JPH04127637A - Modem control system - Google Patents

Abstract

PURPOSE:To control the MODEM so that the state is transited from the divergence state, the divergence process and the quasi-steady-state into the steady- state at all times by changing a coefficient converging speed relating to equalizer coefficient revision calculation after refreshing so as to quicken the coefficient converging speed when a steady-state signal is inputted after an external disturbance. CONSTITUTION:The system is provided with an A/D converter 1, a QAM (orthogonal amplitude modulation) demodulator 2, a low pass filter (LPF) 3, an equalizer (EQ) 4, a timing estimate device, an arithmetic section 6 for evaluation and state measure, a control section (CPU) 7, a timing buffer 8, a phase estimate buffer 9, a phase estimate section 10, a phase rotating device 11, a discriminator (QUA) 12, a phase rotating device 13, an equalizer (EQ) buffer 14, and a decoder 15. In this case, the coefficient converging speed when a steady-state signal is inputted after an external disturbance is quickened by changing a coefficient converging speed relating to equalizer coefficient revision calculation after refreshing. Thus, the divergence of the MODEM operation is easily prevented and the convergence after production of an external disturbance is quickly realized.

Description

[Detailed Description of the Invention] The present invention relates to a modem hang-up control method, and more specifically, it is used in various modulation and demodulation methods, facsimile device modulation and demodulation methods, synchronization detection and timing recovery methods of data transmission methods. The present invention relates to a modem (MODEM) hang-up control method. For example, it is applied to facsimile machines, communication modulation/demodulation systems, mobile communication, etc.

Phase 5hift Keying
: Phase modulation method), D P S K (Differ
initial Phase 5hift Keying
In order to restore transmitted data from a received signal, modulation methods such as differential phase modulation (differential phase modulation) correct the frequency characteristics that the received signal receives on the transmission path, and correct the difference in carrier frequency between the transmitting and receiving sides (frequency offset). and correct phase jitter,
Furthermore, it is necessary to extract a synchronization signal from the received signal and then reproduce the transmitted data in synchronization with this synchronization signal.

In order to correct the frequency characteristics, an adaptive equalizer (A
A adaptive equalizer (adaptive equalizer) is used, and its equalizer coefficients are updated at the symbol rate so as to have the inverse characteristics of the transmission channel characteristics. Also.

Frequency offset and phase jitter correction can be done using Signal Constellation (Signal Notification @: SignalCons)
This is done by rotating the receiving point by the estimated phase shift on the signal (tellation). Furthermore, as a method for extracting the synchronization signal from the received signal.

Taking the PSK modulation method as an example, first, a passband signal (received signal) is multiplied by a carrier wave, passed through a waveform shaping filter to become a complex baseband signal, squared, and this synchronous frequency component is passed through a bandpass filter. Then, timing estimation can be performed using a method to detect this as a synchronization signal and various coefficients (for example, equalizer coefficients) obtained in the process of processing the received signal when reproducing the synchronization signal from the received signal. There is a synchronous reproduction method in which synchronous reproduction is performed by extracting information and accelerating or delaying the sampling frequency of A/D conversion of the received signal based on this information.

However, with these methods, in transmission systems where sudden disturbances such as instantaneous interruptions and phase hits occur frequently, the modem's equalizer coefficients may diverge, the estimation error of the rotation amount may increase, and playback problems may occur. This may cause an error in the synchronization timing,
There was a problem that the modem would hang up because it could not detect the synchronization signal itself.

In response to this situation, a first buffer means buffers various coefficients related to equalizer coefficients, carrier recovery, and timing recovery for a predetermined number of symbol rates, and a first buffer means buffers the coefficient values at the end of the facsimile training sequence. 2, a calculation means for calculating an evaluation value (evaluation scale) using the input and output values of the equalizer, and a monitoring means for monitoring this, and this evaluation scale is determined in advance for each symbol rate. An abnormality in the equalizer is detected by comparing it with a threshold value, and this abnormality causes timing recovery (synchronization signal regeneration), equalizer coefficient update calculation,
When it is determined that the carrier recovery (phase shift correction) function malfunctions, the modem Control methods to prevent malfunctions have already been proposed. However, in this case, the convergence coefficient that adjusts the update rate of the equalizer filter taps is fixed (-constant), so
The convergence speed is slow, and as a result, it takes time for the modem to converge, and if the threshold value of the convergence condition is inappropriate, there is a problem that the modem falls into a quasi-steady state (local minimum) where it does not completely converge. Further, since the threshold value for detecting the beginning and end of a disturbance is fixed, there is a problem in that, depending on the transmission path where the S/N is extremely degraded, the modem is controlled from beginning to end, resulting in unstable operation.

Furthermore, the one described in JP-A-1-279617 relates to an automatic equalization circuit that reduces distortion by shaping the pulse waveform used in digital magnetic recording of PCM signals, etc., and uses a transversal filter. a waveform equalization circuit using a waveform equalization circuit, a level detection circuit that detects the level of an input signal, and a coefficient setting circuit that sets coefficient values of a transversal filter from the input and output of the waveform equalization circuit.

The configuration includes a control circuit that controls the operation of the coefficient setting circuit using the output of the level detection circuit.

In addition, there is also the "automatic equalizer" disclosed in Japanese Patent Application Laid-open No. 2-26406, and the "modem device" disclosed in Japanese Patent Application Publication No. 2-57039.
, ``Automatic Equalizer'' in Japanese Patent Application Laid-Open No. 2-16066, etc. have been proposed, but the content of each of them is to prevent malfunctions of the modem or equalizer by monitoring line signals. Yes, it does not include the concept of refresh or fast convergence.

Purpose The present invention was made in view of the above-mentioned circumstances.
Regarding the equalizer coefficient calculation after refreshing, we speed up the coefficient convergence when a steady signal is input after a disturbance, and introduce a state scale to monitor the modem state to always avoid divergent states and divergent processes. Second, we provide a modem control method that controls the modem to transition from a quasi-steady state to a steady state, and we also obtain a threshold suitable for this transmission system by referring to the time average value of the evaluation scale at the end of training. It was designed to provide a modem control method.

In order to achieve the above object, the present invention provides (1) in a phase modulation method used for data transmission, various coefficients related to equalizer coefficients, carrier recovery, and timing recovery for a predetermined number of symbol rates. buffering the various coefficient values at the end of the facsimile training sequence using a second buffer means, and calculating an evaluation scale using the input and output values of the equalizer by a calculation means; The evaluation scale by the calculation means is monitored by a monitoring means, the evaluation scale is constantly monitored for each symbol rate, and an abnormality in the equalizer is detected by comparing the evaluation scale with a predetermined threshold;
Timing recovery (synchronous signal regeneration) due to this abnormality
, when it is determined that the equalizer coefficient update calculation and carrier recovery (phase shift correction) functions are likely to malfunction, the storage of the coefficient values in the first buffer means is stopped, and the timing recovery and equalizer As coefficients related to coefficient update calculations and carrier recovery calculations, coefficient values obtained before a predetermined number of symbol rates from the time when an abnormality is detected in the evaluation scale are loaded from the first buffer means and used to perform refresh processing. and further demodulates the subsequent data for a predetermined number of samples, and if the number of samples whose evaluation scale exceeds a predetermined threshold within this predetermined sample interval reaches a predetermined number, the demodulation process is continued as it is. If not, the same values as the coefficient values loaded from the first buffer means are refreshed and used for the coefficients related to the timing recovery, equalizer system update calculation, and carrier recovery calculation, and the comparison calculation with the evaluation scale calculation is performed. The end of the disturbance is detected by repeating it, but if the number of times the coefficients are refreshed exceeds the preset number,
It consists of a modem control method in which coefficient values are loaded from a second buffer means and processed, and a steady signal after a disturbance is input by changing the coefficient convergence speed regarding equalizer coefficient update calculation after refreshing. furthermore, (2) stepwise changing the convergence coefficient value regarding the equalizer coefficient update operation after the refresh is performed for each predetermined number of samples;
Furthermore, (3) adaptively changing the convergence coefficient value related to the equalizer coefficient update calculation after the refresh is performed according to the value of the state scale, or (4) in the phase modulation method used for data transmission, Various coefficients related to equalizer coefficients, carrier recovery, and timing recovery are buffered for a predetermined number of symbol rates by a first buffer means, and the various coefficient values at the end of the facsimile training sequence are buffered by a second buffer means. an evaluation scale using the input and output values of the equalizer, the evaluation scale calculated by the calculation means is monitored by the monitoring means, and the evaluation scale is constantly monitored for each symbol rate; An abnormality in the equalizer is detected by comparing the evaluation scale with a predetermined threshold, and this abnormality causes timing recovery (synchronization signal regeneration), equalizer coefficient update calculation, and carrier recovery (phase shift correction) functions to be activated. When it is determined that a malfunction is occurring, the storage of the coefficient values in the first buffer means is stopped, and an abnormality is detected in the evaluation scale as coefficients related to timing recovery, equalizer coefficient update calculations, and carrier recovery calculations. Refresh processing is performed by loading the coefficient values before a predetermined number of symbol rates from the time of detection from the first buffer means, and demodulating subsequent data for a predetermined number of samples. If the number of samples whose evaluation scale exceeds a predetermined threshold within the interval reaches a predetermined number, demodulation processing continues; otherwise, timing recovery, equalizer system number update calculation, and carrier recovery calculation are performed. The first coefficient related to
The end of the disturbance is detected by refreshing and using the same coefficient value loaded from the buffer means and repeating the comparison operation with the calculation of the evaluation scale. If the value exceeds the threshold value, the modem control method loads the coefficient value from the second buffer means and performs the processing. , (5) Calculating the time average value of the evaluation scale during a preset predetermined time after the end of the training, and setting a threshold value by multiplying the time average value by a predetermined time; ) The present invention is characterized in that the time average value of the evaluation scale is calculated, and the time average value is multiplied by a predetermined value to obtain the threshold values with different multiple values.

Hereinafter, the present invention will be explained based on examples.

Note that the phase modulation method consists of the following configuration.

That is, a converting means converts an image signal read from an original into a binary signal, and the binary data converted by the converting means is modulated (phase modulated) in association with phase information of a carrier wave, and converted into a passband signal. A transmission signal means for sending out to a transmission path, and a conversion means for receiving a passband signal transmitted by the transmission means, multiplying the passband signal by a carrier wave, and converting it into a complex baseband signal by passing it through a waveform shaping filter. a reproduction means for reproducing synchronization from the received signal; an adaptive equalization means for equalizing frequency distortion received in the transmission system; and a determination for determining transmission data from a complex baseband signal in synchronization with the previous synchronization signal. With means.

It is comprised of a decoding means for decoding the transmission data, and a correction means for correcting the phase shift (frequency offset and phase jitter) received when the received signal is on the transmission path and converted into a complex basepan 1-signal.

FIG. 1 is a block diagram of a demodulating modem including a transmission system for explaining one embodiment of a modem control method according to the present invention. In the figure, 1 is an A/D converter, 2 is a QAM (Quad
rate Ampli-tude Modulat
ion: orthogonal amplitude modulation) demodulator, 3 is a low-pass filter (LPF), 4 is an equalization layer (EQ), 5 is a timing estimation section, 6 is an evaluation and state scale calculation section, 7 is a control section (CPU
), 8 is a timing buffer (TimingBuff
er), 9 is a phase estimation buffer, 10 is a phase estimation unit, 11 is a rotator, 12 is a determiner (QUA), 13 is a rotator, 14 is an equalization layer (EQ) buffer, and 15 is a decoder (
Decoder).

The equalization layer (EQ) 4 is a transversal type FIR (acyclic type: F]njteImpu that constitutes an equalizer part.
lse Re5ponse) filter, and the filter tap Ci is Ci= C1-r・e・xi*(K) (1
) is updated using the calculation formula. Xi is a conjugate complex number input to the equalization layer 4, e is an error, and r is a coefficient for adjusting the convergence speed (convergence coefficient). The reason why the modem diverges in FIG. 1 is because when a large disturbance is added to the human power X in equation (1), the update calculation of the filter coefficient tends to diverge. Therefore, the divergent operation of the modem can be guaranteed unless the operation of the equalization layer of the receiving modem follows the dynamic characteristics caused by sudden disturbances.

Further, by increasing the coefficient convergence speed when a normal signal is input after a disturbance, the modem malfunction period can be reduced. For this reason, buffering of each coefficient of the modem, calculation of evaluation scales (including status scales), and a control section for these are added to normal receiving modem processing. As for processing, in order to increase the convergence speed and avoid falling into a quasi-steady state, the control unit changes the value of the equalizer convergence coefficient r stepwise or adaptively with reference to the state scale. A section has been added that allows you to do so.

Examples of evaluation scales for detecting disturbance sections include the following.

■ Ter(k) = Te(k)-Te(k-1)
(3) Te(k)=Real(C1・
C, *) (4) First, ■■ of the above evaluation scale will be explained.

(2) is a short-time average value of the error e of the equalizer output, and detects abnormalities occurring in the amplitude of the received signal. In addition, ■) is the difference value of the coefficient ratio between the center tap and the next tap of the equalizer coefficient. This mark (■) indicates the amount of change in phase, and is used to detect rapid phase rotation caused by disturbance.

Furthermore, 5(k) is defined as a state scale indicating the state of the modem. This 5(k) is a parameter indicating the convergence state, quasi-steady state, divergence process, and divergence state of the modem system.

5(k) adjusts the modem equalizer coefficient convergence speed parameter r. Examples of 5(k) include the following:

m:0 The buffer includes a first buffer means for storing equalizer coefficients and coefficients related to timing recovery and carrier recovery for each symbol rate for a predetermined number of samples, and a buffer means for storing the various coefficient values at the end of the facsimile training sequence. a second buffer that rings.

The operation of each block described above will be explained below along with the processing for detecting the start and end of the disturbance section.

FIG. 2 is a flowchart showing an example of detecting the start end of a disturbance section. Below, each step will be explained in order.

After starting operation, the receiving modem starts detecting the start of the disturbance section. Processing of the received signal proceeds at the symbol rate according to normal modem processing.

Determine whether F'laning is complete. If you are training, return to step 5 above.

5teP3: After the training is completed, the coefficients of carrier wave recovery and timing recovery 1 equalizer related to normal modem operation are held in the first buffer in order to guarantee modem operation after disturbance occurs.

%; Furthermore, the coefficient value at the end of the facsimile training sequence is stored in a second buffer.

Next, as evaluation scales, ■E(k), ■Ter(
k) is calculated.

The evaluation scale calculated in step 5 is compared with the threshold values THI and TH2, respectively.

ΣLiヱ7; With THI and TH2 as thresholds, E(k)
)THI or Ter(k, )>TH2, it is assumed that a disturbance has occurred, storage in the first buffer is stopped, and processing shifts to detecting the end of the disturbance section.

Note that here, the first buffer holds values from t=◯ (current time) to t=−L (time L samples ago).

This is because the value of the evaluation scale is calculated from the input/output coefficients of the equalizer FIR filter, which involves a time lag in the number of filter taps until the occurrence of disturbance is detected.

FIG. 3 is a flowchart showing an example of threshold value determination including detection of the start end of a disturbance section. Below, each step will be explained in order.

After starting operation, the receiving modem starts detecting the start of the disturbance section. Processing of the received signal proceeds at the symbol rate according to normal modem processing.

- Determine whether training is complete. If you are training, return to step 1 above.

ΣReturn Table - After training is completed, the coefficients of carrier wave recovery, timing recovery, and equalization layer related to normal modem operation are stored in the first buffer in order to guarantee modem operation after disturbance occurs. Ru.

%; Furthermore, the coefficient value at the end of the facsimile training sequence is stored in the second buffer, and the counter C is set to O.

5step 5; Next, as an evaluation scale, +'D E (k,
), ■Ter(k) are calculated.

%; Compare the evaluation scale counter C calculated in step 5 with a predetermined threshold value TH-COUNT.

Ting Wangyu; In step 6 above, C< T H-C
OUNT, that is, when C is less than or equal to THC0UNT, a fixed threshold value THfix that is preset to a high value is used as the threshold value for detecting the beginning and end of the disturbance, and the counter C is set to C.
+1.

W; In step 6 above, C<TH-COL'
If it is not NT, that is, when C is larger than THCCOUNT, the time average value of the evaluation scale calculated between C=O and C=THCOUNT is multiplied by a predetermined value to detect the start and end of the disturbance. Set the threshold values for each.

As the calculation at this time, the following may be considered, for example. The special number THs=m-avr(l evaluation scale 1) THe=n-avr(l evaluation scale 1) where the threshold value THs for detecting the beginning of the disturbance and the threshold value THe for detecting the end of the disturbance are used. Here, m and n are arbitrary integers, but m > n, which allows for tolerance in detecting the start of disturbance. Furthermore, here,
Although the time average value of the absolute value of the evaluation scale was used as the time average value of the evaluation scale, any value may be used.

5tep9: The evaluation scale calculated in 5tep7+ or 5tep8 is compared with threshold values THI and TH2, respectively.

E(k
)) When THI or Ter(k) > TH2, it is assumed that a disturbance has occurred, storage in the first buffer is stopped, and processing moves to detecting the end of the disturbance section.

Although the method for determining the threshold value for a method that increases the convergence speed has been described here, the threshold value determination process is not limited to increasing the speed, and can also be applied to a modem configuration that does not increase the speed.

FIGS. 4(a) and 4(b) are flowcharts 1 to 4 showing an example of detecting the end of a disturbance section. Hereinafter, each step will be explained in order.

The fading end detection is determined by the first step in the start end detection process.
Forces the modem to receive the coefficient values stored in the buffer of (
(This is called refresh), and when the evaluation scale is the observed value or a certain condition is satisfied, it is assumed that the termination has been detected. Here, the determination is made by introducing three parameters: the predetermined time J for which the modem is operated after refreshing the coefficients, the time J for which the modem continues to operate after entering the convergence state, and the number of refreshes.

Complaint: Set the counter k that counts the number of refreshes to O.

5tep2; Set the convergence-related counter and the refresh-related counter 1+J to 0, and set k=+1.

Compare the number of refreshes k with the threshold TH3.

%; In the above 5tep 3, if k>TH3 is not satisfied, refresh the first buffer with the value L samples before.

In step 3 above, k > T H3
In this case, refresh is performed using the coefficient values in the second buffer.

Execute normal modem processing (does not store data in the first buffer, but changes coefficient values).

Step 7: Calculate the evaluation scale and compare it with the threshold.

5tep9; If the evaluation scale to be compared in step 8 is smaller than the threshold value, i=i+1.

That is, the duration i during which the evaluation value takes a value smaller than the threshold value is counted. If this value i is greater than or equal to R, the end detection process is stopped, the convergence coefficient is fixed to that used in the steady state, and the process moves to the start edge detection process.

Above m: In step 8 above, if the evaluation scale to be compared is larger than the threshold value, set 1=O, and count the duration J after refreshing the coefficient.

5tepl 1; Continuation time J=j+t.

Seagull, □■ Tayo; If duration j=G, the above 5 steps
Returning to step 2, if i<R and j<G, the convergence coefficient is changed (step 13) and the process returns to step 5 and step 6.

5step 13; The flowchart for changing the convergence coefficient is in the fourth step.
It is shown in figure (b). The interval G in which refresh is performed and modem processing is continued is defined as interval length Gl, G2 . G
3, and the value of the convergence coefficient in this interval is rl.

r2. Calculation is performed as r3. In this case, although the explanation has been made by dividing it into three parts, it does not matter how many parts it can be divided into.

In step 12, if i = R and J<G, reset the convergence coefficient and set r=r fi
xed.

Note that in the control method shown in FIG. 4, if convergence does not proceed even with the use of the second buffer, the modem is hung up.

FIG. 5 is a flowchart showing another embodiment of detecting the end of a disturbance section. The difference from the flowchart shown in FIG. 4 is that in step 5, a part for calculating the status scale is added in addition to the evaluation scale, and the convergence coefficient within the refresh interval is adaptively changed based on this status value. be. All other parts are the same. In the case of Fig. 5, the relationship between the convergence coefficient r (k) and the state coefficient 5 (k) is
It can be expressed by a function F: r (k) = F (S(k)).

FIG. 6 is an example of the relationship between r(k) and 5(k) based on the function F. Let the minimum and maximum values of the state scale be S, −
min and Smax, the minimum value and maximum value of the convergence coefficient are associated as r-min and rmax, respectively, and the interval therebetween is linearly correlated. Figure 7 is the 6th
It is an evolution of the correspondence between diagrams, and is a step-by-step ■,
Possible correspondences include log ■, exp ■, and the like.

As is clear from the above description, the present invention has the following effects.

(1) Effect corresponding to claim 1: By detecting disturbance from the input signal, divergence of modem operation can be easily prevented, and convergence can be realized more quickly after occurrence of disturbance.

(2) Effects corresponding to claims 2 and 3; equalizer convergence coefficient r
It is possible to speed up the return (convergence) of modem operation by changing it stepwise or adaptively, and at the same time, by monitoring the status, it is possible to control the modem system so that it is always in a normal state. It is.

(3) Effects corresponding to claims 4 and 5.6: It is possible to set a disturbance detection threshold suitable for the transmission system, and modem control can be operated more efficiently.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram of a demodulating modem including a transmission system for explaining an embodiment of the modem control method according to the present invention, and FIG. 2 is a flowchart showing an embodiment of detecting the start of a disturbance section. FIG. 3 is a flowchart showing an example of determining a threshold including detection of the start end of a disturbance section, FIG. 4 is a flowchart showing an example of detecting the end of a disturbance section, and FIG.
Flowcharts, FIGS. 6 and 7, showing other embodiments of detecting the end of the disturbance section, show the convergence coefficient r (k
) and the state scale S (k). 1-A/D converter, 2 ・QAM (Quad
rate Amplitude Modulation
n: orthogonal amplitude modulation) demodulator, 3: low-pass filter (LPF), 4: equalization layer (EQ), 5: timing estimation unit, 6: evaluation and state scale calculation unit, 7: control unit (CPU), 8...-Timing buffer (Timi
ng Buffer), 9 Phase estimation buffer, 10.
... Phase estimation section, 11 ... Rotator, 12 Judgment unit (QU
A), 13. Rotator, 14... Equalization layer (EQ) buffer, 15 =-Decoder. Patent applicant Rico Co., Ltd. (and 1 other person) Figure (b) (a) Figure

Claims (1)

[Claims] 1. In the phase modulation method used for data transmission, various coefficients related to equalizer coefficients, carrier recovery, and timing recovery are buffered by a first buffer means for a predetermined number of symbol rates, and facsimile training is performed. The various coefficient values at the end of the sequence are buffered by a second buffer means, an evaluation scale using the input and output values of the equalizer is calculated by a calculation means, and the evaluation scale by the calculation means is monitored by a monitoring means. and
The evaluation scale is constantly monitored for each symbol rate, and an abnormality in the equalizer is detected by comparing the evaluation scale with a predetermined threshold, and this abnormality causes timing recovery, equalizer coefficient update calculation, and carrier recovery. When it is determined that the function is heading towards malfunction, the storage of coefficient values in the first buffer means is stopped, and the evaluation scale is used as a coefficient related to timing recovery, equalizer coefficient update calculation, and carrier recovery calculation. performs a refresh process by loading and using coefficient values for a predetermined number of symbol rates before the time when an abnormality is detected in the first buffer means;
Furthermore, demodulation processing of subsequent data is performed for a predetermined number of samples, and if the number of samples whose evaluation scale exceeds a predetermined threshold reaches a predetermined number within this predetermined sample interval, the demodulation processing is continued; otherwise, the demodulation processing is continued. To do this, refresh and use the same coefficient values as the coefficient values loaded from the first buffer means for the coefficients related to the timing recovery, equalizer system update calculation, and carrier recovery calculation, and repeat the comparison calculation with the evaluation scale calculation. If the number of times the coefficient is refreshed exceeds a preset number, the modem control method loads the coefficient value from the second buffer means and performs the process. 1. A modem control method characterized in that regarding an equalizer coefficient update operation after the equalizer coefficient is updated, the coefficient convergence speed is increased when a steady signal is input after a disturbance by changing the coefficient convergence speed. 2. The modem control method according to claim 1, wherein the convergence coefficient value regarding the equalizer coefficient update calculation after the refresh is changed stepwise for each predetermined number of samples. 3. The modem control method according to claim 1, wherein a convergence coefficient value related to an equalizer coefficient update operation after said refresh is adaptively changed according to a value of a state scale. 4. In the phase modulation method used for data transmission, various coefficients related to equalizer coefficients, carrier recovery, and timing recovery are buffered for a predetermined number of symbol rates by the first buffer means, and the above-mentioned coefficients at the end of the facsimile training sequence are Buffering various coefficient values by a second buffer means, calculating an evaluation scale using the input and output values of the equalizer by a calculation means, and monitoring the evaluation scale by the calculation means by a monitoring means,
The evaluation scale is constantly monitored for each symbol rate, and an abnormality in the equalizer is detected by comparing the evaluation scale with a predetermined threshold, and this abnormality causes timing recovery, equalizer coefficient update calculation, and carrier recovery. When it is determined that the function is heading towards malfunction, the storage of coefficient values in the first buffer means is stopped, and the evaluation scale is used as a coefficient related to timing recovery, equalizer coefficient update calculation, and carrier recovery calculation. performs a refresh process by loading and using coefficient values for a predetermined number of symbol rates before the time when an abnormality is detected in the first buffer means;
Furthermore, demodulation processing of subsequent data is performed for a predetermined number of samples, and if the number of samples whose evaluation scale exceeds a predetermined threshold reaches a predetermined number within this predetermined sample interval, the demodulation processing is continued; otherwise, the demodulation processing is continued. To do this, refresh and use the same coefficient values as the coefficient values loaded from the first buffer means for the coefficients related to the timing recovery, equalizer system update calculation, and carrier recovery calculation, and repeat the comparison calculation with the evaluation scale calculation. The end of the disturbance is detected by the modem control method, and if the number of times the coefficients are refreshed exceeds a preset number, the coefficients are loaded from the second buffer means and processed. A modem control method characterized in that a threshold value used for detecting the start and end points is set according to the state of the transmission path. 5. The threshold value is set by calculating a time average value of the evaluation scale during a preset predetermined time after the end of the training, and multiplying the time average value by a predetermined time. modem control method. 6. When setting the threshold by calculating the time average value of the evaluation scale and multiplying the time average value by a predetermined value, the threshold value for detecting the start end of the disturbance and the threshold value for detecting the end end are determined using different multiple values. 6. The modem control method according to claim 5.