128! There also arises the problem that the number of key selections is limited.

The purpose of the present invention is to overcome the above-mentioned two problems, in that it eliminates the transmission of phase information by orthogonal transformation using only real axis components, and also increases the possibility of key selection. It has these basic characteristics.

The receiving side will be explained below. The synchronization signal extraction circuit 13 receives the scrambled output transmitted from the transmitting side via the input terminal 11, extracts the synchronization signal contained therein, and removes this synchronization signal with the filter 12 to maintain frame synchronization. In this state, the scrambled output is supplied to the orthogonal change circuit 14.

The orthogonal change circuit 14 receives the time domain data of the inverse orthogonal transform state performed on the transmitting side and performs orthogonal transform using DCT to obtain an orthogonally transformed spectrum having only real axis components. This frequency domain orthogonal transformed spectrum is then restored by the spectrum switching circuit 15 after the spectrum switching performed on the transmitting side.

Restoration of spectrum swapping is performed under the control of spectrum swapping control circuit 17, and spectrum swapping circuit 15 and spectrum swapping control circuit 17
have almost the same configuration as the spectrum switching circuit 3 and spectrum switching control circuit 6 on the transmitting side, respectively, and perform spectrum scrambling on the transmitting side based on decoding key information corresponding to encoding key information on the transmitting side. After performing re-exchange for spectrum descrambling to restore the spectrum before exchanging, it is supplied to the inverse orthogonal transform circuit 16 one after another in frame units.

The inverse orthogonal transform circuit 16 performs inverse orthogonal transform on the descrambled frequency domain samples that are input one after another in frame units, converts them into time domain data, and outputs the data through a low-pass filter or the like as restored audio output to an output terminal. Output to 19.

The timing generation circuit 18 controls the operation on the receiving side described above.

In this way, confidential communication is performed using only the real axis component of the Fourier transform using DCT.

In this secret communication, there are 256 possible key selections that can be used for encoding and decoding! and this is
This is ₂₅₆ π ⁱ⁼¹²⁹ i times that of the conventional method using DFT, which significantly increases the number.

The above embodiment uses DCT as an orthogonal transformation that uses only the real axis component, but this
Other real axis component orthogonal transformation means such as KLT may be used; in the case of KLT, the orthogonal transformation circuits 2 and 14 and the inverse orthogonal transformation circuits 4 and 16 shown in FIG.
It is clear that this can be easily implemented if configured to perform KLT and inverse KLT conversion operations, respectively.

〔Effect of the invention〕

As explained above, according to the present invention, there is provided means for performing orthogonal transformation on the audio signal to be transmitted using only the real axis components, and then spectrum scrambling and transmitting the resultant signal from the transmitting side to the receiving side. This has the effect that it is possible to realize a secret communication device that can fundamentally eliminate the influence of the phase characteristics of the transmission path and significantly increase the possibility of selecting encryption/decryption keys.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention;
FIG. 2a is a phase vs. frequency characteristic diagram showing the influence of the transmission path when scrambling is not performed, and FIG. 2b is a phase vs. frequency characteristic diagram showing the influence of the transmission path when scrambling is performed. 1...Input terminal, 2...Orthogonal conversion circuit, 3...
Spectrum switching circuit, 4... Inverse orthogonal transformation circuit, 5
...Composition circuit, 6...Spectrum switching control circuit,
7... Timing generation circuit, 8... Synchronization signal generation circuit, 9... Output terminal, 10... Transmission line, 11...
...Input terminal, 12...Filter, 13...Synchronization signal extraction circuit, 14...Orthogonal transformation circuit, 15...Spectrum switching circuit, 16...Inverse orthogonal transformation circuit, 1
7... Spectrum switching control circuit, 18... Timing generation circuit, 19... Output terminal.

[Claims]

1. The transmitting side includes a first pseudo-random signal generating means that generates a first pseudo-random signal at a predetermined code period; Performs low-pass filtering by moving in the positive or negative direction,
SSB signal generation means for selectively outputting one of two single sideband (SSB) signal spectra having a mirror image relationship according to the code polarity; a constant signal generated in synchronization with the code period of the first pseudo-random signal; pilot signal generating means for generating a pilot signal by digitally modulating a signal of a predetermined frequency with a code string signal of a pattern; a first frequency conversion means for converting the received signal into the baseband signal; and a second frequency conversion means for converting the received signal into the baseband signal; the receiving side transmits upon receiving the output of the second frequency conversion means demodulation means for reproducing the operating clock of the side and demodulating the pilot signal; code synchronization control means for detecting the code string signal of the constant pattern from the output of the demodulation means and outputting a synchronization signal; In response to the synchronization signal, a second pseudo-random signal having the same content as the first pseudo-random signal is generated.
a second pseudo-random signal generating means for generating a pseudo-random signal; a signal spectrum obtained by low-pass filtering the output of the second frequency converting means;