JPH0331018B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a method for receiving an information signal consisting of a binary NRZ signal and reproducing the transmitted information with fewer errors, and particularly in a teletext receiver. It is designed to play. Conventionally, to identify a binary NRZ signal, "0" or "1" was determined based on a certain DC level, depending on whether it was above or below that level. Due to waveform distortion of the pulse signal caused by deterioration of line characteristics such as ghosts in the transmission path and deterioration of group delay characteristics, the so-called eye opening ratio (DC level corresponding to the logical value "1" of the input signal) in the code identification section There was a drawback that bit errors caused by waveform distortion, external noise, or internal noise increased as the ratio of the "0" and "1" discrimination range at the sampling point decreased (edited by the Institute of Electronics and Communication Engineers, Communication handbook, published in 1974, No.
(See pages 846-848 and page 1135). An object of the present invention is to provide a code signal restoration method that can appropriately correct code errors such as those described above. To this end, in the present invention, the transmitted received signal is not directly binarized using the conventional method, but the received signal is multivalued by identifying the levels of multi-stage defects. Then, if the level of the received signal is above a certain value, below a certain value, or at an intermediate level, it is determined that there is a high probability that an error has occurred, and it is set to "1" or "0". The error correction effect is increased by outputting the missing bits as lost bits and calculating including the parity bits to restore the lost bits to the original bits. That is, in the present invention, on the transmitting side, a parity bit is added and a code signal consisting of a binary signal of 0 and 1 is transmitted to the receiving side via a transmission path that may cause waveform distortion, and on the receiving side, A coded signal restoration method that enables the coded signal to be restored from the received signal of the transmitted signal, the method comprising: identifying the level of the signal waveform of the received signal into a plurality of finite stages of three or more; Based on the identification result, a bit for which it is unclear whether the level of the signal waveform of the received signal corresponds to 0 or 1 of the binary signal is positioned as a lost bit, and the minimum inter-symbol distance of the code signal is set as d. , the number S of lost bits is within d-1, and the number of error bits is [(d-S-1)/2] bits (where [ ] is a Gaussian symbol, and the number enclosed by the symbol is (meaning an integer that does not exceed the number of signals), the erasure is detected by comparing the code arrangement between the received and level-identified signal and a plurality of code signals prepared in advance on the receiving side. In addition to reproducing the bit, if there is an error bit, the position of the error bit is detected and the codes 0 and 1 corresponding to the position are inverted to codes 1 and 0, respectively, so that the transmission from the transmitting side is The present invention is characterized in that the encoded signal is restored. The principle of the present invention will be explained by taking as an example the case of the (8,4) expanded Hamming code system, which is used in pattern-based teletext broadcasting and consists of 8 bits and 4 information. (8, 4) Since the minimum inter-code distance d of the extended Hamming code is 4, according to the decoding method using normal binary signals, correction when a 1-bit error occurs, 2
It has the ability to detect the existence of an error when a bit error occurs. As a means for identifying the level of a transmitted received signal using multilevel level values as in the present invention,
For example, Japanese Patent Application No. 57-55464 (Japanese Patent Application No. 58-173933)
An A/D converter is used as described in the specification of the publication, and the level of the input received signal is
Output it as a digital signal. According to the method of the present invention, which detects bits whose level is unclear in the received signal as lost bits and determines the bit pattern, in the transmission of the (8, 4) extended Hamming code, the following state occurs in one word. In this case, errors will not be mixed into the output. 1) Loss of 3 or less bits: Lost bits can be restored. 2) 2 lost bits, error bit/bit: Detects that there is an error in this word, and does not take in the code of this word as data. 3) Lost bits/bits, erroneous bits/bits: restore the lost bits and correct the erroneous bits to recover the code of this word. Of course, the present invention also has the conventional ability to correct a 1-bit error and detect the existence of an error (correction is not possible) when a 2-bit error occurs. In the present invention, by detecting "erased bits", the correctable range and the detectable range of errors are expanded. The (8, 4) expanded Hamming code word has the following bit arrangement, and the minimum inter-code distance is 4 bits. Table 1 Code word Information bit Parity bit

[0] 0000 1110 [1] 1000 0000 [2] 0100 1001 [3] 1100 0111 [4] 0010 0101 [5] 1010 1011 [6] 0110 0010 [7] 1110 1100 [8] 0001 0011

[9] 1001 1101 [10] 0101 0100 [11] 1101 1010 [12] 0011 1000 [13] 1011 0110 [14] 0111 1111 [15] 1111 0001 For example, codeword

[0] bit pattern
We will explain how to restore the bit pattern on the receiving side when 00001110 is transmitted. 1) If 3 bits are lost, the receiving side will write 0×00××
If a bit pattern is detected in the form of

Only in the case of [0]. Therefore, the code word

It is possible to restore [0]. In the case of 2 bit loss like 0x001x10,
Of course, in the case of one bit loss such as 0x001110, the same as above,

There is no such bit pattern among the bit patterns corresponding to code words other than [0], so the code word

[0] can be recovered. In addition, except for case 3 described below, when there is no bit pattern corresponding to a bit pattern other than the lost bit, it is assumed that the original bit pattern cannot be restored from this bit pattern as an [error detection]. Therefore, no code word is output in this case. 2) In case of 2 bit loss + 1 bit error, 0x
When two lost bits (×) and one error bit (0) are received, such as 00×10, (×)
There is no bit pattern that matches the code table except for

There can be two types of codes: [0] and a completely different code word [6]. Therefore, it is determined that the received code cannot be determined and there is an error, and the code word is not outputted due to [error detection]. 3) 0 if 1 bit loss + 1 bit error
When a loss of × and an error bit of 0 are given, such as ×000110, there is no match in Table 1 among the bit patterns excluding ×, but the fifth bit (0) is inverted. When it becomes (1),
code word

Since the pattern can match [0], the received signal is

It is judged as [0]. Even if other bits are inverted, there will be no matching bit pattern. Therefore

There is no mistake in judging it as [0]. If we express 1), 2), and 3) above in general form,
When the minimum inter-code distance of the code signal is d, the number S of lost bits is within d-1, and the number of error bits is [(d-S-1)/2] bits (however, [ ] is a Gaussian symbol, meaning an integer not exceeding the number enclosed by this symbol), then the code between the received and level-identified signal and multiple code signals prepared in advance on the receiving side. By comparing the arrays, the lost bits are recovered, and if there is an error bit, the position of the error bit is detected and the codes 0 and 1 corresponding to that position are inverted to codes 1 and 0, respectively. This is equivalent to restoring the code signal transmitted from the transmitting side. Next, an example of a method for detecting a framing code using the present invention will be described with reference to the drawings. FIG. 1 shows a signal waveform when a character signal is superimposed on a certain horizontal scanning period during the vertical retrace period of a television signal. In FIG. 1, 101 is a horizontal synchronizing signal;
2 represents a color burst signal. These signals are the same as regular television signals. 103 is a clock line in for synchronizing the clock of the teletext signal, and 2 bytes are inserted. Reference numeral 104 is a framing code for synchronizing framing, which has a specific pattern indicating the framing code, and 1 byte is inserted. 105 is a 1-byte service identification signal that identifies the type of information being transmitted, and 106 is a 2-byte packet identification signal that indicates the display position and nature (whether or not it is a colored signal, etc.) of the information in the horizontal synchronization period. It has been inserted. 107 is a 31-byte character information signal to be transmitted. In the character signal 107, character pattern information may be sent as is, or a display control code for displaying the corresponding character information on the receiving side may be transmitted. The service identification signal 105, packet identification signal 106, and character information signal 107 are (8, 4)
It has become a symbol. The length a from the falling edge of the horizontal synchronization signal 101 to the first bit of the clock line in 103 is 56
Bit. The length b of one packet signal sent during one television horizontal scanning period is 37 bytes (296 bits). The length c of one television horizontal scanning period is
Set to 364 bits. FIG. 2 shows an enlarged view of the clock line 103 and the framing code 104. The clock line 103 is as shown in the figure.
101010...10 16 bits, and the subsequent framing code 104 has a bit pattern of
It is 8 bits of 11100101. FIG. 3 shows an example of a configuration for detecting this framing code 104. 108 is an input terminal to which a television image signal including a character signal as shown in FIG. 1 is supplied. 109 is a synchronization signal generation circuit that generates various synchronization signals and clock pulses from the horizontal synchronization signal 101 and clock line 103 included in the image signal. Reference numeral 110 denotes a framing pulse generation circuit which individually compares the input signals to see if there is a bit pulse that matches the 8-bit bit pattern of the framing signal. Previously, when the bit patterns matched completely,
Framing codes could only be detected when the bits were different. When detecting framing codes,
A period in the vicinity of the framing code is gated out from the image signal, and this gate period is set to be a part of the clock run-in period + the framing code period + several bits. When detecting the framing code 104 during this period, by applying the method of the present invention, the framing code can be detected more appropriately than before. As shown in Figure 3, A, I, U, E, O, Ka, Ki,
Q, ke are 8-bit periods taken out while moving one bit at a time. In the figure, the numbers shown in ○ for each period indicate the bit pattern of that period and the bit pattern of the framing code. Indicates the intersymbol distance between. That is, the inter-symbol distance between the bit pattern of period A and the bit pattern of the framing code is 5 bits, and the inter-symbol distance between the bit pattern of period A and the bit pattern of the framing code is 3 bits.
The inter-symbol distance between the bit pattern of period C and the bit pattern of the framing code is 4 bits, and so on. As is clear from this figure, the gate is started 6 bits before the start of the framing code 104,
In other words, if we extract the bit pattern by gating from the beginning of period C and search for the framing code from it, the framing code will ensure a minimum inter-symbol distance of 4 bits for the bit pattern of any 8-bit period. can do.
Therefore, the framing code 104 can also be restored by recovering lost bits as explained above using the 8,4 code as an example. Since there may be an error in gate timing due to jitter in the received signal, it is safe to search for the framing code from 3 bits earlier. In other words, a signal selection error of ±3 bits is allowed. In the currently used 8,4 extended Hamming code, there is a possibility that the timing of the framing code can be detected by the code itself following the framing code, for example, the service identification signal 105 and the packet identification signal 106. are doing. That is, when a slip in framing timing occurs between adjacent codes, the number of codes C(k) that has the same pattern as any code word when viewed after the slip timing is (where k: (Number of slip bits) When k = 1, 7, C(k) = 0 (1) When k = 2, 3, 4, 5, 6, C(k) = 16 (2). That is, in the case of equation (1), there is no code, and in the case of equation (2), there are 16 codes for each k. Here, since there are k=5 types, 5×16=80 code words are generated. On the other hand, the total number of slips is 16
Since there are ×16 × 7 = 1792 ways, the probability that the framing timing is incorrect and the phase straddles the original code word and becomes another code word is 80/1792 = 4.4 × 10 ^-2 , that is, It will be about 1/25.
If the framing timing cannot be determined using the received framing code, if one tries to detect the frame timing by checking the following code word, the frame timing can be detected with an error rate of 4.4×10 ^-2 . A more detailed configuration of the above-described framing code detection device is shown in FIG. A synchronization signal reproducing circuit 109 reproduces a vertical synchronization signal and a horizontal synchronization signal from the television signal, and generates a clock pulse 2 whose phase is locked to the clock line in 103.
03, a write start signal 204 and a write clock signal 303, which are generated three bits before the framing code 104, are generated and output. Reference numeral 111 denotes an A/D converter provided to identify the level of the input code signal, which A/D converts the input signal from the input terminal 108 every time the clock pulse 203 occurs, and identifies the level in multiple stages. , outputs level information as a digital signal according to the level value. When A/D conversion is performed using 4 bits, the level information becomes information representing 16 levels from "0" to "15". The details can be the same as those described in the specification of Japanese Patent Application No. 57-55464. 112 is a DMA controller, and 113 is a random access memory, which stores the output of the A/D converter 110, that is, the level information of the input signal whose level is identified in multiple stages according to the level. 114 is a computer, 115 is a display memory,
116 is a cathode ray tube for screen display. Since it is configured as above, input terminal 1
The input signal input from 08 is sent to the A/D converter 11.
0, the level is identified at every timing of the clock pulse 203 from the synchronizing signal generator 109, and the identified level information is extracted. This information is transferred to and stored in RAM 113 via DMA controller 112. At that time, there are 37 bytes of digital signals in one horizontal scanning period, but 35 bytes excluding 2 bytes of clock run-in and 1 byte are now stored. ing.
The reason for writing an extra byte here is to write 3 bits before the framing code 104 when detecting the framing code 104, and to take in 5 bits more at the end of the packet for safety reasons. FIG. 5 shows a flowchart of error correction detection processing when detecting the framing code 104 when the receiver receives one packet worth of signals. When the computer 113 detects that the data transfer from the DMA controller 112 is completed,
Detects the number of lost bits in the input signal data. Here, when a 4-bit converter is used as the A/D converter 111, level information is input in stages from "0" to "15" depending on the level of the input signal, so the level information is “1”, “2”,
When the signal is “3”, “4”, “5”, or “6”, the level of the original input binary signal is determined to be “0”, and “9”,
When it is "10", "11", "12", "13", or "14", the level of the original input binary signal is determined to be "1". Then, the level information is “0”, “7”, “8”,
When it is "15", it is determined that the missing bit is "X".
In this way, a binary data signal block and a lost bit block are prepared in parallel. Next, all binary data and lost bits are temporally shifted in the previous packet data by the number of shifts required to reach the beginning of the framing code, and the first 8 bits are used as the framing code, and the following 272 bits are used as the packet. Treat it as data. However, if the first 8 bits do not become a framing code pattern even if the lost bits are recovered using the method described above, the search for the framing code is restarted from the beginning of the received bits. This search is performed by shifting backwards to 6 bits from the beginning, taking into account bit shifts of ±3 bits. If the framing code cannot be detected even after such an operation, the presence of the framing code is detected from the 8,4 code existing in the period of 3 bytes following the framing code. That is, 2 bytes of data 16 bits after the first bit are examined, and if the 2 bytes are an (8, 4) code containing only erased bits, the timing is determined. Those 2 bytes are (8, 4)
If it is not a sign, shift it one more bit and perform the same operation. The number of times this operation is K
If K=8, it is determined that the framing timing could not be obtained. When the framing timing can be obtained with K<8, the bit position from the received first bit of the actual framing code is the K-th bit. In this way, the K+8th bit from the received bit becomes 34 bytes of packet data. Thereafter, the service identification signal, packet identification signal, etc. are decoded by the aforementioned lost bit recovery and error correction operations, and the desired control is performed. Similar error correction detection including recovery of lost bits is performed for other (8,4) codes. For other pattern data, the level threshold is set to "8", and the upper and lower portions thereof are set to "1" or "0", and the previously determined lost bits are converted into a binary signal and used. Considering that "1" signals are more noticeable on the screen, and "0" signals appear more frequently than "1" signals in the pattern signal itself, lost bits are more noticeable in the pattern data section. A method in which all values are fixed to "0" is visually more preferable. An example of a time chart for detecting the above-mentioned framing timing is shown in FIG. 301 is the timing of the beginning of the received code signal packet that has arrived at the receiver with a normal phase. Kurotsukranin 103 is the latter 3 here.
Take in extra bits. 302 indicates a 14-bit period (3 bits + 8 bits + 3 bits) within which the framing code is searched. 303 is a search range for reproducing the framing timing from the packet identification signal 105 composed of (8,4) codes. Here, the reason why the framing timing was not regenerated using the packet identification signal 105 is because
This is because when the framing signal 104 is erroneous, there is a high probability that the adjacent packet identification signal 105 will be erroneous. Since the present invention is configured as described above,
Ghost cancellers, phase equalizers,
This has the effect that correct code signals can be reproduced without using an automatic slice level adjuster, automatic clock phase adjuster, etc. Furthermore, when applying the method of the present invention to play back the framing code, if an extra signal is received in advance, the framing code can be played back using the method of the present invention. This has the advantage of eliminating the need for a hard framing reproduction circuit. In addition, since the (8,4) code of the pattern-based teletext signal itself has the ability to reproduce the framing timing with an error rate of 4.4 × 10 ^-2 , even if the framing timing cannot be reproduced by the framing code, the subsequent The framing timing can be recovered from the signal.

[Brief explanation of drawings]

Figure 1 is a diagram showing the signal arrangement within one TV horizontal scanning period of a teletext signal, Figure 2 is a diagram showing the signal arrangement near the framing code, and Figure 3 is the configuration of the framing code reproducing circuit. 4 is a more detailed configuration diagram of the framing code reproducing circuit according to the present invention, FIG. 5 is a flowchart showing the procedure for reproducing the framing code, and FIG. 6 is a time chart for detecting the framing timing. . 101...Horizontal synchronization signal, 102...Color burst signal, 103...Clock run in, 10
4...Framing code, 105...Service identification signal, 106...Packet identification signal, 107
...Character information signal, 108...Input terminal, 109
... Synchronization signal generation circuit, 110 ... Framing pulse generation circuit, 203 ... Clock pulse, 2
04...Start signal, 303...Write clock signal, 111...A/D converter, 112...
DMA controller, 113... Random access memory, 114... Computer, 115...
Display memory, 116... Cathode ray tube for screen display,
301... Timing of the beginning of the packet, 302
...Framing code detection range, 303...Framing timing search range.

Claims (1)

[Claims] 1. On the transmitting side, a code signal consisting of a binary signal of 0 and 1 with a parity bit added is transmitted to the receiving side via a transmission path that may cause waveform distortion, and the receiving side , a coded signal restoration method that makes it possible to restore the coded signal from the received signal of the transmitted signal, the level of the signal waveform of the received signal being level-identified into a plurality of finite stages of three or more; Based on the identification result, a bit for which it is unclear whether the level of the signal waveform of the received signal corresponds to 0 or 1 of the binary signal is positioned as a lost bit, and the minimum inter-symbol distance of the code signal is determined as d. Then, the number S of lost bits is within d-1, and the number of error bits is [(d-S-1)/2] bits (where [ ] is a Gaussian symbol, and the symbol is (meaning an integer not exceeding the number enclosed), the code arrangement of the received and level-identified signal is compared with a plurality of code signals prepared in advance on the receiving side. In addition to regenerating the lost bits, if there is an error bit, the position of the error bit is detected and the codes 0 and 1 corresponding to the position are inverted to codes 1 and 0, respectively, so that the data can be transmitted from the transmitting side. A code signal restoration method characterized in that the transmitted code signal is restored.