JPS6233783B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for detecting code error noise in digital signals. The present application is
This is an improved method for detecting the position of noise in the "code error noise suppression method" filed on January 29, 1980 (application number 55-008285).

A major feature of these detection methods is that they do not require the transmission of extra signals unlike conventional error detection codes.

First, an overview of digital communication will be described as background information for understanding the present invention. The signal s(t) whose highest frequency is band-limited to W(Hz) is
The sampled signal sequence is sampled at a sampling period T=1/2W (sec) {..., s(t _i ), s(t _i+1 ), s(t _i
+2 ),...}. However, t _{i +1} −t _i =T. These signal sequences are then quantized and binary encoded, then modulated and sent out onto a transmission path.
On the receiving side, the binary code obtained by demodulation is decoded to produce a sampled signal sequence (hereinafter abbreviated as "received signal") {..., r(t _i ), r(t _i+1 ) , r(t _i+
2 ),...}, and set it to the highest frequency W (Hz)
Regenerate the signal s(t) through an ideal low-pass filter. (Hereinafter, {..., r(t _i ), r(t _i+1 ), r
(t _i+2 ),...} is abbreviated as {r(t _i )}). By the way, when noise is added to the transmission path, a code error occurs, and thereby code error noise is added to the received signal {r(t _i )}. At this time, one code error affects only one received signal and has no effect on other received signals, so the code error noise becomes temporally localized impulsive noise. This is essentially different from normal continuous noise (for example, thermal noise), and is a major feature of code error noise. The present invention actively utilizes this property.

FIG. 1 is a flowchart showing the processing process of the first invention.

1 is a time window setting device, which increases the accuracy of the frequency analysis performed in the next step 2.

2 is a discrete Fourier transformer which converts one block of received signals {r(t ₁ ), r(t ₂ ), ..., r(t _N )}
The short-time frequency spectrum A _r (k)・exp(−j
Calculate φ _r (k)), k=0, 1, ..., N-1. Here A _r (k) is the amplitude spectrum A _r ²
(k) is a power spectrum), φ _r (k) is a phase spectrum, and j=√−1. Note that the transmitted signal corresponding to this received signal is {s(t ₁ ), s
(t ₂ ), r...s(t _N )}, and its short-time frequency spectrum is A _s (k)·exp(−jφ _s (k). At this time, the m-th signal in the received signal Assuming that there is code error noise (hereinafter abbreviated as "noise") of amplitude h only in r(t _n ), {s
(t _i )}, {r(t _i )} and {n(t _i )} the following relationship holds true.

Then, the short-time frequency spectrum of the noise {n(t _i )} is A _o (k)・exp(−jφ _o (k)=h・
It is expressed as exp(-j2πmk/N).

Here, A _o (k)=h represents a flat amplitude spectrum, and φ _o (k)=2πmk/N represents a phase spectrum. Next, when Expression (1) is expressed on a frequency spectrum, it becomes Expression (3).

A _r (k)・exp(−jφ _r (k))=A _s (k)・exp(−jφ _s (k))+h・exp(−j2πmk/N) …(3) At this time, the following relationship There is.

A _r (k)={A _s ² (k) + 2A _s (k)・h・cos(φ _s (k)−2πmk/N)+h ² }〓 …(4) Figure 2 (1) shows the phase spectrum φ _s (k) of the transmitted signal consisting of N = 256 samples of the vowel /a/.
(2) shows the phase spectrum φ _r (k) of the received signal obtained by adding noise of m=51 and h=200 to the transmitted signal, and (3) shows the phase spectrum φ _o (k) of the noise, respectively. .

By the way, the information regarding the position m of the noise is the sine wave component sin (2πmk/N) in the numerator and denominator of equation (5),
and is included in the frequency m/N of the cosine wave component cos (2πmk/N). At this time, the phase spectrum φ _r (k) of the received signal and the phase spectrum φ _o of the noise
Calculating the correlation function with (k)=2πmk/N, we get
The result will be as shown in Fig. 3 1. On the other hand, when we calculate the correlation function between the phase spectrum φ _r (k) of the received signal and the phase spectrum φ _s (k) of the transmitted signal, we find that
become that way. As can be seen by comparing Figure 3 1 and 2, the correlation between φ _r (k) and φ _o (k) is φ _r
It can be seen that the correlation between (k) and φ _s (k) is considerably larger. In other words, the phase spectrum φ _r (k) of the received signal contains a large amount of information regarding the position m of the noise. On the other hand, this is not the case with the amplitude spectrum A _r (k) of the received signal. Therefore, by converting A _r (k) to a small constant amplitude value and performing inverse Fourier transform while leaving φ _r (k) as it is, the noise with a sharp peak at the noise position m as shown in Figure 4, 3 can be generated. A waveform similar to the waveform can be obtained.

Therefore, by detecting the position of the peak of the waveform shown in FIG. 4, the position m of the noise can be determined. The transmitted waveform at this time is shown in FIG. 41, and the received waveform with noise added is shown in FIG. 42.

3 is a constant amplitude setter that converts the amplitude spectrum A _r (k) of the received signal into a small constant value (usually 1).

4 is the amplitude spectrum 1 and the phase spectrum φ _r
This is a discrete inverse Fourier transformer that transforms a frequency spectrum with (k) into a time waveform.

5 is an absolute value unit that takes the absolute value of the time waveform obtained in 4.

6 is a position detector for the maximum amplitude value of the absolute value waveform obtained in 5. As can be seen from FIGS. 1 to 3, this position indicates the position of the noise.

Note that the amplitude spectrum of the received signal is A _r (k)
In this case, the power spectrum becomes A _r ² (k), so even by converting the power spectrum into a small constant amplitude value, the same processing as in the case of the amplitude spectrum is possible.

A comparison between the noise detection method described above and the detection method used in the "code error noise suppression method" filed on January 29, 1980 (application number 55-008285) will be described below. The detection method of Application No. 55-008285 required an operation to subtract a constant value from the amplitude spectrum of the received signal, but the current detection method simply converts the amplitude spectrum to a constant value. Often, subtraction processing is no longer necessary.
Therefore, when considering hardware, this method has the advantage of being simpler.

Finally, applications of the detection method of the present invention will be described.

In PCM communication, except when the bit error rate is extremely high (approximately 10 ^-1 ), noise occurs at random intervals, so most signal values are correct. Therefore,
By detecting the position of the noise using the detection method of the present invention and replacing the value of the signal containing the noise with an interpolated value based on the correct signal value before and after the noise, a practically sufficient correction can be made. The code error noise suppression method using this method has a very great advantage in that it does not require the transmission of extra signals other than the voice signal, unlike the conventional method using error correction codes.

[Brief explanation of the drawing]

FIG. 1 is a flowchart, and FIGS. 2, 3, and 4 are line diagrams. DESCRIPTION OF SYMBOLS 1... Time window setter, 2... Discrete Fourier transformer, 3... Constant amplitude setter, 4... Discrete inverse Fourier transformer, 5... Absolute value unit, 6... Position detector for code error noise.

Claims (1)

[Claims] 1. One block of signals consisting of several received signal sequences is converted into a short-time frequency spectrum by discrete Fourier transform, and after converting the amplitude spectrum to a constant value, a discrete inverse Fourier change is performed. 1. A code error noise detection method using frequency spectrum processing, characterized in that the position of the code error noise in the block is detected by detecting the position of the code error noise in the block. 2. A code by frequency spectrum processing according to claim 1, characterized in that after converting the power spectrum to a constant value, a discrete inverse Fourier transform is performed to detect the position of code error noise in the block. Error noise detection method.