JPS615660A - Agc locking method of modem - Google Patents

Abstract

PURPOSE:To attain high-speed AGC locking and to reduce a training signal period by executing AGC (automatic gain control) processing for plural number of times during one symbol of a training signal to apply locking. CONSTITUTION:The titled system consists of an A/D converter 1 and a signal processor 2, and the processor 2 is provided with a demodulator 4, a roll-off filter ROF5, an AGC is provided at the post-stage of the ROF5 and is executed by base band processing and the locking is conducted by executing plural number of times of the AGC7 during one symbol of the training signal. Thus, high- speed locking is attained and the AGC locking time is reduced. Thus, the training time is reduced also and a communication charge is reduced and the response from the request of transmission to the transmission permission is decreased.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention is an AGC (automatic gain control) for keeping the signal level constant in a digital modem (fjE modulation device).
Regarding the retraction method, AGC that can be retracted particularly quickly
Regarding the retraction method.

Modems (modems) are used to transmit data using voice band lines such as telephone lines. In such modems, since the line conditions are not uniform, a training signal is sent out before data transmission.
The other party's modem performs a pull-in operation to send a signal fj::::::: according to the state of the line.

The demodulation section of a conventional modem consists of an A/D (analog/digital) converter 1 and a signal processor 2, as shown in Figure 8.The A/D converter 1 samples the received signal from the line. After converting it into a digital value, it is input to the signal processor 2, and the signal processor 2 passes the digitized received signal through AGC processing 3, demodulation processing 4, roll-off filter processing 5, and automatic equalization processing 6, and outputs the demodulated signal. I was getting it. That is, in the signal processor 472, the signal level is equalized by AGC processing 3, demodulated by a carrier wave by demodulation processing 4, and further subjected to roll-off filter processing 5.
Automatic equalization processing 6 after waveform shaping (bandwidth limitation) by
The demodulated output was obtained by

In the demodulation section of such a modem, it is necessary to adapt each process in advance before digital transmission according to the line condition, and as shown in Figure 9, the training signal ( A z signal of all "1"s is transmitted, and each process is adapted in advance by connecting the line using this training signal. For this adaptation, it is necessary to adjust the CAPC (carrier phase control) of carrier (Wi transmission) detection, tanning pull-in, AGC pull-in, and automatic equalization, and the training signal inevitably requires the time required for these adjustments. minutes will be sent.

On the other hand, conventional AGC processing is placed before demodulation processing,
The input signal is a random waveform as shown in FIG. For this reason, conventional AGC processing was performed as shown in the equivalent circuit shown in FIG.

That is, the received signal %X is multiplied by the control voltage β by the multiplier 31,
In a predetermined dynamic range, an output signal X' having a uniform level force is obtained. This control voltage β is created by a feedback loop as follows. Output signal X'
is made into an absolute value by the absolute value circuit 32, and this is converted into a negative signal AX'
is subtracted from the reference voltage α2 by an adder 33, and further multiplied by a 1 feedback coefficient α3 by a multiplier 34 to obtain the food pack amount (M difference amount). This feedback amount is added to the integral value (tap value) T of the tap 36 in an adder 35, averaged, and further multiplied by a predetermined coefficient α4 in a multiplier 37, and then a predetermined value α5 is added in an adder 38. The control voltage .beta.

Therefore, in the conventional AGC processing, the error amount is obtained after converting into an absolute value, and then averaged and then limited by a limiter to obtain the feedback control voltage β.

This type of AGC processing was placed before demodulation because in conventional analog modems, the signal level is equalized and then the signal is passed through a demodulator and roll-off filter to achieve a perfect S/N ratio. Correspondingly, it has also been adopted in digital modems.

[Problem that the invention seeks to solve]

On the other hand, if the lead-in time of the AGC line is reduced, the training signal transmission period will be shortened accordingly.
As shown in Fig. 9, the time from when the transmission request R3 is issued to when the pull-in is completed and the transmission is cleared C8 is shortened, and the time to use the line for pull-in (training) is shortened, which improves responsiveness and line charges. It's convenient.

However, conventional AGC processing is placed before demodulation processing, and is a passband processing in which the input signal is a random modulation/modulation waveform, so the amount of error in tap correction also largely depends on the waveform, and the input level Since it is difficult to calculate the average value from short-term information, there is a problem in that the pull-in time cannot be shortened in principle.

[Means for solving problems]

The present invention provides an AGC modem that can speed up the AGC pull-in time.
To provide a GC pull-in method.

For this reason, the present invention obtains a demodulated output by performing a demodulation step of demodulating the received signal with a carrier wave, a roll-off filter step of performing waveform shaping, an AGC step of performing automatic gain control, and automatic and dynamic equalization. an automatic equalization step, the AGC step is provided after the roll-off filter step and executed by baseband processing, and the AGC step is executed multiple times between one symbol of the training signal. It is characterized by pulling in.

Further, in an embodiment of the present invention, the AGC step includes a step of extracting a DC component of the output of the roll-off filter step, and a step of performing gain control based on the extracted DC component. It is said that

Furthermore, in another embodiment (a) of the present invention, the DC component extraction step is characterized in that the output of the roll-off filter step is squared and averaged. In this aspect, the step of performing the gain control is characterized in that an error amount is obtained from the extracted DC component, and the error amount is averaged to obtain the control amount.

[Effect]

In the present invention, since the AGC step is provided after the roll-off filter step, AGC using baseband processing is possible. Therefore, the input analog waveform can be converted to DC and processed, stable and high-speed processing can be performed, and the interval between one symbol Since the AGC step can be performed multiple times, the AGC pull-in can be performed quickly and the training signal period can be shortened.

〔Example〕

Hereinafter, the present invention will be explained in detail with reference to Examples.

FIG. 1 is a configuration diagram of one embodiment of the present invention, and in the figure,
Components that are the same as those in the figure are shown with the same symbols, and 7 is the AGC
The processing steps are shown. In both FIG. 1 and FIG. 8, the demodulation process 4, roll-off filter process 5, AGC process 7, and automatic equalization process are blocks that represent processes performed by the signal processor 2 by executing a program.

In the present invention, the AGC processing step is placed after the roll-off filter processing step 5 and before the automatic equalization processing step 6, and the AGC processing step is not a modulation signal from the line but a roll-off Since it is the output of filter processing step 5, A by baseband processing
GC operation becomes possible. AGC processing for this purpose will be explained below.

FIG. 2 is an equivalent circuit diagram of AGC processing of the configuration shown in FIG.
The processing executed by the program is shown by an equivalent circuit, and FIG. 2(A) is a schematic diagram thereof, and FIG. 2(B) is a detailed diagram thereof.

In the figure, 70 is a voltage control amplifier (step) which multiplies the input signal (ROF output) by a control amount β to generate an AGC output, and 71 corresponds to the multiplier 31 in FIG. This is a power calculation section (step), which is composed of a squaring circuit, and obtains the absolute value of the AGC output, and corresponds to the absolute value circuit 32 in FIG. 10. 72 is an averaging section (step). , between one symbol, AGC
What averages the output, 73 is an error calculation unit (step)
From the output of the averaging section 72, the reference value α2 and the control value α
3 is used to calculate the amount of error, and 74 is a tap correction section (step) consisting of an adder 33 and a multiplier 34 shown in FIG. 10, 75 is a limiter which limits the output of the tap correction section 74 by coefficients α4 and α5, and outputs the control amount β. Yes, the multiplier 37 in FIG.
It is composed of an adder 38.

In this AGC processing step, the ROF output is
l) and Y (Image), so the second
As shown in figure (B), the voltage control amplifier 70 is connected to 70X and Y on the Y side.
It consists of 70Y on the side. In the example shown in FIG. 2(B), since the variable range of the control amount that can be input to the multiplier receiver is small, it is configured with multi-stage (n-stage) multipliers. Also, the power calculation unit 71 includes multipliers 71a and 71b and an adder? The Y side is squared by a multiplier 71a, and the Y side is squared by a multiplier 71a.
1b is squared, and an adder 71c adds them to obtain an absolute value. Furthermore, the averaging unit 72 includes taps 72a, 72b and an adder 72c to average the output between one symbol.
, 72d, the adder 72c adds the two previous inputs and the previous input, and further adds this and the current input to the adder 72.
The values are added and averaged by d. In addition, 76X
, 76Y are limiters, each of which limits the AGC output input to the AGC loop using a coefficient α1.

The AGC processing configuration in FIG. 2 is different from the configuration in FIG.
Basically, a one-symbol averaging section 72 is added, and as a result, an <AGC loop, which will be described later, is executed by baseband processing.

Next, before explaining the operation of the embodiment shown in FIG. 1, the demodulation processing and roll-off filter processing of the arrangement shown in FIG. 1 will be explained with reference to FIGS. 3 and 4.

FIG. 3 is an equivalent circuit diagram of demodulation processing in the configuration of FIG. 1, and in the figure, 40 and 41 are multipliers.

Assuming that the received signal is composed of guibits,
If the cos θ of the carrier wave is multiplied by the received signal by the multiplier 40, the real part X (Real) becomes one 5 in θ of the carrier wave.
If the received signal is multiplied by the multiplier 41, the imaginary part Y'(
1+*age) is obtained.

FIG. 4 is an equivalent circuit diagram of the roll-off filter processing of the configuration shown in FIG.
0Y is an X-side roll-off filter and has the same configuration. 51a, 51b151c/-51n are taps for delay, 52a, 52b, 52cm
52n is a coefficient multiplier, and each tap 513 to 51n
53 is an adder which calculates the residue of the output of each coefficient multiplier 528 to 52n, and outputs the roll-off filter output ROFX.
(ROFY).

This roll-off filter has impulse response characteristics,
As is well known, waveform shaping (bandwidth limiting) is performed.

Next, the AGC pull-in operation in the embodiment configuration shown in FIGS. 1 to 4 will be explained with reference to the pull-in operation explanatory diagram in FIG. 5.

Here, the received signal is 1. 2KHz (1200 baud, 2
400 bpa) will be explained.

Since the training signal is two signals (all "1"), the 1.2KHz modulation spectrum is divided into 1.2KHz and 2.4KHz signals as shown in Figure 5 (A>).
11 ngletone.

After A/D converting this in the A/D converter 1, the signal processor 2 demodulates it at 1.8 KHz in the 111m processing 4, and as shown in FIG. 5(B), <0.6 KHz (2,4
-1,8), 3.0 (0,6+1.8)KHz, 4
．． A demodulated spectrum of 2 (2,4+1.8) KHz is obtained. Furthermore, by roll-off filter processing 5, the spectrum after the roll-off filter becomes as shown in FIG. 5(C).
The frequency is 0.6KHz.

Since the output of this roll-off filter processing 5 is an analog waveform, it is necessary to convert it to DC and perform baseband processing.

Therefore, when squared by the power calculation unit 71 in JAGC processing, the spectrum shown in FIG. 5(D), that is, 0 KHz
(DC) and 1.2KHz.

This 1.2KHz component is the first E during one baud rate.
Therefore, when averaged over one symbol (baud rate) by the averaging section 72, it becomes zero, and the input waveform is converted to DC.

Therefore, the subsequent error calculation 73, tap correction 74, and limiter 75, which are normal AGC control in AGO processing, can be performed by baseband processing. Direct current (
If it can be performed using the baseband), it is not affected by the waveform, so high-speed and stable AGC processing can be performed, and the average level can be reliably calculated with a short number of symbols.

In addition, the following error calculation 73, tap correction 74, limiter 7
The operation of No. 5 is the same as the conventional one.

On the other hand, as shown in FIG. 7, the pull-in rate of AGC improves according to the number of pull-ins, and usually requires about 5 to 6 AGC processes (loop processes).

Also, since AGC tap correction is normally performed once per symbol, even if the control amount is maximized, the pull-in time is 6 symbols (1
If it is 200 baud, it will take 5 m).

In the present invention, since the above-mentioned high-speed AGC processing can be performed, the AGC processing can be performed multiple times between one symbol (baud rate), so that the above-mentioned loop processing can be performed 5 to 6 times within one symbol.

Therefore, as shown in FIG. 6, the present invention performs AGC processing six times (multiple times) using the roll-off filter output obtained by demodulation and roll-off filter processing to complete the pull-in within one symbol. That's what I do.

For example, in 1200 baud (1, 2 &) data transmission, the audio band is 3.4KHz, so sampling at a frequency more than twice this band is required for signal reproduction. 7.2KHz,
Suppose that sampling is performed six times during one symbol. If the roll-off filter processing is performed three times at this time, the output of the roll-off filter processing for three times is written to the RAM (random access memory) of the signal processor 2, and this output is repeatedly used to perform the AGC processing six times. By performing this process once, AGC pull-in is completed for one symbol. That is, at the time of pull-in, the same AGC processing is repeated in a loop to shorten the pull-in time.

On the other hand, after such pull-in is completed, one AGC process is performed for one symbol, similar to normal dynamic AGC processing.
G Perform CtlilJ control. At this time, since the input received signal is not limited to a Z signal like a training signal, it becomes a random signal, and the influence of the DC converting of the averaging section 72 is small, so dynamic AGC similar to the conventional one can be performed.
Control becomes possible.

The number of times of AGC processing is changed between the pull-in time and the dynamic control time by a sequencer (not shown) giving the signal processor 2 whether it is a training period or a data transmission period.
This is done by the signal processor 2 controlling the number of times of AGC processing.

In the above description, one signal processor 2 performs the demodulation and subsequent steps, but the tasks may be performed by a plurality of processors, or a microprocessor may be used.In short, it is sufficient if the processing can be performed digitally.

Although the present invention has been described above using one embodiment, the present invention can be modified in various ways according to the gist of the present invention, and these are not excluded from the present invention.

〔Effect of the invention〕

As explained above, according to the present invention, the demodulation step demodulates the received signal using a carrier wave, the roll-off filter step performs waveform shaping, the AGC step performs automatic gain control, and the demodulated output is performed by performing automatic equalization. The AGC step is provided after the roll-off filter step and executed by baseband processing, and the AGC step is executed multiple times between one symbol of the training signal to acquire This feature allows the AGC step to be performed at high speed with baseband processing, which allows the AGC step to be performed multiple times during one symbol.
Since the C processing is executed, high-speed pull-in is possible, and the AGC pull-in time can be significantly shortened to 1/6 of the conventional one. Therefore, training time can be shortened, communication charges can be reduced, and responsiveness from transmission request to transmission permission can be shortened. In addition, it has the effect of being easy to realize and is useful in practice.

[Brief explanation of drawings]

Figure 1 is a block diagram of an embodiment of the present invention, Figure 2 is an equivalent circuit diagram of AGC processing in the configuration shown in Figure 1, Figure 3 is an equivalent circuit diagram of demodulation processing in the configuration shown in Figure 1, and Figure 4 is an equivalent circuit diagram of demodulation processing in the configuration shown in Figure 1. Equivalent circuit diagram of roll-off filter processing in Figure 1 configuration, 5th
The figure is an explanatory diagram of the operation of the present invention, and Fig. 6 is an AGC according to the present invention.
Processing explanatory diagram, Fig. 7 is an AGC pull-in characteristic diagram, Fig. 8 is a conventional configuration explanatory diagram, Fig. 9 is a training signal explanatory diagram, Fig. 10 is an equivalent circuit diagram of AG and C processing in the configuration of Fig. 8, FIG. 11 is a modulation waveform diagram. In the figure, 1-A/D converter, 2-signal processor, 3.7-AGC processing, 4-demodulation processing, 5-.-roll-off filter processing, 6-.-automatic equalization processing.

Claims (4)

[Claims] (1) It has a demodulation step that demodulates the received signal with a carrier wave, a roll-off filter step that performs waveform shaping, an AGC step that performs automatic gain control, and an automatic equalization step that performs automatic equalization to obtain a demodulated output. The modem is characterized in that the AGC step is provided after the roll-off filter step and executed by baseband processing, and the AGC step is executed a plurality of times between one symbol of a training signal to perform pull-in. AGC retraction method. (2) The AGC step comprises a step of extracting a DC component of the output of the roll-off filter step, and a step of performing gain control based on the extracted DC component. The modem AGC pull-in method described in (1). (3) The AGC pull-in method for a modem according to claim (2), wherein the DC component extraction step is a step of squaring and averaging the output of the roll-off filter step. (4) The step of performing gain control includes determining an error amount from the extracted DC component and averaging the error amount to obtain a control amount. How to connect modem to AGC.