JPS6244449B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to an error correction circuit that removes digital signal errors caused during demodulation when an analog signal is converted into a digital signal and transmitted. The public corporation's PCM-
24 system, INTELSAT's SPADE, etc. have been put into practical use, and voice digital communication is being introduced into the field of mobile radio where transmission path characteristics are time-varying.
In the future, there is a tendency for audio digital transmission to be carried out through transmission lines with increasingly poor conditions. In such an environment, code errors in transmission signals are likely to occur too frequently to be ignored, and in order to ensure a certain level of speech quality, it is necessary to eliminate code errors that occur in some form. Conventional code error removal techniques generally include error correction and the use of an error correction code that adds a redundant signal for error correction to a transmitted digital signal. Although this method is logically sound and its effects are certain, it also has some disadvantages, such as adding a redundant signal, which increases the transmission speed by that amount, or conversely reduces the quality of the audio signal. This is because this technique does not utilize any statistical properties of the original audio signal. The audio frequency component of telephone lines is band-limited to 0.3KHz to 3.4KHz. The international sampling frequency for converting this into a digital signal is 8KHz. A signal sampled at 8KHz has a maximum of 0Hz due to the sampling theorem.
It has the ability to transmit a wide range of frequency components up to ~4KHz. Therefore 0Hz~0.3KHz and 3.4KHz~4K
It can be seen that the total 0.9KHz band of each component with Hz is originally a redundant component. The purpose of this invention is to use the aforementioned 0.9KHz redundant band for code error correction to correct and eliminate code errors in a normal digital audio signal that has not been modified at all on the transmitter side only on the receiver side. The goal is to provide the following. In a digital transmission system that transmits the sample value time series of a sampled transmitted analog signal in the form of a digital signal, (a) a demodulation buffer that demodulates the original reproduced sampled time series from the digitally transmitted received signal and a demodulation buffer output from the demodulation buffer output; (b) an error candidate detector that outputs an error candidate detection output when a digitally transmitted received signal subjected to disturbance is detected near the signal identification boundary line; c) An error amount detection filter whose pass band is outside the specified band of the transmitted analog signal, which demodulates components outside the specified band generated due to demodulation errors from the demodulation buffer output or the output of the digital-to-analog converter. an error amount detection filter that detects the amount of error; (d) according to the output of the error amount detection filter, and only when the error candidate detector outputs an error candidate detection output, either the demodulation buffer output or the digital-to-analog converter output; An error correction circuit consisting of an error removal circuit that corrects one of the two is obtained. According to this invention, not only one channel of telephone line but also many telephone channels can be grouped together.
In systems that digitize and transmit analog signals whose components fall within a certain frequency band, such as FDM signals, the redundant band is used on the receiving side to remove noise caused by code errors. can do. And since this removal technique does not require any changes on the sending side, it is
It can be directly applied to received signals such as PCM-24 system, satellite communication SPADE, and TDMA system. Next, the present invention will be explained in detail with reference to the drawings. FIG. 1 is a block diagram of a transmitting section of a conventional audio digital transmission device, and FIG. 2 shows waveforms of each section in FIG. Figure 1 shows microphone 1
The sound is converted into an electrical signal, and the filter 2 whose characteristics are shown in Figure 6 converts the sound into an electrical signal of 0.3KHz to 3.4KHz.
The spectrum is shaped within the band. Figure 2 10
The filter output as shown at 0 is sampled at 8KHz by the sampling circuit 3, resulting in a form as shown at 101 in FIG. The amplitude value Si of this signal is determined by an analog-to-digital converter (A/D converter) as shown in Fig. 2.
The sample value time series is converted into a 7-bit digital representation such as 02. i.e.

There are 7 bits m(i) ₇ to m (i) ₁ that satisfy the formula. These 7-bit parallel signals are converted into parallel signals by the parallel/serial converter 5 as shown in FIG.
is converted into a serial signal such as At this time, in order to preserve the original 7-bit division, 1 bit "Sy" for word synchronization is added before m(i) ₁ , and 8 bits become one signal unit. This signal is further transferred to the converter buffer memory 6.
The signal is input into the signal and converted into a parallel signal of 2 bits at a time as shown in FIG. 2 104. This signal is converted into a four-phase modulated wave by a four-phase modulator 7. FIG. 2 105 shows the modulated wave divided into an in-phase component I and a quadrature component Q. The modulated wave is sent out from the antenna 8. On the other hand, on the receiving side, as an example, signals as shown in FIG. 2 105 are received one after another, but the signal points are disturbed by reception noise. FIG. 3 shows that the received signal 30 has noise n
It shows how it moves to point 31 due to _I 32 and n _Q 33. If point 31 is demodulated, a code error will obviously occur. It is known that n _I and n _Q are generally distributed in a Gaussian manner around the received signal 30, and most of the points 31 that cause code errors as shown in FIG. Occurs nearby. Putting this in reverse, signal identification boundary line 3
It can be said that the signals occurring in the vicinity of 4 and 35 are error candidates that have the possibility of code errors caused by noise and other disturbances. FIG. 4 is a diagram showing the principle of the error candidate detector for the four-phase modulated wave described above, and an error candidate detection output is output when a received signal occurs in a region 40 near the signal identification boundary line. FIG. 5 is a block diagram of the receiving section of the digital audio transmission apparatus of FIG. 1, and its operation is the same as that of the block of FIG. 1 in reverse order. Therefore, each part waveform 104, 103, 102, 1
01 is equal to that in FIG. antenna 1
The received signal from 6 is returned to a parallel signal of 2 bits each by a 4-phase modulation detector 9, which will be explained later.
A converter buffer 10 that converts the waveform shown in the figure from 104 to 103 and the waveform 103 to 10 in FIG.
A demodulation buffer consisting of a serial/parallel converter 11 converts the data into 7 bits each, and the digital/analog converter 12 converts the sample value back to an analog signal. Then, through an ideal low-pass filter 14 with a cutoff of 4KHz, the original correct sound is restored and output as sound from the speaker 15. Here, the frequency components of the output of the digital-to-analog converter 12 are, of course, only those between 0.3 KHz and 3.4 KHz as shown in FIG. 6, unless a code error occurs. Note that T shown in the figure indicates the sampling period. FIG. 7 shows the waveform 70 of the demodulated analog signal and its frequency components when a code error occurs. The waveform 70 can be considered as the sum of the original waveform 70 and a rising wave 72 due to a code error. Then, the former takes the form of a band-limited spectrum 73, and the latter takes the form of a spectrum 74 that extends over the entire band characteristic of impulsive noise. From this figure, the occurrence of numbering errors and their magnitude are as follows:
It can be estimated based on the occurrence and magnitude of the spectrum 74 in the figure. For this, spectrum 74
Frequency characteristics 8 as shown in Fig. 8 that only appear selectively
Filters with 0 are valid. Such a filter will hereinafter be referred to as an error amount detection filter. Here, it is possible to estimate the time when a code error occurs in this filter as well, but because the filter band is narrow compared to the transmission band, the output waveform is very smooth and it is difficult to detect the peak. Not used for time detection. FIGS. 9, 10, and 11 are block diagrams for explaining the components and operations of one embodiment of the present invention. Figure 9 shows a detector 9 for four-phase modulated waves.
It uses the most common synchronous detection method in 000, and includes multipliers 91 and 92, π/2 phase shifter 93, reference carrier wave 94, filters 98 and 99, and A/D converter 9.
5, 96, a 32KHz clock 97 connected to a terminal 902, and the amplitude values of the in-phase component and quadrature component of the modulated wave input from the input terminal 90 are output to the output terminal 90.
It is output as a digital code from the 0,901 terminal. FIG. 10 shows the error candidate detector 8000 described above. Figure 11 7020, 70
Received signal points such as 21, 7022, and 7023 are expressed as digital values of both in-phase and quadrature components by a detector 9000 as shown in FIG.
The signals enter input terminals 900 and 901 in FIG. This signal is used for read-only memory (Read Only memory).
Memory≡ROM) 74 address. This ROM outputs "1" when both the in-phase and quadrature components corresponding to the area 40 in FIG. 4 are added as addresses. This output becomes an input to a shift register 76, and is sequentially shifted to the right by a 4KHz clock from a terminal 78, as shown at 701 in FIG. 1 sample of audio is shown in Figure 2 1
As can be seen from 04, it corresponds to a set of four signal points.
Therefore, after all four signal points corresponding to one audio sample are input to the ROM 74 by the function of the 4-frequency divider 75, the shift register 76 is reset at the timing 700 in FIG. 11 for the next new audio sample. Ru. The OR circuit 77 detects whether or not an error candidate is included in the bits m(i) ₇ to m(i) ₂ of the received signal. If an error candidate is included, it is recorded by the data latch circuit 80. In addition, in this example, m(i) ₇ ~
Error candidates are detected for bits up to m(i) ₂ , but since m(i) ₁ is the lower bit, even if there is an error, the amount that appears in the output of the digital-to-analog converter 12 is small. , is a bit that is difficult to detect by an error amount detection filter, but it poses no practical problem, so no error candidate for the corresponding bit is detected in this example. Therefore, the necessary number of bits can be selected depending on the device to which it is applied. FIG. 12 is a diagram showing a block diagram of an embodiment of the present invention, and FIG. 13 is a diagram for explaining waveforms and operations at various locations in FIG. 12. 9000 in the figure
is the four-phase modulated wave/detector in Fig. 9, and 5000 is the fifth
Demodulator 800 corresponding to block 5000 in the figure
0 is the error candidate detector 1000 in FIG.
This is an error detection filter with frequency characteristics as shown in the figure. Block 2000 is an error removal circuit. First, as shown in FIG. 13 5001, the shaded area contains noise due to code errors. Demodulator 5
After the output of 000 passes through the delay line 2001, it is delayed and sent to the analog subtracter 20 as shown in FIG.
Enters one input terminal of 02. On the other hand, Fig. 13 8
The output of the error candidate detector 8000 as shown in 001 also passes through the delay line 2004, and then
As shown in FIG. 8002, the signal is input to one input terminal of the multiplier 2003 with a delay. Error amount detection filter 1000
Due to its narrowband nature, the output of
01,2004 and enters the other input terminal of the multiplier 2003 as shown in FIG. 13, 1001. The multiplier 2003 is configured such that the output of the error amount detection filter 1000 is the output of the subtracter 20.
The error candidate detector 8000 serves to gate the input to the other input terminal of 02. Therefore, the output of the multiplier 2003, as shown in 8003 in FIG. 13, has the same shape as the noise component due to the code error in the waveform 5002 in FIG. 13. Yotsute subtractor 20
As shown in FIG. 13 (2004), noise due to code errors is removed from the output of 02. Here, in the embodiment of FIG. 12, the delay line 2001,
By configuring 2004 as a shift register, the subtracter 2002 as a digital subtracter, the multiplier 2003 as a digital multiplier, and the error amount detection filter as a digital filter, the D/A converter in the demodulator 5000 is connected after the output terminal 2004. can be moved to Such a modification allows for the main digitalization of the circuit, making it economical. Although the embodiment shown in FIG. 12 is a case of four-phase phase modulation, the technique of the present invention can of course also be applied to baseband signals of coaxial transmission. If we consider baseband transmission in which binary serial data of +1 and -1 are sent one after another as shown in FIG. 2, block 6000 in FIG.
It becomes the shape of The input/output characteristics of the error candidate detector are such that the area 140 in FIG. 14 is the error candidate detection area. FIG. 15 is a block diagram of an embodiment of the present invention for the ±1 binary baseband transmission. The error amount detection filter 1000 and error removal circuit 2000 in the figure are the same as those in FIG. Demodulator 6000 is identical to block 6000 of FIG. In the embodiment of FIG. 10, the error candidate detector 8000 opens the input terminal 901 and outputs "1" when the address corresponding to the area 140 of FIG. 14 is applied to the input terminal 900 of the ROM 74. What has changed. As explained above, according to the present invention, in contrast to the method of converting an analog signal into a digital signal and transmitting the digital signal,
By adding the error correction circuit of the present invention only to the receiving end of the transmitting side, it is possible to remove impulsive noise superimposed on the demodulated analog signal due to code errors. As explained earlier, this method uses the redundant band of the original analog signal source, so it is characterized in that it does not require any waveform manipulation on the transmitting side.

[Brief explanation of the drawing]

Figure 1 is a diagram showing a block diagram of the transmitting section of a normal audio digital transmission device, Figure 2 is a diagram to explain waveforms at various locations in Figure 1, and Figure 3 is a diagram to explain code errors due to disturbance. FIG. 4 is a diagram showing the error candidate detection area, FIG. 5 is a block diagram of the receiving section of a normal audio digital transmission device, and FIG. 6 is a diagram showing the frequency characteristics of a filter that passes audio on the transmitting side. Figure 7 shows changes in demodulated waveform due to code errors,
FIG. 8 is a diagram showing the frequency characteristics of the error amount detection filter; FIG. 9 is a diagram showing a normal four-phase synchronous detector; FIG. 10 is a block diagram of the error candidate detector; 11 is a diagram for explaining waveforms at various locations in FIG. 10, FIG. 12 is a diagram showing a block diagram of an embodiment of the present invention, FIG. 13 is a diagram for explaining waveforms at various locations in FIG. 12, and FIG. 14 is a diagram for explaining waveforms at various locations in FIG. The figure is ±1 of 2
FIG. 15, a diagram illustrating an error candidate detection area for value baseband transmission, is a diagram showing another embodiment of the present invention. In the figure, 1000 is an error amount detection filter;
2000 is an error removal circuit, 5000 and 6000 are demodulators, 8000 is an error candidate detector, and 9000 is a four-phase modulated wave detector.

Claims (1)

[Claims] 1. In a digital transmission system that transmits a sample value time series of a sampled transmitted analog signal in the form of a digital signal, (a) demodulating the original reproduced sampled time series from the digitally transmitted received signal; a demodulator consisting of a demodulation buffer and a digital-to-analog converter that obtains a demodulated analog signal from the demodulation buffer output; (c) an error amount detection filter whose pass band is outside the specified band of the transmitted analog signal, which detects components outside the specified band caused by demodulation errors as the output of the demodulation buffer or the digital analog signal; an error amount detection filter that detects the amount of demodulation error from the output of the converter; (d) according to the output of the error amount detection filter, and only when the error candidate detector outputs an error candidate detection output; An error correction circuit comprising: an error removal circuit for correcting either one of the outputs of a digital-to-analog converter;