KR20010002657A - Method for channel distortion compensating - Google Patents

Abstract

PURPOSE: A method for compensating channel distortion is provided to detect and remove moving ghost inserted into inputted signal and to execute equalizing using data part as well as training sequence if moving ghost is inserted into inputted signal. CONSTITUTION: A method for compensating channel distortion includes several steps. The first step is to calculate DC value from inputted data and read the DC value predetermined times periodically and detect maximum and minimum and store them.(406) The second step is to move values of memories one by one and to store the minimum in the last memory.(407,408) The third step is to determine whether difference between the maximum and the minimum is larger than predetermined first threshold value(409). If the difference between the maximum and the minimum is smaller than predetermined first threshold value, a fourth step is to determine whether difference between values stored in the memories and maximum is larger than prescribe second threshold value.(412) If the difference between the maximum and the minimum is larger than predetermined first threshold value or if the difference between values stored in the memories and maximum is larger than prescribe second threshold value, a fifth step is to determine that there is moving ghost on channel and execute equalizing as data mode.

Description

Method for channel distortion compensating

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the transmission and reception of digital televisions (DTVs), and more particularly, to a method for compensating for channel distortion by removing moving ghosts inserted in a process of receiving a signal transmitted from a broadcasting station.

In general, a DTV system removes a received signal using an equalizer if the received signal contains a ghost signal generated on a channel.

That is, in order to help equalization in the receiver, the DTV system carries a training sequence in a field sync section in every field. Therefore, the receiver uses this training sequence to compensate for distortion occurring on the channel.

1 shows a structure of one frame of a general DTV, and FIG. 2 shows a structure of field synchronization in more detail.

Here, one data segment will be described first, and is composed of a data segment synchronization signal of 4 symbols and data of 828 symbols. A field consists of 313 data segments, and the 313 data segment consists of one field sync segment and a 312 general data segment including a training sequence signal.

In this case, as shown in FIG. 2, one field sync signal has a length of one data segment, a data segment sync pattern exists in the first four symbols, and then a pseudo random sequence. ), PN 511, PN 63, PN 63, and PN 63 exist, and the next 24 symbols include VSB mode related information. Here, the second PN 63 of the three PN 63 section is changed in polarity every time. That is, '1' is changed to '0' and '0' is changed to '1'. Therefore, it may be divided into even / odd fields according to the polarity of the second PN 63.

3 is a modeling of a general channel without a motion ghost. In general, that is, when there is no motion ghost on the channel, the HDTV receiver receives (original signal + ghost signal) as shown in FIG. 3.

The ghosts carried on the channel are removed using the training sequence shown in FIG. 2 loaded in the field synchronization in the equalizer portion of the receiver.

In this case, as shown in FIG. 3, the ghost signal is almost always constant from the standpoint of the HDTV receiver. Therefore, the receiver can sufficiently remove the ghost even with the training sequence included in every field synchronization.

However, when the motion ghost is loaded on the channel as shown in FIG. 4, since the state of the ghost is changed every moment, for example, as shown in FIG. 4, the motion ghost generated by the plane is 1-> 2-> Because they change in the order of 3, the training sequence for each field alone does not effectively eliminate ghosts.

In other words, the conventional method of removing ghosts generated on a channel using only a training sequence cannot effectively cope with movement ghosts on a channel, resulting in a performance degradation of a system such as a broken picture received by a receiver.

Therefore, a patent (application number: 98-62826) has been proposed by the present applicant to solve this problem.

The present invention complements the above patent.

An object of the present invention is to provide a channel distortion compensation method for detecting and removing a motion ghost inserted into a received signal.

Another object of the present invention is to provide a method for compensating for channel distortion that detects and removes a slow moving motion ghost in a received signal.

It is still another object of the present invention to provide a method for compensating for channel distortion by performing equalization using a data part in addition to a training sequence when a motion ghost is inserted into a received signal.

1 illustrates a typical VSB data frame structure

FIG. 2 is a diagram showing the configuration of the data field synchronization portion of FIG.

3 is a model of a general channel without motion ghost

4 is a model of a typical channel with a motion ghost

5 is a block diagram showing a channel distortion compensation device according to the present invention

6 is a detailed block diagram of the equalizer of FIG.

7A and 7B are flowcharts for performing a channel distortion compensation method according to an embodiment of the present invention.

8A to 8D are views illustrating the slicer interior of FIG. 6.

9 (a) to 9 (c) show a blind equalization scheme in FIG.

10 is a flowchart for performing a channel distortion compensation method according to another embodiment of the present invention.

11 is a flowchart for performing a channel distortion compensation method according to another embodiment of the present invention.

Explanation of symbols for main parts of the drawings

100: Micom 101: Tuner

102: demodulation unit 103: reset unit

200: channel decoding unit 201: A / D converter

202: Sync detector 203: VSB mode detector

204: input MSE calculator 205: DC calculator

206: equalizer 207: output MSE calculation unit

The channel distortion compensation method according to the present invention for achieving the above object, the step of calculating the DC value from the input data, reading the DC value for a predetermined number of times at a predetermined period and detecting and storing the maximum value and the minimum value among them; And a step of storing the minimum value in the last remaining memory while moving the values of the memories one by one, and comparing whether the difference between the maximum value and the minimum value stored in the step is greater than a first predetermined threshold value. And if it is determined that the difference between the maximum value and the minimum value is smaller than the first threshold value, comparing the values stored in the plurality of memories with the currently detected maximum value, respectively, to determine whether the difference is greater than the second predetermined threshold value. Determining a difference between the maximum value and the minimum value in the step that is greater than or equal to the first threshold value. And determining that a motion ghost exists on the channel if it is determined that the difference between the value stored in the memories and the maximum value detected by the current reading is greater than the second threshold value, and performing equalization in the data mode.

If the synchronization signal included in the input signal is not detected, the demodulator for demodulating the received data and the channel decoder including the equalizer are reset.

When the equalizer determines that the average square error (MSE) of the equalized signal is greater than the signal power, the demodulator for demodulating the received data and the channel decoder including the equalizer are reset.

The presence or absence of a motion ghost existing on a channel is detected while changing the period of reading the DC value.

Performing equalization in the data mode is characterized in that the equalization is performed by changing the step size of the equalizer to the maximum.

The step of performing equalization is characterized in that, if it is determined that there is no motion ghost and the step size is determined to the maximum, the step size is changed in the order of medium and minimum with a predetermined time interval while turning off the data mode.

The channel distortion compensation method according to the present invention includes calculating a DC value from input data, reading the DC value for a predetermined number of times at a predetermined period, detecting a maximum value and a minimum value among them, calculating and storing an average value, and a plurality of Storing the average value in a memory remaining last while moving values of the memories one by one, comparing whether the difference between the maximum value and the minimum value stored in the step is greater than a first predetermined threshold value; If it is determined that the difference between the maximum value and the minimum value is smaller than the first threshold value, comparing the values stored in the plurality of memories with the average value calculated by reading the current value and determining whether the difference is greater than a second predetermined threshold value; The difference between the maximum value and the minimum value in the step is greater than the first threshold or stored in the memories in the step. If the value of the difference of the average value calculated to determine the current reading greater than the second threshold value to determine that a ghost movement present on a channel is characterized in that comprises a step of performing equalization in a data mode.

Other objects, features and advantages of the present invention will become apparent from the following detailed description of embodiments taken in conjunction with the accompanying drawings.

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

In general, in a DTV transmission, a constant DC value is inserted before a DTV signal is sent for carrier recovery at a receiver. The DC value is represented as a pilot in the frequency spectrum, thereby performing carrier recovery at the receiver. Help.

Since the DC value by the pilot is always constant, it can be seen that the DC value is almost constant when the DC value is calculated from the data input from the receiver when there is no motion ghost on the channel.

Therefore, in the present invention, when there is no motion ghost on the channel to determine the presence of the motion ghost on the channel by using a constant DC value of the input data, especially the presence of a very slow moving ghost If it is determined that there is a motion ghost, the mode of the equalizer is changed to a data mode that uses not only field synchronization with a training sequence for equalization but also general data.

In addition, if it is determined that the equalizer may diverge after determining whether the equalizer diverges using Mean Square Error (MSE) values calculated at the input and output of the equalizer, the mode of the equalizer is blind mode. By switching to (Blind Mode), the equalizer is adaptively operated according to the channel state. Here, the blind mode is to sequentially perform two-level slicing, four-level slicing, and eight-level slicing at predetermined time intervals.

As described above, the present invention not only effectively and quickly removes the movement ghost on the channel that changes every moment, but also enables equalization even when the equalizer is likely to diverge so that the performance of the receiver according to the channel state can be maximized. do.

FIG. 5 is a block diagram of the present invention for performing such a channel distortion compensation method. After tuning a desired channel among RF signals received through an antenna, the RF signal of the tuned channel is converted into an intermediate frequency (IF) signal. An output tuner 101, a demodulator 102 for demodulating the IF signal of the channel tuned by the tuner 101, a channel decoder 200 for removing ghosts of the data demodulated by the demodulator 102, The microcomputer 100 controls the ghost removal of the channel decoder 200, and the reset unit 103 resets the demodulator 102 and the channel decoder 200 under the control of the microcomputer 100. do.

The channel decoding unit 200 may include an analog / digital converter (ADC) 201 for converting input data into 10-bit digital data, and a synchronization detector for detecting a synchronization signal from the 10-bit digital data. 202, a VSB mode detector 203 for detecting a VSB mode from the 10-bit digital data, an input MSE calculator 204 for calculating an MSE of the 10-bit digital data, and inserted into the 10-bit digital data DC calculation unit 205 for calculating the DC value, equalizer 206 for removing ghosts and the like contained in the 10-bit digital data, and MSE calculation unit 207 for calculating MSE of the equalized data. It is composed.

Here, the microcomputer 100 receives the operation results of the synchronization detector 202, the VSB mode detector 203, the input MSE calculator 204, the DC calculator 205, and the output MSE calculator 207. Determine whether to operate the equalizer 206 in training sequence mode, data mode or blind mode, and accordingly the blind mode on / off signal, data mode on / off signal, etc. Output to (206).

FIG. 6 is a block diagram illustrating an internal configuration of the equalizer of FIG. 5.

That is, FIG. 6 includes a feedforward filter and a feedback filter. Here, the feed forward filter portion serves to remove the proximity ghost, and the feedback forward filter portion serves to remove the residual ghost generated in the long ghost and the feed forward filter. Here, if the step size is increased, the convergence speed of the equalizer 206 is faster, but the residual error is larger after convergence. On the other hand, if the step size is smaller, the convergence speed of the equalizer 206 is slower but it remains after convergence. The advantage is that the error is small.

7A and 7B are operation flowcharts of the microcomputer 100 for controlling equalization.

In the present invention configured as described above, the operation mode of the equalizer 206 is assumed to be three modes of a training sequence mode, a data mode, and a blind mode, and the step size is Max, Mid, and Min. Suppose you have three modes of Min).

In this case, the training sequence mode, data mode, and blind mode will be described briefly as follows.

The training sequence mode is a method of performing equalization by using a training sequence included in a field sync interval for each field as shown in FIG. 1, since the training sequence is already known as the training sequence and the actual training sequence. This method compares training sequences of input data to obtain distortion information of a channel, and then applies the information to input data again to compensate for distortion on the channel.

The training sequence mode enables equalization at the receiver very effectively if there is no motion ghost on the channel, i.e. if the distortion on the channel is always constant.

However, since the training sequence comes only once every field, the distortion information of the channel obtained with the first training sequence is the second training in the case of FIG. Since the distortion information of the channel obtained with the first training sequence is applied to the data part that is continuously input after the first training sequence, since the channel information is changed before the distortion information of the channel is obtained again with the sequence, the distortion information of the wrong channel is compensated to the data part. Will result.

Therefore, when the state of the channel is rapidly changing due to the presence of a motion ghost on the channel, a data mode is applied to quickly compensate for the changing channel state by using not only the training sequence but also the data portion for equalization. That is, the data mode performs equalization using training sequences and 8-level slicing.

On the other hand, the blind mode is a method of performing equalization by slicing rather than relying on the information of the training sequence when the equalizer is likely to diverge.

To this end, first, when the power is turned on, a channel change, a system reset, or the like is initialized (step 401). That is, assuming that there is no motion ghost on the channel, the data mode and the blind mode are turned off, and the step size of the equalizer 206 is also set to minimum. The initial values of Counter_1, Counter_2, and Counter_3, which are counters used to determine the presence or absence of a motion ghost, are set to zero. In addition, the time is set to 20 msec, and the memory (memory [20] to memory [1]) is set to 0 at an initial value.

After the initialization is performed, the synchronization detector 202 detects data segment synchronization (ie, horizontal synchronization) and field synchronization (ie, vertical synchronization) from the input data and outputs the detection result to the microcomputer 100. That is, the sync detector 202 detects a data segment sync signal and a field sync signal from A / D converted 10-bit digital data. When both sync signals are detected, the sync detector 202 is set to '0'. If it is not detected, the nSyncLock signal is made '1' and output to the microcomputer 100.

The microcomputer 100 determines whether the nSyncLock signal is 0 (step 402). If the nSyncLock signal is '0', it means that both the data segment synchronization signal and the field synchronization signal are detected. Proceed to

However, when the nSyncLock signal is '1', since the synchronization signal is not detected, the demodulator 102 and the channel decoder 200 of FIG. 5 are reset (step 403). This is because subsequent data processing has no meaning in a state where various synchronization signals necessary for digital data processing are not detected. That is, when the synchronization signal is not detected, the noise may be severe or may be generated even when the demodulator 102 and the channel decoder 200 are incorrectly set. Thus, the demodulator 102 and the channel decoder 200 may be reset. It will be done again from the beginning.

If the nSyncLock signal is determined to be '0' in step 402, the DC calculator 205 calculates a DC value contained in 10-bit digital data, and the microcomputer 100 reads the DC value. The microcomputer 100 increments the value of Counter_1 by 1 every time the read DC value is read once (step 404). The DC values thus read are stored in the memory inside the microcomputer 100, and this is performed until the value of Counter_1 becomes 10. That is, Counter_1 indicates the state of how many times the microcomputer 100 reads the DC value, and increases by 1 each time the DC value is read once.

Then, it is determined whether the value of Counter_1 is 10 (step 405). In the present invention, the number 10 is used as a reference, but this is a value that can be arbitrarily set by the designer.

If the value of Counter_1 is not 10 in step 405, the process proceeds to step 417 in which it is determined whether there is a motion ghost. This is because even if the microcomputer 100 does not read the DC value 10 from the DC calculation unit 205 to determine whether there is a motion ghost on the channel, the system is stabilized by checking whether the equalizer 206 diverges. To make it work.

On the other hand, if it is determined in step 405 that the value of Counter_1 is 10, the value of Counter_1 is reset to 0 again, and the maximum value (maximum DC_Value) of the 10 DC values stored in the internal memory by the microcomputer 100 reads 10 times. And minimum DC_Value are stored separately in variables DCmax and DCmin (step 406).

Then, the values of the 20 memories set in the initialization step 401 are moved one by one (step 407). In other words, the value of memory [19] is moved to memory [20], the value of memory [18] is moved to memory [19], and the value of memory [1] is moved to memory [2]. After each movement of the memory values, the minimum value DCmin detected in the step 406 is finally stored in the memory [1] (step 408).

The values stored in the memories are used to detect a moving ghost with a small change in DC over time, that is, a very slow moving. That is, it is easy to determine that there is no motion ghost on the channel due to the small change of the DC value with time in the slow moving motion ghost. Make this work optimally.

At this time, before detecting the slow moving ghost using the memories, the value of the DCmax-DCmin is first calculated to detect the fast moving ghost first (step 409).

That is, if DCmax-DCmin? 6 in step 409, it is determined that there is a motion ghost on the channel, and the value of Counter_2 is changed to 1 (step 410). Here, Counter_2 is a kind of flag indicating the presence or absence of a motion ghost on a channel. If the value of Counter_2 is 0, it means that there is no motion ghost on the channel, and if it is 1, there is a motion ghost. The reference value 6 is an experimentally determined value, and may be determined differently by an experimenter.

After the value of Counter_2 is changed to 1, the value of Counter_3 is set to 3 (step 411). In this case, once Countermax is determined that DCmax-DCmin ≥ 6 and there is a motion ghost on the channel, DCmax- 3 times in a row without determining that there is no motion ghost on the channel immediately after DCmax-DCmin <6 DCmin <6 only plays the role of judging the disappearance of the motion ghost that appeared on the channel. The reason is that even if there is a motion ghost on the channel, sometimes DCmax-DCmin ≥ 6 may occur. In this case, even though there is a motion ghost on the channel, once DCmax-DCmin <6 This is because the image is broken when the data mode is turned off by determining that there is no motion ghost on the channel. On the contrary, even if the motion ghost disappears on the channel, even if the data mode is turned on for a while, the received image is not broken. Here again, the value 3 set in the present invention as the reference value of Counter_3 is a value that can be arbitrarily changed by the designer.

On the other hand, if it is determined in step 409 that DCmax-DCmin <6, it may be determined that there is no motion ghost on the channel, but in this case, the motion ghost may be very slow on the channel. It is determined whether there is a ghost moving very slowly on the phase (step 412).

That is, the difference in the DC value read 10 times from the microcomputer 100 may not be able to determine the moving ghost moving very slowly on the channel, so that the microcomputer 100 reads the DC value every 10 times. The minimum value, that is, the values of the memory [20] to the memory [1] storing the DCmin value are compared with the maximum value currently read and calculated, that is, the DCmax value.

Here, the values of the memory [20] to the memory [1] are stored in the minimum value every time the microcomputer 100 reads the DC value detected by the DC calculation unit 205 ten times. In addition, in the initialization step 401, time = 20 msec was initialized. This is to set the period in which the microcomputer 100 reads the DC value from the DC calculator 205 in 20 msec units. This value can also be set by the designer.

In addition, the minimum value among the DC values read by the microcomputer 100 10 times every 20 msec is the value recorded in the memory [1] after moving the memory [20] to the memory [1] one by one, and thus the memory [20] to the memory [1]. It can be seen that the values stored in are the minimum values stored 20 times in 200msec units.

Therefore, using the values stored in the memory [20] to the memory [1] it can be seen that the DC value variation on the channel for a total of 200msec × 20 = 4000msec (about 4 seconds). In the present invention, the number of memories is 20, but this can also be arbitrarily set by the designer.

At this time, the DCmax value detected by reading is compared with the values of the memory [20] to the memory [1] already stored, and if the difference is 6 or more, even if it is determined that DCmax-DCmin < It is determined that there is a moving ghost that changes the DC value. In addition, even when the detected DC max value and the values of the memory [20] to the memory [1] are compared with each other, the difference is less than 6, and it is used as the basis for determining that there is no motion ghost on the channel.

In step 412, diff = abs (memory [i]-DCmax) is taken, which means that abs takes an absolute value. The reason for taking the absolute value is that the value of DCmax detected at the present time may be smaller than the value stored in the memory [20] to the memory [1].

Therefore, if the diff value is greater than 6 in step 412, it is determined that there is a ghost moving very slowly on the channel, and the value of Counter_2 is set to 1 (step 410), and the value of Counter_3 is set to 3 (step 411). .

On the other hand, the DCmax value currently detected in step 412 is compared with the values stored in the memory [20] to memory [1], and if the difference is less than 6, it is determined whether the value of Counter_3 is 0 (step 413).

The reason for determining whether the value of Counter_3 is 0 in step 413 is as described above, although the difference between DCmax and DCmin among the values read ten times in the microcomputer 100 is less than 6, and DCmax and the memory [20]- Even if the difference is less than 6 even in comparison with the values stored in the memory [1], it is determined that there is a motion ghost on the channel and does not immediately turn off the data mode.

Therefore, if the value of Counter_3 is not 0 in step 413, the value of Counter_3 is decreased by 1 and the next routine is performed (step 414). If the value of Counter_3 is 0, it is determined that there is no motion ghost on the channel. Resets to zero (415).

After resetting the value of Counter_3 to 0 in step 415, the value of time = 20msec set in the initialization step 401 is changed (step 416).

In the present invention, the time value has four modes of 20msec, 30msec, 40msec, and 50msec, but this can be arbitrarily changed by the designer. The time value here is a period in which the microcomputer 100 reads the DC value calculated from the DC calculator 205.

That is, when it is determined that there is no motion ghost on the channel by first reading the DC value at 20msec period, if it is determined that there is no motion ghost on the channel, the time value is changed to 30 msec again and the microcomputer 100 is 30 msec. Each DC value is read from the DC calculator 205 to determine whether there is a motion ghost on the channel.

In this way, the time value is changed to 50msec and the DC value is read to determine whether there is a motion ghost on the channel.

Changing this time value helps to detect a slowly changing motion ghost in step 412.

If it is assumed that the time value is 50 msec, the microcomputer 100 reads the DC value 10 times in a 50 msec period and repeats the operation of storing the minimum value in the memory [1].

Therefore, the values stored in the memory [20] to the memory [1] are moved after the memory [20] to the memory [1] by one of the minimum DC values read by the microcomputer 100 ten times every 50 msec. In the end, the values stored in the memory [20] to the memory [1] are the minimum values read and stored 20 times in 500 msec units.

Therefore, by using the values stored in the memory [20] to the memory [1], it is possible to know the DC value variation on the channel for a total of 50 msec x 20 = 10000 msec (about 10 seconds).

As described above, when the time value is 20 msec, the DC variation on the channel for about 4 seconds can be known as described above, but when the time is 50 msec, the DC variation on the channel for about 10 seconds can be known. This means that a time of 50 msec can detect motion ghosts that change slowly on the channel much more than a time of 20 msec.

On the other hand, after any one of steps 405, 411, 414, and 416 is performed, it is determined whether the current value of Counter_2 and the VSB mode of the data input from the VSB mode detector 203 (step 417). ). If the value of Counter_2 is 1 in step 417 and the VSB mode of the data input from the VSB mode detector 203 is 8VSB (= 8TVSB) above the ground, it is determined that a motion ghost exists on the channel. The step size of the machine 206 is made to the maximum (step 418). In other words, if there is a motion ghost on the channel, the state of the ghost changes every moment, and the data mode is turned on to use not only the training sequence but also the data portion for equalization, and the ghost size is maximized by maximizing the step size of the equalizer. To be able to quickly follow The reason for checking the terrestrial 8VSB mode is that the VSB mode has an 8VSB mode used for terrestrial broadcasting and a 16VSB mode used for cable broadcasting, because the motion ghost exists only in the terrestrial broadcasting channel. If there is no motion ghost, the step size is reduced.

However, if either of the two conditions is not satisfied in step 417, it is determined whether the step size is maximum (step 419). This case corresponds to the case where there is no motion ghost on the channel from the beginning or the motion ghost exists but disappears in the current channel.

At this time, if the current state of the step size of the equalizer 206 is the maximum in step 419, it means that there is a motion ghost on the channel but now disappears, or the equalizer 206 has performed the blind mode. This is because the above two cases maximize the step size.

In addition, if the step size is not maximum, this is the case that there is no motion ghost on the channel from the beginning, or the channel decoding unit 200 is reset for some reason.

Therefore, if it is determined in step 419 that the step size is not the maximum, the data mode is turned off and the step size is made to the minimum (Min) (step 420).

On the other hand, if it is determined in step 419 that the step size is maximum, then the data mode is not turned off because there is no motion ghost on the channel, and at the same time, the step size is made to be minimum while the data mode is turned off. (Step 421). After a certain time, the step size is changed to the minimum (Min) (step 422). This part is also known as gear shift, because changing the step size from maximum to medium to minimum can reduce the equalizer's convergence time more than changing the maximum from the maximum to the minimum.

After any one of the steps 418, 420, and 422 is performed, the MSE value MSEOUT is checked at the output of the equalizer 206 to determine whether the equalizer 206 is divergent. (Step 423). That is, the microcomputer 100 determines whether the MSE value MSEOUT calculated by the output MSE calculator 207 of the channel decoder 200 is greater than the signal power of 8 VSB (step 424).

Here, the output MSE value of the equalizer 206 is greater than the signal power, which is the case where the equalizer 206 completely diverges, and in this case, convergence is almost impossible even if the equalization is performed in the blind mode. Therefore, if it is determined in step 424 that the output MSE value MSEOUT is greater than the signal power, the demodulator 102 and the channel decoder 200 are reset so as to quickly escape from the divergence state (step 425).

In other words, the blind mode is operated when there is a possibility of divergence, and when the equalizer 206 is completely diverged, it is more effective to reset the demodulator 102 and the channel decoder 200 again from the beginning. Here, the reason for resetting the demodulator 102 is that carrier recovery may not be performed in the demodulator 102 if the output MSE value is greater than the signal power.

Thus, the present invention determines whether to perform equalization in blind mode only if the output MSE of equalizer 206 is not greater than the signal power.

That is, when it is determined in step 424 that the output MSE value MSEOUT is smaller than the signal power, the output MSE value MSEOUT is a threshold value (the threshold value for determining whether the threshold equalizer 206 diverges here). It is determined whether the value is greater than the value set by the designer by experiment (step 426).

If the output MSE value MSEOUT of the equalizer 206 is greater than the set threshold, it is determined that the equalizer 206 is likely to diverge, and if it is small, the equalizer 206 is likely to converge. have.

Therefore, when the output MSE value MSEOUT of the equalizer 206 is smaller than the set threshold in step 426, the process returns to step 402 without repeating the blade mode and repeats the processes described so far.

On the other hand, if it is determined in step 426 that the output MSE value MSEOUT of the equalizer 206 is greater than the set threshold value, the equalizer 206 is operated in the blind mode so that the system can quickly return to a stable state.

To this end, the step size of the equalizer 206 is maximized to allow quick convergence (step 427), and after the step size is maximized, two-level slicing is performed ( Step 428).

After the two-level slicing proceeds to some degree, a four-level slicing is then performed (step 429), and after a four-level slicing is performed to some extent, an eight-level slicing is then performed (step 430).

Here, how long after 2-level slicing goes to 4-level slicing and how long after 4-level slicing goes to 8-level slicing can be arbitrarily set by the designer, but experimentally, the field synchronization period is about 24.2 ms. Is suitable.

In FIG. 6, the slicer 312 is further composed of four detailed slicers. That is, 2, 4, 8, 16 slicers. The concept for these is shown in Figs. 8A to 8D. As shown in (a) to (d) of FIG. 8, 2, 4, 8 and 16 levels of slicing are performed according to each slice mode. At this time, the 16-level slicer of Figure 8 (d) can be used in cable broadcasting, not used in terrestrial broadcasting.

9 (a) to 9 (d) show the scheme applied to the blind equalization of the present invention.

After the two-level slicing proceeds for 24.2 ms (field synchronization interval) as shown in FIG. 9A, the four-level slicing proceeds as shown in FIG. 9B, and again, the four-level slicing is performed for 24.2 ms. After progressing, as shown in FIG. 9C, 8-level slicing proceeds.

That is, in FIG. 6, the output data of the equalizer 206 is sliced according to the slice mode selected by the select signal of the multiplexer 314 and the output data of the equalizer 206. The difference between is obtained by the error detector 315 and this value is used as an error signal so that the feed forward filter 301 and the feedback filter 307 perform equalization.

After 2-> 4-> 8-level slicing is completed, the process returns to step 402 and repeats the same loop.

On the other hand, Figure 10 is another embodiment of the present invention, only the portion different from the above described Figure 7 will be described. At this time, in FIG. Subsequent operations are the same as in FIG. 7B, and thus, detailed descriptions thereof will be omitted.

That is, in FIG. 7, if the value of Counter_1 indicating how many times the microcomputer 100 reads the DC value from the DC calculator 205 is not 10, the microcomputer 100 proceeds to the step of determining whether there is a motion ghost. If the microcomputer 100 fails to read the DC value 10 times, that is, if Counter_1 = 10, the process returns to step 504 and repeats the process of reading the DC value until Counter_1 = 10.

According to the experimental results of implementing the algorithm of FIGS. 7 and 10 described above, the performance is almost the same.

On the other hand, Figure 11 is another embodiment of the present invention, only the portion different from the above described Figure 7 will be described. Likewise, in FIG. Subsequent operations are the same as in FIG. 7B, and thus, detailed descriptions thereof will be omitted.

That is, there may be various methods for the values stored in the memory [20] to the memory [1]. In FIG. 7, the minimum value among the DC values read by the microcomputer 100 into the memory [20] to the memory [1] 10 times. In FIG. 11, the microcomputer 100 stores an average value of DC values read by the microcomputer 100 as another example.

To this end, if it is determined in step 605 that the value of Counter_1 is 10, the value of Counter_1 is reset to 0 again, and then the maximum value among the 10 DC values stored in the internal memory by the microcomputer 100 is read 10 times. DC_Value) and minimum DC_Value are stored separately in the variables DCmax and DCmin, and the average value of 10 DC values is stored in the variable DCvar (step 606).

Then, the values of the 20 memories set in the initialization step 601 are moved one by one (step 607). In other words, the value of memory [19] is moved to memory [20], the value of memory [18] is moved to memory [19], and the value of memory [1] is moved to memory [2]. After moving the memory values, the average value DCvar detected in step 606 is stored in the memory [1].

Then, the motion ghost is detected by calculating what the value of DCmax-DCmin is (step 609).

That is, if DCmax-DCmin? 6 in step 609, it is determined that there is a motion ghost on the channel, the value of Counter_2 is changed to 1 (step 610), and the value of Counter_3 is set to 3 (step 611).

On the other hand, if it is determined in step 609 that DCmax-DCmin <6, it is first determined whether there is a ghost moving very slowly on the channel before turning off the data mode (step 612).

That is, each time the DC value is read by the microcomputer 100 every 10 times, the average value, that is, the values of the memory [20] to the memory [1] storing the DCvar value are compared with the average value calculated by reading the current value, that is, DCvar value. .

At this time, the current read and detected DCvar value is compared with the values of the memory [20] to the memory [1] already stored, and if the difference is 6 or more, even if it is determined that DCmax-DCmin <6 in the step 609, the channel is slow. It is determined that there is a moving ghost that changes the DC value. In addition, even when the detected DCvar value and the values of the memory [20] to the memory [1] are less than 6, it is used as the basis for determining that there is no motion ghost on the channel.

Therefore, if the diff value is greater than 6 in step 612, it is determined that there is a ghost moving very slowly on the channel, and the value of Counter_2 is set to 1 (step 610), and the value of Counter_3 is set to 3 (step 611). .

On the other hand, the DCvar value currently detected in step 612 is compared with the values stored in the memory [20] to the memory [1], and if the difference is less than 6, it is determined whether the value of Counter_3 is 0 (step 613).

Since the subsequent operation is the same as in FIG. 7 described above, a detailed description thereof will be omitted.

As such, the present invention detects not only fast-moving motion ghosts on the channel but also very slow-moving motion ghosts, and trains the equalizer 206, particularly when the equalizer 206 does not diverge without a motion ghost on the channel. Operate in sequence mode. Also, if there is no motion ghost on the channel, but the equalizer 206 is likely to diverge, operate in blind mode, and the equalizer 206 diverges while there is a motion ghost, including a ghost moving slowly on the channel. If not, in the data mode, if there is a motion ghost on the channel and the equalizer 206 also diverges, the data mode and the blind mode are operated simultaneously.

As described above, according to the channel distortion compensation method of the present invention, it is possible to effectively and quickly remove the motion ghost existing on the channel, and in particular, to detect and remove the motion ghost moving very slowly on the channel. In addition, the equalizer can be operated as quickly and stably as possible even when there is a possibility of divergence or when the system diverges.

Those skilled in the art will appreciate that various changes and modifications can be made without departing from the spirit of the present invention.

Therefore, the technical scope of the present invention should not be limited to the contents described in the embodiments, but should be defined by the claims.

Claims (8)

A channel distortion compensation method for removing ghosts present on a channel by using an equalizer that performs equalization in at least one of a training sequence mode, a data mode, and a blind mode, Calculating a DC value from input data, reading the DC value for a predetermined number of times at a predetermined period, and detecting and storing a maximum value and a minimum value among the DC values; Storing the minimum value in a memory remaining last while shifting values of the memories one by one; Comparing whether the difference between the maximum value and the minimum value stored in the step is greater than a first predetermined threshold value; If it is determined that the difference between the maximum value and the minimum value is smaller than a first threshold, determining whether the difference is greater than a second predetermined threshold by comparing the values stored in the plurality of memories with the currently detected maximum value; And If it is determined in this step that the difference between the maximum value and the minimum value is greater than the first threshold value or in this step that the difference between the value stored in the memories and the currently detected maximum value is greater than the second threshold value, it is determined that there is a motion ghost on the channel. And performing equalization in a data mode. The method of claim 1, And a demodulator for demodulating the received data and a channel decoder including the equalizer if the synchronization signal included in the input signal is not detected. The method of claim 1, And reconstructing a demodulator for demodulating received data and a channel decoder including the equalizer when it is determined that the average square error (MSE) value of the equalized signal is greater than the signal power in the equalizer. The method of claim 1, And detecting the presence or absence of a motion ghost existing on the channel while changing the period of reading the DC value. The method of claim 1, Performing equalization in the data mode And performing equalization by changing the step size of the equalizer to the maximum. The method of claim 1, wherein performing equalization And determining that there is no motion ghost and determining the maximum step size, while changing the data mode, changing the step size in the order of medium and minimum at regular intervals. The method of claim 1, wherein performing equalization And turning off the data mode even if it is determined that there is no motion ghost, and turning off the data mode if the determination that there is no motion ghost continues a plurality of times. A channel distortion compensation method for removing ghosts present on a channel by using an equalizer that performs equalization in at least one of a training sequence mode, a data mode, and a blind mode, Calculating a DC value from input data, reading the DC value for a predetermined number of times at a predetermined period, detecting a maximum value and a minimum value therefrom, and calculating and storing an average value; Storing the average value in a memory remaining last while shifting values of the memories one by one; Comparing whether the difference between the maximum value and the minimum value stored in the step is greater than a first predetermined threshold value; If it is determined that the difference between the maximum value and the minimum value is less than a first threshold value, determining whether the difference is greater than a second predetermined threshold value by comparing the values stored in the plurality of memories with an average value currently calculated; And If it is determined that the difference between the maximum value and the minimum value is greater than the first threshold value in the step or the difference between the value stored in the memories and the average value currently calculated in the step is greater than the second threshold value, it is determined that there is a motion ghost on the channel. And performing equalization in a data mode.