KR910000652B1 - Electronic switching dtmf signal receiving method - Google Patents

Abstract

The method comprises a first step for eliminating dial tone utilizing a filter (30) for detecting and discriminating efficient digit tone, a second step for modulating signal sampled by 8KHz descend sampling to 4KHz sampled signal, a third step for dividing sampled signal into high frequency band and low frequency band group digit tone, a fourth step for analyzing autocorelation of the digit tone, and a fifth step for transmitting result of signal detecting by deciding existancy of signal and frequency of signal.

Description

DTMF signal receiving method of electronic exchange

1 is a basic block diagram of a DTMF signal receiver according to the present invention.

2 is a frequency contrast table for the DTMF digit code.

3 is a statistical distribution of partial autocorrelation coefficient values of a speech signal.

4 is a distribution diagram of the polarization coefficient values of eight tones of a DTMF signal and noise signal.

5 is a k ₁ change band according to signal-to-voice ratio (SVR) and k ₁ detection bands for 8 DTMF tones.

6 shows the change in k ₁ according to signal-to-noise ratio (SNR) and k ₁ detection bands for 8 DTMF tones.

7 is a variation of the partial autocorrelation coefficient K ₂ of eight DTMF tones contaminated with noise.

8 is a variation of the partial autocorrelation coefficient K ₂ of eight DTMF tones contaminated with voice signals.

9 is a detailed flowchart of a DTMF signal reception method according to the present invention.

FIG. 10 is a typical embodiment of a timing diagram according to the DTMF signal reception scheme of FIG.

FIG. 11 is a block diagram illustrating an example of a design of a digital DTMF signal receiver implemented by a time domain processing scheme in accordance with the present invention. FIG.

The present invention relates to a DTMF (Dual Tone Multi Frequency) signal reception method of an electronic exchange, and in particular, digit tones used for dialing in a telephone set and pseudo signals such as noise and voice signals sometimes present near frequencies such as these digit tones. The present invention relates to a DTMF signal reception method of a digital electronic exchange that can safely identify valid digit tones among them.

In a switching system, when a telephone set used by ordinary subscribers transmits digits through the key pad, they are transmitted from subscriber telephone sets as they use DTMF signals (700 Hz to 1700 Hz) in the voice band (300 Hz to 3400 Hz) to convey digit information. There is a need for a high performance receiver capable of detecting tonal signals, which include a DTMF tone detector that can specifically meet the following two essential system requirements.

First, the receiver must have high sensitivity enough to detect and identify tone frequencies. Second, the receiver distinguishes effective digit signals between pseudo signals such as noise or voice signals and digit tones that actually carry digit information. You should be able to do it.

Of course, such a receiver may show a slight characteristic difference from the system requirements in the circuit characteristics of the signal detector itself. Therefore, in real practical receiver implementation, there is no choice but to compromise the ideal system requirements. There are several typical ways for detection.

One of them is the digital filter method ("A study on the Digital Push-button Signaling Receiver") by S. Kikuchi, H. Kitano, and J. Matsuda, Review of the ECL. , Vol. 27, Nos. 7-8, pp. 630-641, July-August 1979) use limiters as a way to meet the above system requirements. Based on the fact that it contains a fixed amount of signal energy, a pseudo signal containing multiple frequency components does not have enough energy to exceed a predetermined threshold in one frequency channel, but is effective digits. The tones are aggregated at all frequencies of interest, so they cross the threshold immediately, and a voice filter is added to pre-adjust the signal on the voice channel. Although the function to neutralize the receiver (Disable), the effective digit neotgido tone signal creating a level under too sensitive, if exceeds the level, and also disable the voice channel, causes the receiver to receiver due to noise.

Another method is the Discrete Fourier Transform (DFT) method, which is a frequency conversion method for a received signal, which is a second harmonic component for eight tone frequencies of DTMF signals to satisfy the above system requirements. ("Add DTMF Generation and Decoding to Dsp-μP Designs" by P. Mock, EDN, pp. 205-220, March 21, 1985).

This type of operation is based on the fact that the second harmonic component can detect the energy of a relatively large speech signal.

In other words, even though the voice signal is recognized as valid digit tones because it causes digit simulation (noise or voice signal to look like a DTMF signal), it is regarded as a malfunction due to voice if the second harmonic component is larger than a preset threshold. That's the way it is.

The present invention is completely different from the DTMF signal reception schemes described above, and calculates autocorrelation coefficients from received signals and based on these, the well-defined parameter, Partial Autocorrelation (PARCOR) coefficient value ( "PARCOR Speech Analysis-Synthesis System" by N. Kitawaki, K. Itoh, and F. Itakura, Review of the ECL, Vol. 26, Nos. 11-12, pp. 1140-1455, November-Oecember, 1978). By identifying the presence or absence of the tone frequency and determining the effective digit, the partial autocorrelation values of the effective digit tones are distinct from the partial autocorrelation values of the pseudo signals such as noise and voice signals. This is based on the fact that the coefficient values of the digit tones, which are very stable even when contaminated by pseudo signals and which have a sufficient difference from each other, are unique values. So that the various characteristics can remove some of the restrictions, which can suffer from the existing receiving method of the one effectively applicable by making for an optimal determination of the digit detection and tone frequency valid digit.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 shows the basic elements of a DTMF signal receiver according to the present invention, which passes the band-separated sample signal in the falling sampler 20 through a low pass filter limited to 3.4 kHz and then for every two samples. The sample signal sampled at 8 kHz is sampled down by the sample signal sampled at 4 kHz, and then the sample passed through the frequency group separation filter 21 {x (n)}, (n = 1, 2, ..., N- From 1), the autocorrelation coefficients {r ₀ , r ₁ , r ₂ } in the autocorrelation calculator 22 are calculated as follows.

Where N is 48 or 52.

This is because it is suitable for the reception method according to the present invention, which will be described later.

On the other hand, the partial autocorrelation coefficient k ₁ , k ₂ } in the autocorrelation calculator 23 is calculated by the following equation based on the autocorrelation coefficients r ₀ , r ₁ , r ₂ obtained above.

Here, the autocorrelation coefficient {k ₁ , k ₂ } ("A Study on the Feasibility of PARCOR-type Receiver in DTMF Signaling" by MGKyeong, JIBaek, HHLee, Proc. Of 1987 IEEE Region 10 Conference, Vol. 3 of 3 , pp.38.2.1-38.2.5, August 1987), and according to the receiver of the present invention, at the rear end of the low-band frequency group separation filter 37 of FIG. 11, {k ₁ (L), k ₂ (L)} and {k ₁ (H), K ₂ (H)} at the rear end of the high-band frequency group separation filter 34, respectively.

However, one frequency group hayeoteumeuro receiving system of the present invention is such that when actually using any of the k ₂ (L) and k ₂ (H) for the signal detection with respect to k ₂ in accordance with the definition formula of the k ₂ {r ₁ , r ₂ } does not need to be calculated.

When the analysis of the partial autocorrelation coefficients is completed, {k ₁ (L), K ₁ (H)} ² _{1 = 1} analyzed by the signal detection and effective digit determination circuit 24 is a comparison process based on predetermined thresholds. If the signal is detected and the signal is detected, the detected digit is read through the signal detection process of reading the digit information from the k ₁ detection band contrast table for the eight DTMF tones shown in FIG. 5 or FIG. Lastly, the tone is tested to satisfy the system requirements for the minimum duration of the DTMF signal and the interdigit period between the digits. go through the digit determination process where _{{S 0, S 1, S} 2} of the signal detection and the effective digit determination circuit 24 for the presence or absence of a signal obtained by the signal detection process, The signal member (S ₀₎ signal is present, respectively, as state variables for transmitting information (S ₁₎ is determined on hold denotes the (S ₂₎ the determination hold (S ₂₎ is a signal by come in the whole body of the period in which a signal is continued twisting moment signal This is to compensate for the occurrence of temporary malfunction during detection.

The operation of the receiver should be designed so that the tone frequency is detected and the effective digit determination is made effectively based on the fact that the coefficient values of the eight effective digit tones are each unique as described above.

On the other hand, the partial autocorrelation coefficient changes in the level of the input signal because the value of the amplitude level of the signal is offset from the definitions of k ₁ and k ₂ regardless of the amplitude level of the input frequency tone. Regardless of whether the same detectability (Detectability) is always maintained.

Figure 2 is a frequency comparison table for DTMF digit codes used for dialing in a telephone set by 16 combinations generated from four low-band frequency tones and four high-band frequency tones used in DTMF signaling. The digit code is displayed.

The detection of DTMF signals is performed by calculation of the partial autocorrelation coefficients k ₁ and k _2. The distribution of the partial autocorrelation coefficients for the noise (male and female voice) signals and noise signals that can be mixed with the DTMF signal is determined. Figure 3 shows the distribution of partial autocorrelation coefficients for male and female voice signals, and Figure 4 shows the distribution of partial autocorrelation coefficients for noise signals and eight DTMF signal frequencies.

3 is a graph showing the statistical distribution of the partial autocorrelation coefficients of a speech signal, where the k ₁ coefficient is extremely sharply distributed near +0.1 and the k ₂ coefficient is conversely near -0.9. It can be seen that the distribution of the autocorrelation coefficient closer to +1.0, which is the voice signal of the male having a lower frequency component than the voice.

4 is a graph showing the distribution of the values of the partial autocorrelation coefficients k ₁ and k ₂ for the noise signal and the eight tone signals, and is also well isolated from the partial autocorrelation coefficients for the noise signal. The k ₁ values for each tone frequency not only have a constant value, but also maintain a constant value even in voice and noise.

Therefore, only the k ₁ and k ₂ autocorrelation coefficients that the received signal is calculated through the partial autocorrelation analysis process can effectively distinguish and identify effective digit tone signals from the pseudo signals, and detect noise. In this case, as shown in FIG. 4, the pseudo signal determination can be made perfectly with only the k ₂ value, and even when the voice signal is input, the effective digit tone and the pseudo signal can be safely removed from the threshold for the predetermined high frequency digit tones. Identification can be distinguished.

5 shows the change in the k ₁ coefficient and the detection band for each corresponding frequency tone under the condition that the digit tone signals are contaminated by external and internal receiver noise and the tone signal is shifted by ± 2%. It can be seen from these coefficients that the k ₁ coefficients remain almost constant with no significant change in noise and that the coefficient values for each of the corresponding frequency tones are sufficiently well isolated. At the time of signal detection, perfect frequency decoding of the detected signal is possible.

FIG. 6 illustrates k ₁ under the condition that the tone signal is shifted to ± 2% when the voice signal is added to the tone signal by a crosstalk phenomenon that may be induced between adjacent telephone lines while the digit tone signals are transmitted. It provides a change in coefficient and a corresponding detection band for each corresponding frequency tone, and it can be seen that the k ₁ coefficient has a constant value without being greatly influenced by the change in voice voltage magnitude as in the case of FIG. It can be seen that the coefficient values for each of the corresponding tone frequencies are sufficiently well isolated, so that a perfect frequency decoding can be obtained for signal detection from these values.

FIG. 7 shows the change in k ₂ coefficient values for eight tone frequencies under the condition that the tone signals are contaminated by external and internal noise of the receiver while the tone signals are being transmitted and the rated frequency of the tone signal is shifted to ± 2%. It is possible that the value of k ₂ coefficient varies somewhat due to the error of the rated frequency, but it is changing below a certain threshold and it is clearly different from the k ₂ value of noise, so it can be a useful parameter to identify noise and digit tones first. in case Al can and in fact the noise to k ₂ coefficient value is the minimum value of noise, as the fourth compared help in claim 7 also has with larger size then k ₂ than a threshold value of a digit tones shown in the seventh degree Full noise immunity can be provided.

8 shows eight tones under the condition that the rated frequency of the tone signal is shifted to ± 2% and received when a voice signal is applied on the tone signal due to crosstalk that may be induced between adjacent telephone lines while the tone signals are being transmitted. This figure shows the change of k ₂ coefficient values with respect to frequency, and it can be seen that k ₂ coefficient values are also slightly changed due to the error of rated frequency but are changing below a certain threshold, which is a useful parameter for analyzing and identifying voice signals. In the case of negative, the k ₂ coefficient value can also provide a significant immunity to the negative because most of the k ₂ coefficient values are above the threshold of k ₂ for the tone signal shown in FIG. Can be used as a parameter.

In the reception method according to the present invention, the k ₂ threshold is set slightly higher so that the signal can be safely detected without missing a tone signal, and the k ₁ thresholds are held very strictly to falsely detect a pseudo signal as a tone signal when the signal is detected. Try to reduce the probability.

9 is a detailed flowchart of a digital DTMF signal receiving method according to the present invention. The variable IA containing information on a signal-free period between digits and information on whether a digit tone exists for previously received sample data are shown. The variable LA, which contains information on the presence or absence of digit tones for sample data received before variable LA, initializes the variable LA2, removes the dial tone component from the input signal, and selects 4kHz for the output samples of the dial tone rejection filter. The falling sampled sample obtained by the falling sampling is subjected to a preprocessing process that is divided into a high band frequency group and a low band frequency group, and the autocorrelation coefficients {r ₀ , r ₁ , r ₂ } are calculated for each frequency group. On the basis of the coefficients, the partial autocorrelation coefficient {k ₁ , k ₂ } is calculated.In the high-band frequency group, {k ₁ (H), k ₂ (H)} is calculated. In the low-band frequency group, {k ₁ (L), k ₂ (L)} is calculated.

They are then completely identified in the first test, k ₂ (which can use either H or L) if only noise is detected through a signal detection process that continuously compares the three piece autocorrelation coefficients. Go to the first stage of preprocessing and receive the next sample data and repeat the following process. At this time, the variable IA value is incremented from 0 to 1 to contain information about the no-signal period between digits. Since there is no information, the variable KU containing information on whether there is a digit tone for the current sample data is set to the signal member S ₀ , and then the information of the variable KU is converted into a variable as it goes to the first stage of the preprocessing to repeat the following process. The information of variable LA is passed to variable LA2.

On the other hand, if k ₁ (H or L) passes the threshold test, then k ₁ (H) is compared with its threshold, and if it fails the threshold test, it is assumed that a voice signal has been received and the variable IA value as in the first test. Increments 1 from the current value, sets the variable KU to the signal member S ₀ , and then returns to the first stage of the preprocessing.

However, in the k ₁ (H) test, if the value of k ₁ (H) is less than the threshold, it is finally determined that a digit tone signal has been received when k ₁ (L) is within these ranges when compared with the upper and lower limits of the threshold. If the variable KU is set to the signal presence S ₁ and the process proceeds to the effective digit determination process, but fails the test of k ₁ (L), the variable KU is regarded as a temporary malfunction due to crosstalk between adjacent telephone lines. Set to (S ₂ ) and proceed to the effective digit discrimination process, where the judgment hold (S ₂ ) means a reservation for the presence of digit tones, if it passes k ₂ (H or L) and k ₁ (H) Clearly, the presence of a tone signal but a failure in the k ₁ (L) test means a case of temporary malfunction due to crosstalk, etc.

However, for the digit tone detected by the partial autocorrelation coefficient test, the code number corresponding to the detected digit is stored from the detection band table of the k ₁ coefficient values for the eight tone frequencies shown in FIG. 5 and FIG. As a result of signal detection, if signal presence S ₁ or judgment hold (S ₂ ) is detected, it goes to the next valid digit determination process. If two consecutive detection results are signal presence S ₁ or {S ₁ , S ₂ , S ₁ } case, among digits meet the conditions for the no-signal period is determined as a valid digit has been newly transmitted and sikimyeo outputs the digit where the variables LA2 and variable KU signal exists S ₁ and the variable LA determines that hold S ₂ is temporary cross talk Because of this, the information on the non-signal period between digits should be compensated.

Thus, variable IA compensates by having the current value minus one.

On the other hand, after the effective digit tone signal is determined, the flow chart of the reception method resets the variable IA to ø, returns to the first stage of the preprocessing process, receives the next sample data, and repeats the same process. In the DTMF signal reception method of the figure, the effective digit determination is performed within 24-40 ms. The sample data sampled at 4 kHz takes N = 48 or N = 52 corresponding to 12 ms or 13 ms each time the process is repeated. I'm trying to.

(48 × 0.00025Sec = 12ms; 52 × 0.00025Sec = 13ms).

Also, the variable IA, which represents the non-digit period between digits, is set to IA≥4 so that 4x12ms = 48ms, since the non-digit period between digits is at least 40ms.

FIG. 10 shows an example of a typical timing diagram according to the DTMF signal reception method of FIG. 9, and event A is a case where multiple tones are detected, and at least two signals are present from the signal detection process in the flowchart of FIG. Because (S ₁ ) is not continuously detected and the tone duration is not valid, the previously detected digit # (n-1) remains the same, and event B is the case where the nth digit is detected. Since the no-signal period between the time and the digits between # (n-1) and #n is valid, it is determined that a new digit has been received and a digit code number corresponding to digit #n is outputted, where the condition is m≥ ø.

Event C is the case where the end of digit #n is detected and digit information #n is maintained. Event D is the case where digit # (n + 1) is detected. The corresponding tone digit code number is output, and event E represents an acceptable drop out of the duration of digit # (n + 1), in which case no signal duration between digits is short, Move to the next sample data block without printing the digit code number.

Event F is a case where a voice signal is received between digit sending periods. After the end of the digit # (n + 1) is detected, the signal F passes through the no signal period, and then passes through the signal detection process of FIG. It is a case of making a no-signal decision as a signal, and event G detects digit # (n + 1), and both the tone duration and the no-signal period between digits are valid and outputs the corresponding digit code number.

FIG. 11 illustrates an example of a design of a digital DTMF signal receiver using a time domain processing scheme in connection with the present invention. The receiver includes a dial tone removal filter for removing a dial tone and limiting a band of an input noise signal as described above. 30) and the receiver circuit converts the samples sampled at 8 kHz from the falling sampler 31 into samples sampled at 4 kHz, and uses the high band frequency band separation filter 34 and the low band frequency band separation filter 37 ) Are divided into two branches (32) and (33) for separate processing into the high band frequency group digit tone and the low band frequency group tone, respectively. Are inputted to the autorelation calculator 35 and the piece autorelation calculator 36, which are cascaded with the high band frequency group separation filter 34, and passed through the low band frequency group separation filter 37 at the branch 33. Flags are the low-frequency group to the separation filter 37 and the cascade-connected with the autocorrelation calculator 38 and the side auto-correlation calculator 39 is input to.

The output values from the two autocorrelation calculators 36 and 39 are then connected to the signal detection and valid digit determination circuit 40.

Operation of each block of the receiver is performed by pre-processing the linearized input sample signal {X (n)} from the compressed non-linear Pulse Code Modulation (PCM) signal and first dialing the dial tone in the dial tone removal filter 30. After removing and sampling the sample signals that have been sampled and transmitted at 8 kHz from the falling sampler 31, the samples are sampled in 4 kHz sampled and then divided into two (32) and (33) for band separation. The autocorrelation coefficients are then calculated from the sample data sampled down by the autocorrelation calculators 35 and 38 of each frequency group and separated by a high band / low band frequency group separation filter (34, 37). The signal autonomous correlation circuit 40 calculates the partial autocorrelation coefficients from the partial autocorrelation coefficients 36 and 39 and the previously calculated partial autocorrelation coefficients. In A signal detection process for determining the presence or absence of a signal and frequency values of the tone, and the validity of the detected digit tones.

As described above, the conventional first reception method has a problem in that it is sensitive to a change in the input signal level for accurate signal detection, whereas the digital signal reception method according to the present invention is a partial autocorrelation input signal used for digital tone detection. It is always constant regardless of the level, which maintains the same detection capability under any fluctuation of the signal level and provides almost perfect immunity against the malfunction response of the receiver which is easily caused by pseudo signals such as noise and voice. In addition, by using samples sampled at 4 kHz for signal detection, the number of channels that a receiver can process is doubled compared to the reception method using the samples sampled at 8 kHz. By effectively using only one parameter, The advantage of simplifying the signal detection and effective digit tone discrimination circuit is much simpler than the existing effective digit tone discrimination circuit due to the reduced overhead, and the DTMF signal reception method according to the present invention is entirely digital signal processing, so it is a single chip. Since the DSP can be effectively implemented using a DSP (Digital Signal Processor), implementing a receiver with such a DSP chip has advantages in minimizing and manufacturing a DTMF signal device and improving cost and reliability.

Claims (4)

In the signal receiving method of the electronic exchange, the dial tone is removed using the dial tone removal filter 30 for detecting and determining the effective digit tone, and from the sample signal sampled at 8 kHz by the falling sampling 31. Converting to a sample signal sampled at 4 kHz, separating high and low band frequency group digit tones by high band / low band frequency group separation filters 34 and 37, and high band / low Performing autocorrelation analysis by the autocorrelation calculators 35 and 38 and the piece autocorrelation calculators 36 and 39 that are cascaded to the band frequency group separation filters 34 and 37, and The output of the branch is input to the signal detection and valid digit determination circuit 40 and then received by the steps of outputting the result of signal detection by determining the presence or absence of a signal and the frequency values of the tone. DTMF signal reception method. The method of claim 1, wherein k ₁ (H), k ₂ (H) calculated by the high band piece autocorrelation calculator 36, and k ₁ (L) calculated by the low band piece autocorrelation calculator 39, with respect to the _{k 2 (L) k 2 (} H or L), of the electronic switch, characterized in that detecting the effective digit tone signal to continuously test the three parameters _{k 1 (H), k 1} (L) in order DTMF signal reception method. The method of claim 1, further comprising: decoding a tone frequency from a detection band table of k ₁ for eight tone frequencies, and setting four variables KU, LA, LA2, and IA containing information on the presence or absence of a digit tone. And setting the number of samples N = 48 or N = 54 to confirm the existence of the signal, and comparing the history of the adjacent digit detection results of the variables KU, LA and LA2 while checking the four variables. DTMF signal receiving method of an electronic exchange, characterized in that the signal processing is performed by determining the detected digits as valid digits when it is confirmed that two signal presences are satisfied while the signal-free period of the variable IA is satisfied. Of claim 1 to claim 3, wherein the signal member (S ₀₎ and a signal is present (S ₁₎ and the determined setting a hold (S ₂₎ step, (S ₀₎ in order to correct the transient error at the time of signal detection The step of setting the signal presence (S ₁ ) and the judgment hold (S ₂ ), and the previous digit determination result and the subsequent digit determination result centering on the determination hold (S ₂ ) is the digit tone as the signal presence (S ₁ ). And a step of confirming whether there exists and correcting a temporary detection error by declaring it as a valid digit, if present.

Images