JPS647703B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a training signal identification method for identifying the next signal when an automatic equalization system pulls in a received signal using a training signal preceding phase modulation data. .

Conventionally, a method has been used in which a training signal preceding phase modulation data is equalized by an equalizer, the equalized output signal is compared with a reference signal, and an error phase component is corrected to pull in a received signal. FIG. 1 shows an example of the configuration. That is, input the input signal to the equalizer EQL1 and calculate the phase,
The amplitude is equalized, and the equalized output signal is passed through the multiplication section 2 at the feedback point of the integral loop of the carrier automatic phase adjustment circuit CAPC5, and then inputted into the judgment circuit 3, where the error with respect to the reference signal is judged, and the error signal is sent to the subtraction section. 4, and input it into the first integrating circuit of CAPC 5 consisting of first and second integrating circuits (not shown), and integrate the frequency offset due to steady phase rotation.
Next, the signal is put into a second integration circuit and integrated to remove jitter and phase fluctuation. The addition outputs of these two integrating circuits are fed back to the determination circuit 3 via the multiplication section 2 so as to make the phase difference applied to the determination circuit 3 zero, and the outputs are branched and input to the multiplication section 6, and the outputs are branched to the multiplication section 6, and the outputs are fed back to the judgment circuit 3 through the multiplication section 2.
This is intended to automatically set up the phase of the received carrier signal to the reference signal by multiplying it by the error signal from the signal and feeding it back to the equalizer EQL1.

The input signal input to such an automatic equalization system is supplied from a data transmission device such as a modem, but momentary interruptions may occur in the line.
When such a momentary interruption occurs, it is necessary to identify whether the next signal to come is a training signal or a data signal. Then, in the case of a training signal, a pull-in sequence on the receiving side is executed, and in the case of a non-training signal (in the case of a data signal), a predetermined timing is executed to prevent the equalizer error from diverging to infinity due to the elapse of the instantaneous interruption time. Executes an instantaneous interruption sequence that provides phase, tap coefficients, etc. This makes it possible to increase the instantaneous interruption resistance of the automatic equalization system.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a training signal identification method for identifying the next signal during an interruption in order to increase the interruption tolerance of an automatic equalization system.

In order to achieve the above object, the training signal identification method of the present invention uses two different phases in the phase plane.
Receive a signal including a training signal having segments whose values change alternately and a data signal having values of arbitrary multiple phases in a phase plane, and equalize degraded components of the received signal by the training signal. A training signal identification method for an automatic equalization system that creates a signal and equalizes the data signal, comprising: an extraction unit that extracts a frequency component corresponding to the segment; and a detection unit that detects the level of the extracted signal component. and when the signal component level detected by the detection unit is constant over time, the detected signal is determined to be the training signal, thereby identifying it from the data signal. This is a characteristic feature.

The present invention will be described in detail below with reference to examples.

As a training signal suitable for the above purpose
The format of Figure 2a as recommended by CCITT will be used. In other words, in the first segment with no signal,
The second segment alternately sends signals at positions A and B on the phase plane, followed by a third segment that sends signals at positions C and D randomly. The second segment is used to detect the training signal and is the main part of the invention. The third segment is used for tap correction of the automatic equalizer. The fourth segment occurs randomly over multiple phases and causes the descrambler to retract.

The spectrum of the AB pattern of the second segment is a frequency pattern consisting of a 1.7 KHz carrier signal and signals on both sides of the carrier signal, 0.5 KHz and 2.9 KHz, as shown in FIG. On the other hand, a data signal has a chevron-shaped frequency pattern covering these three frequencies.

To detect this, first extract the 1.7KHz carrier using a bandpass filter, then 0.5KHz,
The 2.9KHz component is demodulated with a 1.7KHz carrier and a carrier with a phase difference of 90 degrees from this, and then passed through a 1.2KHz bandpass filter, squared, and extracted with a 2.4KHz bandpass filter. When the time change of the output waveform is plotted, in the case of the training signal, both 1.7KHz and 2.4KHz are at a constant level as shown in Figure 3a, while in the case of the data signal, both frequencies are at a constant level (Figure 3B). As shown in the figure, there are large level fluctuations.

This characteristic can be used to distinguish between the two.

FIG. 4 is an explanatory diagram of the configuration of an embodiment of the present invention according to the above-described principle.

In the same figure, Fig. 2a before inputting to the equalizer
The input signal shown in is input to the A/D converter 8 and converted into a digital signal.

If the input signal after the instantaneous interruption is a training signal, the frequency spectrum corresponding to the AB pattern described above is input to the multipliers 9 ₁ and 9 ₂ , and is multiplied by a 1.7KHz carrier having a phase difference of 90°. The frequencies on both sides are demodulated and 1.2KHz is output. Bandpass filter 1 whose output is 1.2KHz
0 ₁ and 10 ₂ are squared by absolute value squaring circuits 11 ₁ and 11 ₂ and then added in an adder 12, and then input to a multiplier 17 via a 2.4 KHz band filter 13.

On the other hand, the output of the A/D converter 8 is branched, and the control output of the loop consisting of the multiplier 15 and the automatic gain adjustment AGC 16 is input to the multiplier 17 and multiplied by the output of the bandpass filter 13, resulting in the effect of reciprocal calculation. is obtained, and a constant output is sent to the effective value detection circuit 19 regardless of the increase or decrease in the input signal level to obtain a DC output. The DC output is compared with the output delayed for a certain period of time through the delay circuit T21 in the adder 22, a difference is taken, and the difference output is sent to the adder 27 as a positive value output by the absolute value squaring circuit 25.

Further, the 1.7KHz carrier is directly extracted from the output of the A/D converter 8 by the 1.7KHz bandpass filter 14, and the same procedure as above is followed. First, the multiplier 18 performs a reciprocal calculation and sends a constant output to the effective value calculation circuit 20, and the DC output is detected by a delay circuit 23 and an adder 24 as an output difference delayed by a certain period of time. It is input to the adder 27 as a positive value output, and added to the aforementioned 2.4KHz output. The result of addition of these 2.4 KHz output and 1.7 KHz output is fed back through the time constant circuit 28 to perform integration, and the integrated output is sent to the adder 31. On the other hand, the outputs of both effective value detection circuits 19 and 20 are branched and added in an adder 29, an average value is determined in an average value detector 30 and sent to an adder 31, and this average value is subtracted from the integrated output. In the case of the training signal, as shown in Fig. 3a, the temporal change at both 2.4KHz and 1.7KHz is almost 0, and therefore the integral output added by the adder 31 is almost 0, so the adder As shown in FIG. 5a, the output gain G of 31 is a constant negative gain A relative to the average value of 0 dB.

The above is the case of a training signal, but in the case of a data signal, it has characteristics that cover the spectrum shown in Figure 2 c, so it has a 2.4KHz output and a 1.7KHz output.
The outputs are output from the respective effective value detection circuits 19 and 20 in substantially the same manner as in the case of the training signal. The difference from the training signal is that the output difference after a certain time delay is output to the delay circuits 21 and 21, respectively.
23 and added as positive side outputs in the absolute value squaring circuits 25 and 26, it does not become 0, and as mentioned above in Fig. 3b, there is a large level fluctuation at each frequency, so the values before and after each sampling are The output difference is detected and added as a positive value. Therefore, as shown in FIG. 5b, what was initially a negative gain G with respect to the average value of 0 dB is gradually added with a positive value as the sampling progresses, and at a certain point the gain G changes to positive. In this way, it is possible to clearly identify whether the input signal after a momentary interruption is a training signal or a data signal, depending on whether the gain G changes from negative to positive over time.

As explained above, according to the present invention, when there is a momentary interruption in an automatic equalization system that pulls in a received signal using a training signal that precedes phase modulation data, the training signal that precedes the phase modulation data is The frequency component of the binary signal included in the signal is extracted, the level of the extracted signal component is detected, and the level of the signal component is determined to be constant to identify it as a training signal. As a result, in the case of a training signal during a momentary interruption, the reception side pull-in sequence is executed, and in the case of a non-training signal, such as a data signal, the equalizer as described above diverges as the momentary interruption time passes. This feature greatly contributes to improving the system's ability to withstand instantaneous interruptions.

[Brief explanation of drawings]

Fig. 1 is an explanatory diagram of a conventional automatic equalization system, Figs. 2 a to c are explanatory diagrams of training signals used in the present invention, Fig. 3 a and b are explanatory diagrams of the principle of the present invention, and Fig. 4 is an explanatory diagram of the present invention. Explanatory diagram showing the configuration of the embodiment of the invention, No. 5
Figures a and b are explanatory diagrams showing the effects of the present invention, in which 8 is an A/D converter, 9 ₁ , 9 ₂ , 15, 17,
18 is a multiplier, 10 ₁ and 10 ₂ are 1.2KHz bandpass filters, 13 is a 2.4KHz bandpass filter, 14
is a 1.7KHz bandpass filter, 19 and 20 are effective value detection circuits, 21 and 23 are delay circuits, and 12 and 2 are
2, 24, 27, 29, 31 are adders, 11 ₁ ,
11 ₂ , 25, and 26 are absolute value square circuits, 28 is a time constant circuit, and 30 is an average value detection circuit.

Claims (1)

[Claims] 1. Receive a signal including a training signal having a segment in which binary values of different phases in the phase plane alternately change, and a data signal having values of arbitrary plural phases in the phase plane. , a training signal identification method for an automatic equalization system that equalizes the data signal by creating a signal that equalizes a degraded component of a received signal using the training signal, and extracting a frequency component corresponding to the segment. an extraction unit; and a detection unit that detects the extracted signal component level, and when the signal component level detected by the detection unit is constant over time, the detected signal is the training signal. A training signal identification method characterized in that the training signal is distinguished from the data signal by determining that the training signal is different from the data signal.