JPS637509B2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention selects an appropriate number of frequencies from among a plurality of predetermined frequencies, and determines which frequency is being used to transmit a multi-frequency signal using the selected frequencies. The present invention also relates to a multi-frequency signal receiver that automatically determines and receives signals on its own.

Hereinafter, a specific explanation will be given using an example of an in-band transmission method that uses the voice band as the frequency and transmits call connection information, etc., but this is for the convenience of explanation and is not intended to limit the present invention in any way. do not have.

Now, in the so-called in-band transmission system exemplified here, one or more sets of sine wave signals with predetermined frequencies are transmitted from one station to another station to convey a certain set of control information. It can be done. The most representative signaling system among these was established in 1964 by the Consultative Committee on International Telegraph and Telephone ( CCITT ) .
international T e´le´graphique et T
e´le´phonique；International Telegraph and
Telephone Consultative Committee; hereinafter simply
It is written as CCITT. ), and the present invention will be explained below using this as an example. In this signal method No. 5, the signal is set every 200Hz from 700Hz.
Six frequencies arranged up to 1700Hz are predetermined, and these six frequencies (hereinafter referred to as selection signals)
A multi-frequency signal obtained by appropriately combining two of the frequencies is used to transmit so-called connection information. In this signal system No. 5, in addition to these six frequencies, 2400Hz and
A monitoring signal consisting of 2600 Hz is prepared and is used for various controls such as starting up the in-band signal receiver and billing device of the other station and when ending a call. These signals are exchanged between telephone offices according to a certain procedure. This signaling system No. 5 was announced in 1973 by the International Telecommunications Union (The International Telecommunications Union).
Green Book Volume V1-2 entitled “Telephone Signaling and Switching” published by Telecommunication Union.
Since it is described in detail in Book, Volume V1-2), further details will not be given here.

Now, the basis of signal detection by such a multifrequency signal receiver is whether the received signal belongs to a predetermined multifrequency signal, and if so, what kind of content it is. It is to judge. Therefore, such multi-frequency signal receivers conventionally include a plurality of bandpass filters designed to have the center frequency of each of the plurality of sine waves prepared in advance as described above, followed by rectification. It was usually constructed using circuits, low frequency devices, threshold circuits, etc.

Although this circuit configuration may essentially be an analog configuration or a digital configuration, recent advances in semiconductor technology have led to the development of LSI (Large-Scale Integrated Circuit) technology and the rapid development of digital signal processing technology supported by this. As a result, the conventional analog signal processing technology is being replaced by digital signal processing technology. This is because digital circuits are easier to integrate into ICs or LSIs, and can be expected to provide long-term stable and highly accurate signal processing capability, which is a feature of digital circuits.

By the way, there are three methods for receiving digital multi-frequency signals: (1) Digital filter method (2)
Fast Fourier Transform Method (3) A discrete Fourier transform method is known.

Regarding the digital filter method, for example, P. Kaul and H. “An All Digital Telephone Signaling Module” by Mr. and Mrs. H. Lieberman.
A paper entitled ``Module)'' was published in ``Proceeding of the National Electronics Conference 73''.
Electronics Conference 73), pp. 245-248. Although the method presented in this paper has achieved the band filter characteristics using a digital filter, it merely replaces the various functions of an analog circuit with a digital circuit, and is still at the stage of development of the above-mentioned technologies. At a certain point in time, the filter order has to be high, which has the disadvantage of increasing the circuit scale.

Next, regarding the fast Fourier transform method, for example, there was research by Kikuchi, Imagawa, and Iwase. ”
He gave a lecture titled, and it was also published on page 189 of the collection of papers from the same lecture. The feature of this fast Fourier transform method is that instead of the bandpass filter used in the digital filter method, a plurality of (hereinafter N
(even number). ) point input data is subjected to fast Fourier transform ( F ast Fourier Transform, hereinafter
It is written as FFT. ) to obtain N frequency components. However, this FFT method
This method is not suitable when it is desired to obtain only an output for a specific frequency component, and has the drawback that the calculation process is complicated.

Regarding discrete transformation methods, for example, I. Koval (I.Koval) and Jii. Mr. G. Gara and his wife, “Digital MF Receiver Using Discrete Fourier Transformation,”
Discrete Fourier Transform)”
Transactions on
Communications) magazine COM-21 pp.1331-
The one published in 1335 (December 1973) is known. This conventional discrete Fourier transform method is
Instead of the band filter used in the digital filter method , a discrete Fourier transform (hereinafter abbreviated as DFT ) is applied to the input data of a plurality of (hereinafter referred to as an even number) points, This method calculates a specific frequency component, and the circuit size is smaller than the digital filter method, and the calculation process is simpler than the FFT method. Furthermore, since only the frequency components of a multi-frequency signal can be obtained, unnecessary calculations are eliminated. It has the advantage that it does not require the use of any other method.Therefore, it is considered to be the most promising method among these three methods.

However, in this conventional DFT method, the standard sampling frequency of the audio signal is 8KHz, and the disadvantage is that the determination must be performed multiple times to prevent malfunctions due to noise, instantaneous interruptions, etc. In other words, it is effective to increase the number of DFT points for the purpose of improving frequency characteristics, but
Increasing the number of DFT points increases the time required for one judgment, resulting in a decrease in the number of judgments that can be made within the standard time and a decline in the ability to prevent malfunctions. This does not solve the problem.

Therefore, the purpose of the present invention is to reduce the number of determinations that can be made within the specified time.
An object of the present invention is to provide a digital multi-frequency signal receiver with improved frequency characteristics by increasing the number of DFT points.

To achieve this objective, the digital multi-frequency signal receiver of the present invention includes K (an integer of 2 or more) sets of multipliers, window function generators, threshold generation circuits, DFT circuit groups, and comparison circuits. K sets of these are connected in parallel, and a time difference of T/K (T is the integration time of DFT, in other words, the time required for N-point DFT) is given to the outputs of the wind function generators. In other words, each DFT circuit group of K sets has a different start time by T/K.
Processing can now be executed in parallel. Furthermore, one determination circuit processes the outputs from the K sets of comparison circuits in a time-division manner with a period T/K. As a result, the time required for multiple (hereinafter referred to as M) determinations is shortened, the number of DFT points is increased, and the frequency characteristics are improved, without shortening the time required for one determination.

Note that when handling input signals that have already been converted into digital quantities, such as PCM input,
The received multi-frequency signal can be input directly to the multiplier, but if the multi-frequency signal being transmitted is an analog signal, it is converted into the desired digital multi-frequency signal through an analog/digital converter, an expander, etc. All that is required is to input the converted signal to the input terminal (usually the above-mentioned multiplier) of the digital multi-frequency signal receiver of the present invention.

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

FIG. 1 is a general structural conceptual diagram of a conventionally used digital multi-frequency signal receiver of the DFT system that receives the selection signal, and shows the case of analog input. An input signal input to an input terminal 100 is sampled by an AD converter 101 at 8 KHz, which is a standard sampling frequency for audio signals, and then converted into a digital quantity. By the way, for signals such as audio signals where the probability of occurrence of small amplitudes is high and the probability of occurrence of large amplitudes is low, it is convenient to quantize relatively coarsely in large amplitude parts and relatively finely in small amplitude parts. . In other words, non-linear quantization that changes the quantization step according to the input amplitude is normally performed in an AD converter, and the input signal converted into a digital quantity by the AD converter 101 depends on the non-linear characteristics of the AD converter. This means that you have . Therefore, the data is expanded by the decompressor 102, which has a characteristic that is an inverse function of the decompressor. Therefore, as a matter of course, in the case of linear PCM input, the above processing is unnecessary.

The signal that has become a linear PCM in this way is sent to the multiplier 1
04, a window function is applied for the purpose of alleviating spectral distortion that would be caused by cutting out a signal of a required length in subsequent DFT processing. This window function is obtained by a separately prepared window function generator 103. The signal multiplied by the window function in this manner is sent to a DFT circuit group 105 consisting of six DFT circuits and a threshold generation circuit 106. Said DFT circuit group 10
In 5, each DFT circuit has the DFT definition formula (Equation (1))
Accordingly, X ₁ = _N 〓 ^k=1 a _k W ^-k1 (1) (However, {a _k } _k = 1, 2, ..., N: Input data series N: Number of DFT points (even number) X ₁ : Frequency 1 component) W=exp(-j2π/N) Frequency components _X1 corresponding to each frequency are determined from the multifrequency signal and sent to the comparison circuit 107. On the other hand, the threshold generation circuit 106 calculates the cumulative power of the input signals input to the threshold generation circuit, and it is not until the Nth input data is input that it is possible to judge which of the multi-frequency signals has arrived. ,
A threshold signal is generated and sent to the comparison circuit 107 as a reference for comparison with each DFT calculation result _X1 . By the way, as is clear from equation (1), the frequency component _X1 calculated by the DFT circuit group 105 shows an accurate value only after input data of N points are input. Therefore, comparison circuit 107
Now, only after the input data of N points are input, the above
Comparing the outputs from the six DFT circuits in the DFT circuit group 105 and the output from the threshold generation circuit 106,
The result is sent to the determination circuit 108.

The determination circuit 108 determines whether or not the received signal belongs to a predetermined multi-frequency signal, and if so, for example, in the case of communication method No. 5, it determines which of the six predetermined frequencies. It determines whether the frequency is being used, whether the signal duration exceeds the standard, etc., and outputs the results.

Figure 2 shows the frequency characteristics of the DFT circuit of a digital multi-frequency signal receiver using the conventional DFT method as explained using Figure 1, with the vertical axis representing the attenuation amount and the horizontal axis representing the multi-frequency signal. The frequency deviation from a predetermined nominal frequency is taken and normalized by the amount of attenuation at the nominal frequency. In Figure 2, 10 is a DFT score of 80.
Indicates the main lobe in the case of a point,
20 indicates the main lobe in the case of 160 points, and 21 indicates the first side lobe in the case of 160 points. Furthermore, the sampling frequency is 8K.
Hz, and a triangular window is used as the window function.

Incidentally, the frequency characteristics of a triangular window with a time width T and a maximum value of 1 are given by the following equation.

F(f)=T/2{Sin(πfT/2)/(πfT/2)
} ² (2) Therefore, the zero point of the frequency characteristic of the triangular window is given by the frequency f that satisfies fT/2=m (m is an integer) (3). Therefore, the interval between zero points is Δf _z =2/T (Hz) (4). Here, if the sampling frequency is f _s and the number of DFT points is N, then T=N/f _s [sec] (5), so the interval between zero points is given by the following equation.

Δf _z = 2f _s /N [Hz] (6) Therefore, if the sampling frequency f _s is 8KHz,
If the number of DFT points N is selected as 80 points, the interval between zero points is 200Hz.
Since this is the same as the frequency interval of the multi-frequency signal, the influence of spectrum distortion due to adjacent frequencies is reduced, which is advantageous. Therefore, D.F.T.
Usually, 80 points were selected as the standard number of points, but as can be seen from Figure 2, the equivalent bandwidth of the main lobe is better with 160 points than with 80 points, in other words, for the same attenuation. It can be seen that since the frequency shift becomes narrower, the influence of spectrum distortion caused by other multi-frequency signals becomes smaller. but,
Since the detection time increases in proportion to the number of DFT points,
In the case of 160 points, the detection time is twice as long as in the case of 80 points, and this was the biggest drawback of this conventional DFT method.

FIG. 3 is a conceptual configuration diagram showing an example of implementation in the case of analog input of the digital multi-frequency signal receiver of the present invention that receives the selection signal, and doubles the number of DFT points without increasing the detection time. This is what it aims to do. That is, one set of AD converters 201
For the decompressor 202, a multiplier, a window function generator, a threshold generation circuit, a DFT circuit group consisting of six DFT circuits, and a comparison circuit, two sets of conventional configurations were prepared and arranged in parallel. death,
A time difference of T/2 (T is the integration time of DFT) is given to each set of wind functions, and the outputs that are shifted by T/2 from each comparator circuit are output with a period of T/2 by one judgment circuit. Perform time-sharing signal processing
We succeeded in increasing the number of DFT points from 80 points to 160 points, improving detection accuracy while keeping the detection time required. Of course, the received multifrequency signal is straight from the beginning.
In the case of PCM input, the AD converter and expander can be omitted.

In this embodiment, an AD converter 201 and an expander 20
2 is branched into two, one is multiplied by the output of the window function generator 212 by the multiplier 211, and the DFT consisting of six DFT circuits is multiplied by the output of the window function generator 212.
The signal is sent to a circuit group 213 and a threshold generation circuit 214.
The other one is multiplied by the output of the window function generator 222 by a multiplier 221, and a DFT circuit group 223 consisting of six DFT circuits and a threshold generation circuit 224
sent to. In the DFT circuit group 213, DFT ₁₁
~DFT ₁₆ 's six DFT circuits each perform a predetermined signal when a signal that matches the set frequency is received.
DFT processing is performed and the result is sent to comparison circuit 215. At the same time, the threshold generation circuit 2
In step 14, the cumulative power of the signal from the start time to the end time of the N-point DFT is calculated, and whether it belongs to a predetermined multifrequency signal or not.
Alternatively, the threshold value, which is the standard for determining what kind of content belongs, can be set to a predetermined fixed value depending on the magnitude of the cumulative power of N points of the signal, or a constant at the cumulative power can be set. A judgment is made as to whether to use the multiplied variation value, a threshold value is determined, and the threshold value is sent to the comparison circuit 215. Comparison circuit 21
5, DFT circuit group 2 consisting of six DFT circuits
The output of each of the 13 DFT circuits (DFT ₁₁ to _{DFT 16} ) is compared with the output from the threshold generation circuit 214, and the comparison result is output.

Further, similar processing is performed on the other set with a time difference of T/2, and the comparison circuit 225 outputs the result.

If the time required for N-point DFT in the determination circuit 203 is T, then the outputs of the comparator circuits 215 and 225 are processed in a time-sharing manner at a period of T/2, and the received signal is processed by the output from each comparator circuit. For example, if it is determined that two frequencies are to be used, it is determined whether or not the frequency is a specific selected frequency among six predetermined frequencies. That is,
Check whether two of the outputs of the six DFT circuits in each DFT circuit group exceed the threshold, or whether this N
By repeating the point DFT many times, it is determined whether the signal duration exceeds the standard, and when the input signal ends, a signal indicating the contents of the two received frequencies and a signal indicating that the signal has been received are output. do.

To explain the embodiment of FIG. 3 in more detail,
FIG. 4 shows the signal processing method. Here, FIG. 4a shows the analog input signal to the receiver, FIG. 4b shows the signal processing of the path passing through the multiplier 211 shown in FIG. 3, and FIG. 4c shows the multiplier 211 shown in FIG. Vessel 22
Let us show the signal processing of a route passing through 1.

First, when an input signal as shown in FIG. 4 a is input to the input terminal 200 on the left end of FIG. 3, the AD converter 201
After being sampled at the sampling frequency by , it is converted into a digital quantity and expanded by an expander 202 .
The sample value sequence obtained in this way is the input signal sequence {a ₁ ,
a ₂ , ..., a _k , ..., a _N , ...} (hereinafter abbreviated as { _ak } for simplicity). Then, the input signal sequence {a _k }, which is a series of N signal groups, is the wind function value W ₁ generated by the wind function generator 212 in the multiplier 211 (a triangular window is shown in FIG. 4b). ) and are multiplied by 6
Input to DFT circuits DFT ₁₁ to DFT ₁₆ respectively,
The signal is subjected to a discrete Fourier transform (DFT) and then enters a comparator circuit 215. On the other hand, the other signal branched out from the multiplier 211 enters the threshold generation circuit 214.

The threshold generation circuit 214 calculates the cumulative power of the input data used for the N-point DFT, determines a threshold based on the magnitude of the cumulative power, and outputs it to the comparison circuit 215. For the reason mentioned above, the comparator circuit 215 compares the input from each DFT circuit with the threshold value generating circuit 2 every time the Nth input data is input.
14 is compared with the input from the threshold generation circuit 21.
Instruct the determination circuit which DFT circuit passed the signal outputted in step 4 above the threshold value. DFT
The signal processing by the DFT method, which has been explained focusing on the group having the circuit group 213, is referred to as DFT1 and is illustrated graphically in FIG. 4b.

Similarly, DFT2 shown in FIG. The same is true. That is, in the present invention, the outputs W ₁ and W ₂ from the wind function generator have a time difference, or a phase difference of T/2 (T is N The time required for DFT of a point is given in advance, and predetermined signal processing is performed using a time division method. Then, in DFT1, the first signal processing DFT1-1 shown _{in FIG} . The components corresponding to each frequency of the multi-frequency signal are calculated. At the same time, the cumulative power of the signals from a ₁ to a _N is determined, a threshold value is determined from the power, the threshold value is compared with the calculation result of DFT1-1, and the comparison result is output. Then, the second signal processing DFT1-2 is performed on N points from a _N+1 to a _2N , and the same as DFT1-1 is performed.
A comparison is made between the DFT calculation result and a threshold value, and the result is output.

By the way, as mentioned above, the threshold generation circuit determines the threshold, the comparison circuit compares the threshold with the output of each DFT circuit in the DFT circuit group, and the determination circuit performs the determination only after N points of input data are input. Therefore, in DFT1, the DFT calculation result is compared with the threshold value at T×m (m is an integer) from the start of reception, and the result is output.

However, in DFT2, the output from the wind function generator is delayed by T/2 compared to DFT1 (see Figure 4 b and c), so the input signal sequence from a _1+N/2 to a _N +N/2 The first signal processing DFT2-1 is performed on data at N points, and components corresponding to each frequency of the multifrequency signal are calculated.
Although signal processing DFT2-0 is described in FIG. 4c, this signal processing is an N/2-point DFT and is therefore ignored. Further, at the same time as the signal processing DFT 2-1, a threshold value is determined from the cumulative power of the signals from a _1+N/2 to a _N+N/2, the DFT calculation result and the threshold value are compared, and the result is output. Next, similar processing is performed on the N-point input signal sequence from a _1+N+N/2 to a _2N+N/2 . Therefore, in DFT2, DFT is performed at T×m+T/2 (m is an integer) from the start of reception.
The calculation result is compared with the threshold value, and the comparison result is output.

In this way, DFT1 and DFT2 alternately output the comparison results between the DFT calculation result at N points of the input signal sequence and the threshold value every T/2. Therefore, if one judgment circuit processes the output of DFT1 and the output of DFT2 alternately at a cycle of T/2, the total time required for M judgments will be shorter than when there is only one DFT processing process. will be significantly shortened. For example, when comparing the time required for three determinations with the conventional method, the conventional method processes only DFT1, so the signal processing requires DFT1-1 as shown in Figure 4b.
1. It was necessary to perform DFT1-2 and DFT1-3 in this order, which took a total of 3T. However, according to the present invention, when signal processing is performed, DFT1-1, DFT1-1,
It is sufficient to perform DFT2-1 and DFT1-2, and the total time required is 2T.

The above describes the DFT processing process in two steps using a multi-frequency signal receiver that receives the selection signal of signal method No. 5 as an example.
Reduce the time required to prepare a set and make M judgments.
This is an example of improving the frequency characteristics by increasing the number of DFT points to twice the conventional one. By the way, it is also possible to prepare many more sets of DFT processing processes to shorten the time required for M determinations. In theory
If K sets of DFT processing steps are prepared and the number of determinations M is increased as much as possible, the time required for determination will be approximately 1/K of the conventional one.

Of course, the principle of this invention can also be applied to other signal systems, in which case
It is sufficient to prepare the number of DFT circuits for a predetermined frequency.

However, as a guideline for realistic implementation, let us assume that the above signal system
Regarding the multi-frequency signal receiver that receives the selection signal No. 5, the DFT processing process is unnecessarily increased.
Increasing the number of DFT points not only increases the circuit scale, but also makes the slope of the frequency characteristic curve in Figure 2 steeper. This is not appropriate because it makes the method of determination difficult. In addition, in the case of signal method No. 5,
Since the interval between the nominal values of the multifrequency signal is 200Hz,
Since the detection accuracy of the digital multi-frequency signal receiver will be better if the integer multiple of the interval between the zero points of the frequency characteristic shown by equation (6) is 200 Hz, it is recommended to select an integer multiple of 80 as the number of DFT points. desirable. Therefore,
Because it is necessary to design a multi-frequency signal receiver with a certain margin in the standard, two sets of DFT processing steps are required,
It seems appropriate to choose a score of 160 points.

As described above, according to the present invention, the total time required for M determinations in a digital multifrequency signal receiver can be shortened, and therefore, the number of DFT points can be increased and frequency characteristics can be improved. For example, if you prepare K sets of DFT processing steps, the output from each window function generator,
By giving a time difference of T/K to the start time of DFT and performing time-division signal processing using the regular T/K, the time required for M determinations, which conventionally required only MT, can be reduced to (1+(M- It can be shortened to 1)/K)T.

In the above explanation, a triangular window was used as an example of the window function. The triangular window has the advantage of simplifying the circuit configuration of the window function generator (for example, 212 and 222 in FIG. 3) used in the present invention. Although some may complain that the quality of signal transmission is somewhat unsatisfactory, the configuration of the present invention provides characteristics that are sufficiently high quality to satisfy the standards even when using a triangular window. It will be done.

Of course, if higher frequency resolution of principal components than a triangular window and ability to detect small amplitude components are required, it is of course possible to use other window functions such as a Hanning window or a Hamming window. In this case, the read-only memory (abbreviated as Read Only Memory ) used for the window function generator is
Also called ROM. ) must be taken into consideration, but it is only necessary to change the window function generator, and no change is required to the configuration of the present invention itself. The explanation using FIGS. 4b and 4c can also be applied as is, with only the change in the shape of the window.

[Brief explanation of the drawing]

FIG. 1 is a conceptual diagram of a conventional digital multi-frequency signal receiver using DFT. Figure 2 is the first
The frequency characteristics when using a triangular window in the digital multi-frequency signal receiver shown in the figure are expressed by the number of DFT points.
In the case of 80 (10 in the figure) and in the case of 160 (20 in the figure)
is shown for the main lobe 21 (the main lobe 21 is a side lobe). FIG. 3 is a conceptual configuration diagram showing an example of the implementation of receiving a selection signal with the digital multi-frequency signal receiver of the present invention, and 201 is an AD
Converter, 202 is an expander, 211 and 221 are multipliers, 212 and 222 are window function generators,
213 and 223 consist of 6 DFT circuits
DFT circuit group, 214 and 224 are threshold generation circuits;
215 and 225 are comparison circuits, and 203 is a determination circuit. FIGS. 4a, 4b and 4c are diagrams showing an example of signal processing of the digital multi-frequency signal receiver of the present invention. FIG. 4a shows an example of an input signal to the input terminal 200 at the left end of FIG. Now, if the signal processing in one group in FIG. 3, for example, the group including the DFT circuit group 213, is shown in FIG. 4b, then FIG. 4c is the signal processing in the group including the DFT circuit group 223. . Although the window functions applied in FIGS. 4b and 4c are triangular, this is merely an example, and the present invention is not limited to this.

Claims (1)

[Claims] 1 Select an appropriate number of frequencies from among a plurality of predetermined frequencies that are different from each other, and use the appropriate number of frequencies to multiply the transmitted multi-frequency signal by the output of a separately prepared wind function generator. a multiplier; a DFT circuit group consisting of a plurality of discrete Fourille transform (DFT) circuits connected in parallel to the output end of the multiplier and for calculating the magnitude of each frequency component of a multifrequency signal; and the multiplier. a threshold generation circuit connected to the output of the threshold generation circuit;
A comparison circuit that compares all the outputs of each DFT circuit that constitutes the DFT circuit group, and a judgment circuit that uses all the outputs of the comparison circuit as input and determines whether a normal combination of multifrequency signals has arrived. In a digital multi-frequency signal receiver, K sets of circuit parts each consisting of the multiplier, the window function generator, the threshold generation circuit, the DFT circuit group, and the comparator circuit are prepared and arranged in parallel. A time difference of T/K (T is the integration time of DFT) is given to each output, and the outputs from the K comparison circuits are processed by one judgment circuit in a time-sharing manner with a period of T/K. A digital multi-frequency signal receiver with special features.