JP3497435B2 - Data transmission error monitoring system, data transmission device, data reception device, and data transmission error monitoring method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for monitoring a transmission bit error when a data packet is relayed and transmitted over a plurality of sections. The present invention relates to a monitoring system for monitoring a transmission error in each relay section in a serial transmission system, a transmitter thereof, a receiver thereof, or a monitoring method.

[0002]

2. Description of the Related Art In recent years, the amount of data packets transmitted between various devices such as a computer, a storage device, and a router has been increasing, and a Gbps-class transmission line capacity is required. In some cases, the distance between the devices exceeds 100 m, and in some cases reaches several hundred km. For this reason, there are increasing examples of performing serial data transmission by using an optical fiber in the transmission path between devices. Generally, for serial optical transmission of a packet, code conversion suitable for optical transmission is performed, and at the same time, conversion into a continuous data string in which an appropriate number of "idle sets" are sandwiched between packets. One of such methods uses nB / mB code conversion (n and m are natural numbers of n <m).

In this method, every n bits is converted into an m-bit codeword according to a predetermined rule, and then "1" is set.
And the continuation of "0" are prevented for the convenience of clock extraction, and the occurrence frequencies of "1" and "0" are made as equal as possible to facilitate the adjustment of the discrimination level of the optical receiver. further,
A "special codeword" is defined from codewords not used for code conversion, and an "idle set" including this is inserted in the gap between packets.

The Manchester code, which is well known as a transmission path code, has the original data "0" set to "01" and "1" set to "1".
It is a 1B / 2B code that is converted to 0 ". 4B / 5
B code is FDDI (Fiber-distributed data interfac)
e: IS09314-2, IS09314-3.1992) and Fast Ethenet (100B
ASE-X: ANSI / IEEE 802.3u, 1995), 8B / 1
OB code is Fiber Channel (ANSI X3T)
11, Fiber Channel Physical and Signaling Interface,
1994) and Gigabit Ethernet (GigabitEtherne
t: ANSI / IEEE 802.3z, 1998).

An error monitoring method in a conventional data transmission system using 8B / 10B code conversion will be described with reference to FIGS. 11 to 14 by taking Gigabit Ethernet as an example. In addition, 8B / 10 used here
Details of the B code conversion rule are described in, for example, JP-A-59-10056. FIG. 11 shows an example of a conventional data transmission system, in which 110 is a transmitter, and 1 is a transmitter.
20 is a repeater, 130 is a receiver, 5a, 5
b is an optical transmission line. Further, 111 is an encoder that performs 8B / 10B code conversion, 121 is a reproducer that reproduces a received data string, and 129 and 139 are decoders that perform 8B / 10B code inverse conversion.

A data packet is input to the input end 1 of the transmitter 110. The encoder 111 converts each byte (8 bits) into a 10-bit code word based on the 8B / 10B code conversion rule, and further, between packets converted into code words (including the case where there is no packet input). After inserting a plurality of 20-bit idle sets into the optical transmission line 5a, a continuous data string is transmitted.

FIG. 12 shows the format of such a continuous data string.
-Shows the mat. 10-bit codewords each D^k, D^{k + 1},… (K is a natural number) Are output continuously. kth codeword D^kIs a bit D^k ₀, D^k ₁, ..., D^k ₉ Are transmitted serially in this order, followed by the next codeword
D^{k + 1}First bit of D^{k + 1} ₀Is transmitted.

FIG. 13 shows an example of the 8B / 10B code conversion rule. For convenience, each of the 256 original data is set as Dx. It is expressed as y (0 ≦ x ≦ 31, 0 ≦ y ≦ 7). When converting this into a 10-bit codeword, two conversion rules are used according to the history (whether "0" or "1" was more frequent) so that the occurrence frequencies of "0" and "1" are equalized. Use properly. This history is called running disparity (RD).
When there are many "0", it is converted into the code word in the "current RD-" column in the figure, and conversely, when there are many "1", "curren"
t RD + ”column code word.

In a 10-bit code word, the number of "1" s and "0" s is the same among 1024 ways, or one is 2
As many as three are selected, and among the latter, codewords with many "1" s are assigned to the "current RD-" column, and codewords with many "0" s are assigned to the "current RD +" column. Assuming "the situation where there are many" 0 "s" in the initial state, the history (RD) at the boundary between codewords is "RD-" (the situation where there are one "0") and "RD +" ( There is one more "1")
Will be either.

For example, the original data "00000000" (D0.0) is converted into "1001110100" if there are many "0" s and "0110001011" if there are many "1" s. This 10-bit codeword is "0" and "1"
The history (RD) does not change because the number of

On the other hand, for example, the original data "00000011" (D3.0) is converted into "1100011011" if there are many "0" s and "1100010100" if there are many "1" s. This 10-bit codeword is "0" and "1"
The history (RD) is inverted because the numbers of the two are not equal.

In addition to these 256 kinds of original data, a special code word K28.5 is defined. Bit patterns "0011111" and "1100" included in this
000 "is not included in any code word Dx.y corresponding to 256 original data, and does not occur even if it crosses the boundary between any two code words. Therefore, on the receiving side, the above bit pattern is By detecting, the delimiter position of the 10-bit code word in the continuous data string can be specified.

FIG. 14 shows a format of a code word string when a packet is transmitted in a Gigabit Ethernet using such 8B / 10B code conversion. All packets are 2 bytes in size (2 codewords = 2
It is adjusted to an integer multiple of 0 bit), and the history (RD) after the end of the packet is the same as when the packet started
D- ". nth packet P _n
In the period I _n from the end of the transmission of ₁ to the start of the transmission of the next packet P _{n + 1} , one or more sets of 20-bit idle sets are inserted.

This idle set has a special code word K2.
It is a combination (set) of 8.5 and code word D16.2. -/ K28.5 / D16.2 / (expression by code word) 00111110101001000101 (bit string to be transmitted) "-" added to the head of the expression by code word is "RD-" just before the transmission. It means that. K28.5
Since D and 16.2 are code words that invert the history (RD), the history (RD) after sending the idle set is always returned to the same "RD-" as before sending.

Next, the packet thus converted into a continuous data string is relayed and transmitted from the transmitter 110 to the receiver 130 via a repeater (repeater) 120. A method of monitoring a transmission bit error in the conventional data transmission system shown in FIG. 11 will be described. In the repeater 120,
The regenerator 121 amplifies the signal propagated through the optical transmission line 5a and attenuated, and if necessary, performs waveform shaping by identification reproduction and clock transfer, and then outputs it to the optical transmission line 5b. The clock transfer is to transmit a continuous data string by replacing the clock cycle extracted from the input data string with the free-running clock cycle held by the repeater 120 itself, and the jitter (clock cycle fluctuation) is generated in each repeater. Can be prevented from accumulating. Since the free-running clock cycle is slightly different for each repeater, it is adjusted by inserting and removing the idle set between packets.

Here, the output of the regenerator 121 is also input to the decoder 129 to perform 8B / 10B code inverse conversion from the continuous data sequence. First, from the continuous data string, the special code word K
28.5 is detected, the codeword delimiter and the history (RD) at that time are recognized, and then the original 8-bit data is extracted from the received 10-bit codeword based on the recognized history (RD). Decrypt. When a single bit error occurs in the optical transmission line 5a, the received codeword and the history (RD) do not match and the error occurrence can be detected.

For example, D2.0 "1011010100" transmitted by "RD-" does not change the history (RD).
The code word "-" follows. If a transmission bit error occurs at the 10th bit and it is decoded as "1011010101" (D2.2), this code word inverts the history (RD), so the next code word should be the "RD +" code word. There is a contradiction. In the receiver 130 as well, an error occurrence can be similarly detected by inputting the continuous data string input from the optical transmission path 5b to the decoder 139.

[0018]

However, in such a conventional data transmission error monitoring system as described above, it is often impossible to detect continuous bit errors, and whether or not a bit error has occurred before or after the repeater is
There is a problem in that it cannot be specified by the receiver. The two problems will be described below.

The history (RD) is "0" in the continuous data string,
In order to indicate which one is more, "2" consecutive "01" in the same code word is "10" due to consecutive bit error.
Even if it is detected, there is no contradiction in the history (RD). For example, if a transmission error occurs in the first and second bits of "0100101011" (D2.0) transmitted by "RD +", "10"
When "00101011" is received, it is considered that the codeword D1.0 is normally received, and the error occurrence cannot be detected.

Further, in the conventional data transmission system described with reference to FIG. 11, even if the receiver 130 detects an error occurrence, it cannot be specified in which of the optical transmission lines 5a and 5b it is caused. In order to identify the faulty section, it is necessary to provide another line for monitoring whether or not the decoder 129 of the relay 120 has detected an error and notify the receiver 130 of it.
Maintenance cost is required. The present invention has been made in view of these circumstances, and an object thereof is to detect a continuous bit error, and further, when a packet is relay-transmitted over a plurality of sections, whether or not an error occurs in each section. It is an object of the present invention to provide a data transmission error monitoring system which can easily detect a data transmission error and which is economically efficient.

[0021]

In order to achieve such an object, a data transmission error monitoring system and a data transmission error monitoring method according to the present invention provide a predetermined nB data packet every n bits (n is a natural number). / MB code conversion rule (m is a natural number of n <m) is used to convert to an m-bit code word, and i bits (i is m It is premised on a data transmission error monitoring system in which one or a plurality of multiple idle sets are inserted and these are serially transmitted as a continuous data string from a data transmission device to a data reception device.

In the data transmission apparatus, the parity counter divides the continuous data sequence into n '(n' is a factor of i, 1 <n '<m) data series and divides the data series by 1 for each data series. The number of bits of parity is counted, and the encoder converts n'number (n 'bits) of parity output from the parity counter into an m-bit parity code word according to a predetermined n'B / mB code conversion rule, and a replacer Then, one or a plurality of consecutive idle sets are detected from the continuous data string, and this is replaced with a j-bit (j is a multiple of i) parity set including the parity code word output from the encoder and transmitted. Further, in the data receiving device, the detector detects the parity set from the received continuous data string, and the decoder detects n ′ from the parity set.
The parity counter is restored, the received continuous data sequence is divided into n'data series by the parity counter, the parity set is excluded for each data series, and the parity is counted. The counter outputs are compared with each other.

The number of data series n'may be m or more. In this case, in the data transmission device, the parity counter uses the number of continuous data strings of n '(n' is a factor of i, m ≦
n ') data sequences are divided into 1-bit parities for each data sequence, and the encoder encodes a plurality of n' (n 'bits) of parities so as to be smaller than m bits. It may be divided into count values, and these count values may be converted into m-bit parity code words by a predetermined code conversion rule corresponding to the number of bits.

The n ′ data series are selected so that the bit interval on the serial continuous data string of the data belonging to at least one series periodically repeats a plurality of values and the parity remains unchanged before and after the idle set. But it is okay. Furthermore, (n, m, i, n ′) = (8,10,2
0, 5) and the minimum bit interval may be 4. Further, the parity code word may always match any one of the code words generated by the nB / mB code conversion rule, and the parity set may include an identification code word in addition to the parity code word. .

[0025]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of a data transmission error monitoring system according to a first embodiment of the present invention. In this data transmission error monitoring system, a transmission bit error is monitored when a continuous data string is relay-transmitted from a data transmission device (transmission end) 11 to a data reception device (reception end) 13 via an optical transmission path 5. . In addition, in the present invention, as described above, a predetermined nB / mB code conversion rule (m is a natural number of n <m) for each n bits (n is a natural number) of the data packet.
Is converted into an m-bit code word by the above-mentioned method, and an i-bit (i is a multiple of m) idle set consisting of 1 to a plurality of m-bit code words is set to 1 to between the data packets converted into the m-bit code word. A description will be made on the premise of a data transmission error monitoring system in which a plurality of data are inserted and serially transmitted as a continuous data string from a data transmission device to a data reception device.

In the data transmitter 11, the continuous data string is divided into n '(n' is a factor of i, 1 <n '<m) data series, and 1-bit parity is set for each data series. Parity counter 2 for counting and parity counter 2
'N (n' bits) parity (count value) output by
To an m-bit parity code word according to a predetermined n'B / mB code conversion rule, and 1 or a plurality of consecutive idle sets are detected from a continuous data string, and this is output from the encoder 3. J bits (j is i
And a replacement unit 4 for sending the data after replacing it with a parity set.

The data receiving device 13 includes a detector 6 for detecting a parity set from the received continuous data sequence and a decoder 7 for restoring n'parities from the parity set.
And a parity counter 8 that divides the received continuous data sequence into n ′ data series and excludes the parity set for each data series to count the parity, and the output of the decoder 7 and the output of the parity counter 8. A comparator 9 for comparison is provided.

Next, as an operation of the present embodiment, a case of handling continuous data based on the Gigabit Ethernet standard will be described as an example. To the input terminal 1 to the data transmitter 11, continuous data obtained by 8B / 10B-encoding a packet based on the Gigabit Ethernet standard is input. Figure 2
Shows the format of this continuous data string. In the following example, the idle set is 20 bits (i = 20), the data packet is 8 bits (n = 8), the code word is 10 bits (m = 10), and the number of divisions of the continuous data string (data series This corresponds to the case where the number) is 4 (n ′ = 4).

As described with reference to FIG. 14, the idle set can be divided into code sets of 20 bits (i = 20). That is, code sets S ^k , S ^{k + 1} , ... (K is a natural number) each having 20 bits are continuously input. Hereinafter, the kth code set S ^k
The p-th bit (0 ≦ p ≦ 19) of is written as S ^k _p . This continuous data string is guided to the parity counter 2, divided (separated) into four (n '= 4) data series, and 1-bit parity is counted for each series. This parity (count value) is reset to the initial value by the signal input from the replacing device 4 described later.

Next, how to select four data series will be described. For example, if the k-th code set S ^k is an idle set, (S ^k ₀ , S ^k ₁ , ..., S ^k ₁₉ ) = (D ^2k ₀ , ..., D ^2k ₉ , D ^{2k + 1} ₀ , ..., D ^{2k +1} ₉ ) = (0,0,1,1,1,1,1,0,1,0,1,0,0,1,0,0,0,1,0,1).

Simply interleave every 4 bits,
When separated into four data series, data series # 0 (S ^k ₀ , S ^k ₄ , ..., S ^k ₁₆ ) = (0, ₁ , ₁ , 0, 0) Data series # 1 (S ^k ₁ , S _{^{k 5, ..., S k 17}} ) = (0,1,0,1,1) data series _{^{# 2 (S k 2, S}} k 6, ..., S k 18) = (1,1,1,0, 0) data series _{^{# 3 (S k 3, S}} k 7, ..., S k 19) = a (1,0,0,0,1).

The bit intervals on the continuous data sequence of each data series are all equal to 4, which is the so-called "BIP".
4 ”(Bit Interleaved Parity 4). Here, assuming even parity ('1': parity is '1' when the number of appearances is odd and '0' when it is even), the parity of data series # 0 and # 3 changes before and after this idle set. However, the parity of the data series # 1 and # 2 changes.

Therefore, for example, S ^k ₁ and S ^k ₂ are exchanged, and the data series # 0 (S ^k ₀ , S ^k ₄ , ..., S ^k ₁₆ ) = (0, 1, 1, 0, 0) data series # 1 ′ (S ^k ₂ , S ^k ₅ , ..., S ^k ₁₇ ) = (1,1,0,1,1) data sequence # 2 ′ (S ^k ₁ , S ^k ₆ , ..., S ^k ₁₈ ) = (0,1,1,0,0) data sequence _{^{# 3 (S k 3, S}} k 7, ..., S k 19) separated = (1,0,0,0,1) four data series that You may.

The bit intervals on the continuous data sequence of the data series # 1 ', # 2' are (3, 4, 4, 4, 5) (5, 4, 4, 4, 3) cyclically repeated, respectively. ing. Since the number of occurrences of "1" is an even number in all data sequences, the four parities do not change before and after this idle set.

The encoder 3 of FIG. 1 inputs the 4-bit (n '= 4) parity present value for each code set and outputs 4B.
A 10-bit (m = 10) parity code word is generated by / 10B code conversion. FIG. 3 shows an example of the 4B / 10B code conversion rule. The 4-bit parity'abcd '(a, b, c, dε {0, 1}) is converted into a 10-bit parity codeword P'00aAbBcCdD'.

Here, the uppercase letters indicate symbols in which the lowercase letters are reversed. For example, the parity “0000” is converted into “0001010101”. This is equal to the code word D23.2 when the history is "more one'1 '" (RD +) in the 8B / 10B code described above. In the figure, this was described as + D23.2.

Similarly, the parity from "0000" to "111"
All 16's of 1'are matched with any of the 8B / 10B codewords. FIG. 4A shows a circuit configuration example of the encoder 3 that realizes the 4B / 10B code conversion. In addition to the "00aAbBcCdD" described with reference to FIG. 3, for example, "00AaBbCcDd" 00Aab
BcCDd "00BbAaDdCc 'etc., abcd
The order of may be exchanged, or a code conversion in which the order of uppercase letters and lowercase letters is arbitrarily inverted may be used.

The replacer 4 of FIG. 1 detects an idle set (/-K28.5/+D16.2/) of the Gigabit Ethernet standard from the continuous data string input from the input terminal 1 and keeps it approximately constant. In the cycle, the second code word (+ D16.2) in the idle set is replaced with the parity code word P′00aAbBcCdD ′ output from the encoder 3. That is, the idle set (/-K28.5/+D16.2/) is replaced with the parity set (/-K28.5/P/) and the result is sent to the optical transmission line 5. FIG. 2B shows the format of the data string output from the replacer 4.

Further, the replacer 4 sends an instruction to the parity counter 2 to reset it when the idle set is replaced with the parity set. Therefore, the parity set that has been replaced and inserted includes the parity information from the code set immediately after the previous replacement and insertion to immediately before the parity set. As described above, if the bit interval on the continuous data string is cyclically repeated multiple values and the data series is cleverly separated so that the parity does not change before and after the idle set, it is necessary to remove the idle set due to replacement. No parity correction circuit is required. If this is not the case, parity correction associated with the idle set removal is required on the transmission side or the reception side.

In the data receiving device 13 of FIG. 1, the continuous data string from the optical transmission line 5 is input to the detector 6 and the parity counter 8 as well as to the output terminal 10 as it is. The detector 6 detects the parity set from the data string, guides the parity code word P (10 bits) to the decoder 7, and notifies the parity counter 8 of the parity set detection. The decoder 7 restores the original 4-bit parity count value and outputs it to the comparator 9.

On the other hand, the parity counter 8 divides the input continuous data sequence into the above-mentioned four data series and counts the parity for each series. Then, based on the parity set detection notification from the detector 6, the count value up to immediately before the parity set is output to the comparator 9, the counter is initialized, and counting is performed again from the code set immediately after the parity set. The comparator 9 outputs the 4-bit parity value on the transmission side output from the decoder 7 and the parity counter 8
And compare it with the 4-bit piedy value on the receiving side,
Detect the presence of errors.

FIG. 4B shows the circuit configuration of the detector 6.
20 consecutive bits of the input data string are input to Q0 to Q19. AND of output stage only when parity set P is detected
Output is generated at the gate. Further, the circuit configuration of the decoder 7 is shown in FIG. According to the 4B / 10B code conversion rule shown in FIG. 3, no logic circuit is necessary for detecting'abcd '.

As described above, the data described with reference to FIG.
According to the transmission error monitoring system, a continuous data string
Can be divided into data series and check each parity
Therefore, bit errors occur continuously in the optical transmission line 5.
Can also be detected. 4 data series mentioned above
For # 0, 1, 2, and 3, the bit interval W in the same sequence is
All 4's, so correct continuous errors up to 4 bits
it can. On the other hand, in # 0, 1 ', 2', 3 the same sequence
Is the minimum bit interval W in the sequence # 1 '(S^k ₂And S ^k _FiveBetween)
And # 2 '(S^k ₁₈And S^{k + 1} ₁3), so 3 bits
The continuous error up to can be detected correctly.

Since the data transmitter 11 adds the parity information to the transmission data, it can be specified that the data receiver 13 detects a transmission error in the optical transmission line 5. FIG. 5 is a block diagram showing a partially modified example of the data transmission error monitoring system according to the first embodiment. In the figure, 11 is a data transmitter and 13 is a data receiver, which is an embodiment of the present invention described with reference to FIG. The same components as those of the conventional data transmission system described with reference to FIG. 11 are designated by the same reference numerals, and the description thereof will be omitted.

In the repeater 220, the regenerator 121
A signal propagated through the optical transmission line 5a and amplified is amplified, and if necessary, waveform shaping is performed by identification reproduction, and the signal is sent to the optical transmission line 5b via the data transmission device 11. Receiver 230
Then, the continuous data string input from the optical transmission line 5b is input to the decoder 139 via the data receiving device 13. The data receiving device 13 detects only the occurrence of a transmission error in the optical transmission line 5b. On the other hand, by inputting the continuous data sequence transmitted through the data receiving device 13 to the decoder 139, it is possible to detect a bit error between the encoder 111 and the decoder 139 as in the conventional case. This is because, as described with reference to FIG. 3, all the parity codewords P have the history (RD) as “RD +”.
Since the 4B / 10B code conversion rule is selected so that it is mapped to the 8B / 10B codeword that is inverted from RD to “RD-”, the correct history (RD) at the time of encoding even after the idle set is replaced with the parity set. Is maintained.

Therefore, the receiver 230 determines in which section the optical transmission lines 5a and 5b an error has occurred by combining and determining the two error detection information output from the data receiving device 13 and the decoder 139. It is possible. As in the conventional system, a relay 220 is provided with a decoder,
Furthermore, it is not necessary to transmit the information to the receiver via another line. In addition, the above-mentioned four data series # 0, 1 ',
If 2 ′ and 3 are used, the parity count value does not change before and after the idle set, so that the regenerator 121 of the repeater 220 can perform clock transfer (idle set insertion / removal). That is,
It is possible to prevent the accumulation of jitter by the repeater.

In the above description, the number n'of the data series to be separated is 4, but the number is not limited to this, and the maximum n'can set the n'B / 10B code conversion rule as follows. You should choose. First, as described above, it is desirable that the parity of each sequence be unchanged with respect to the insertion / extraction of the idle set. For that, the idol set is i
Given the bit length (i is a natural number), n'needs to be a factor of i.

The parity code word generated by the n'B / 10B conversion is the fiber channel standard or gigabit.
In the Ethernet standard or the like, it is desirable that it does not match the code word used as a special code set (for example, idle set) in combination with the special code word K28.5. If this condition is not satisfied, for example, as shown in FIG. 6, the wiring in the data receiving device 13 is changed to change the output terminal 10
Is changed to output continuous data through the detector 6 ', and when the special code set is detected by the replacing unit 4, it is replaced with a different predetermined code set in advance, and the original special code set is detected by the detector 6'. It is necessary to return to and output.

The parity code word generated is 8B / 10.
It is desirable to satisfy the B code history (RD) rule. If this condition is not satisfied, it is necessary to adopt the configuration shown in FIG. 6 again, and when the detector 6 ′ detects the parity set, restore the original idle set and lead it to the output terminal 10. Further, it is convenient that the parity code word always matches any one of the code words generated by the 8B / 10B code conversion. If this condition is not met, a "out-of-sync" error can occur when a continuous data string containing a parity set is input to a conventional device. This is because, for example, in the Gigabit Ethernet standard, even if the running disparity (RD) is correct, if a codeword that does not exist in the 8B / 10B code conversion is detected intermittently a plurality of times, the state shifts to "out of sync". This is because. When this condition is not satisfied, it is still necessary to take the configuration shown in FIG.

The 4B / 10B conversion code described with reference to FIG. 3 satisfies all of these conditions. On the other hand, according to the Gigabit Ethernet standard, if the codeword following the special code /-K28.5/ does not match any of the second codewords defined as the special code set (0rdered Set), it is regarded as an idle set. I have decided to decrypt it.
Therefore, if the code conversion rule of FIG. 3 is used, a continuous data string including a parity set may be directly input to the conventional device. That is, since compatibility is maintained and it is possible to mix with conventional devices, even in a large-scale system, the system can be partially renewed as necessary, which is economical.

Next, how to select the maximum n'that satisfies all the above-mentioned conditions will be explained, and it will be shown that it is "5" when the 20-bit length idle set is used. It has a factor of 20 (i = 20) and 10 (m = 10)
Since the maximum value of smaller n ′ is “5”, it suffices to show the existence of the 5B / 10B code that satisfies the condition. The second codeword of the idle set to be replaced with the parity set is "1"
Including 4 and 6 "0", the history from "RD +" to "R
D- ". Therefore, 1B / 2B code conversion (Manchester code) is performed with four consecutive 2-bit sets, and the remaining 2 bits are set to "0" and "0".

That is, 4-bit data'abcd '
Is encoded as '00aAbBcCdD', 'aA00bBcCdD', 'aAbB00cCdD', 'aAbBcC00dD', 'aAbBcCdD00', the running disparity condition is satisfied. Composing two consecutive bits is
Since inversion of "1" and "0" occurs frequently, 8B / 10
This is because it often coincides with any code word generated by the B coding rule.

Of these, '00aAbBcCdD' When ‘AAbBcC00dD’ Satisfies all the conditions.

FIG. 7 shows a 5B / 10B conversion code rule created by combining two such 4B / 10B conversion code rules. The 5-bit parity'stxyz '(s, t, x, y, zε {0, 1}) is converted into the 10-bit parity code word'tTxXyY00zZ' (when s is “0”) or “00xXyYtTzZ '(s. Is "1").

For example, the parity "00000" is converted into "0101010001". This agrees with the 8B / 10B code + D10.7. Similarly, the parity from "00000" to "111"
All 32 patterns of 11 'match with any of 8B / 10B codewords. Moreover, it does not overlap with the second codeword in the special code set of the Gigabit Ethernet standard.

FIGS. 8A to 8C show circuit configuration examples of the encoder 3, the detector 6 and the decoder 7 for realizing the 5B / 10B code shown in FIG. 7, respectively. Although it is more complicated than the circuit configuration of the 4B / 10B code described with reference to FIG. 4, it can be realized with at most about 50 gate circuits.

On the other hand, if such a 5B / 10B code conversion rule is used, a continuous data string can be separated into five data sequences and the respective parities can be transferred. If the idle set is simply interleaved every 5 bits and separated into 5 data sequences, data sequence #a (S ^k ₀ , S ^k ₅ , S ^k ₁₀ , S ^k ₁₅ ) = (0, 1, 1, 0) data series _{^{#b (S k 1, S k}} 6, S k 11, S k 16) = (0,1,0,0) data series _{^{#c (S k 2, S k}} 7, S k 12, S ^k ₁₇ ) = (1,0,0,1) data series #d (S ^k ₃ , S ^k ₈ , S ^k ₁₃ , S ^k ₁₈ ) = (1,1,1,0) data series #e ( S ^k ₄ , S ^k ₉ , S ^k ₁₄ , S ^k ₁₉ ) = (1,0,0,1).

This is the so-called "BIP5" (Bit Interl
eaved Parity5). Since all the bit intervals W in the same data series are 5, continuous errors up to 5 bits can be detected correctly. However, the data series #a, #c, #e
Since '1' appears even number of times, the parity does not change, but since #b and #d are odd numbers, the parity changes. Therefore, when selecting five parity-invariant data sequences for the idle set, the maximum value of the minimum bit interval W in the same data sequence is at least 4 or less.

To make the parity invariant with respect to the idle set, for example, S ^k ₁ and S ^k ₃ are exchanged, and the data sequence #a (S ^k ₀ , S ^k ₅ , S ^k ₁₀ , S ^k ₁₅ ) = (0 , 1,1,0) data series #b ′ (S ^k ₃ , S ^k ₆ , S ^k ₁₁ , S ^k ₁₆ ) = (1, 1, 0, 0) data series #c (S ^k ₂ , S ^k ₇ , S ^k ₁₂ , S ^k ₁₇ ) = ( ₁ , 0, ^0, ₁ ) Data sequence #d ′ (S ^k ₁ , S ^k ₈ , S ^k ₁₃ , S ^k ₁₈ ) = (0, ₁ , ₁ , 0) Data series #e (S ^k ₄ , S ^k ₉ , S ^k ₁₄ , S ^k ₁₉ ) = ( ¹ , 0, 0, 1) may be separated into five data series.

Since the number of appearances of "1" is an even number in any data series, the five parity count values do not change before and after the code set. Since the minimum bit interval W in the same sequence is 3 of the data sequence #b ′ (between S ^k ₃ and S ^k ₆ ) and the data sequence #d ′ (between S ^k ₁₈ and S ^{k + 1} ₁ ), It is possible to correctly detect continuous errors of up to 3 bits.

Further, not only the data series #b and #d but also the elements of the data series #c which originally do not need to be replaced are replaced, and between #b and #c or between #c and #. The minimum bit interval W can be set to “4” (that is, the maximum value) by using a clever technique of limiting the replacement between d. An example is shown below.

Data series #a (S ^k ₀ , S ^k ₅ , S ^k ₁₀ , S ^k ₁₅ ) = ( ₀ , ₁ , ₁ , 0) Data series #B (S ^k ₁ , S ^k ₇ , S ^k ₁₁ , S ^k ₁₆ ) = (0,0,0,0) data series #C (S ^k ₃ , S ^k ₈ , S ^k ₁₃ , S ^k ₁₇ ) = (1,1,1,1) data series #D (S ^k ₂ , S ^k ₆ , S ^k ₁₂ , S ^k ₁₈ ) = (1, 1, 0, 0) Data sequence #e (S ^k ₄ , S ^k ₉ , S ^k ₁₄ , S ^k ₁₉ ) = ( 1, 0, 0, 1) In this way, continuous errors of up to 4 bits can be correctly detected.
It should be noted that if the number of data series n'is large, the number of random errors that can be correctly detected per unit time is large, so that it is possible to correctly observe even a small bit error rate.

In the first embodiment described in detail above, it is assumed that the Gigabit Ethernet standard using 8B / 10B code conversion (the idle set is 20 bits) is used, but the present invention is not limited to this. It is not something that will be done. For example, the idle set of the Fiber Channel standard using the same 8B / 10B code conversion is 40 bits long (4
Code word) '00111110101010100001010101010101010101010' (/-K28.5/+D21.4/-D21.5/-D
21.5 /).

Therefore, by the same method as this, 5
The data series of the book (n ′ = 5) are converted into (S ^k ₀ , S ^k ₅ , S ^k ₁₀ , S ^k ₁₅ , S ^k ₂₀ , S ^k ₂₅ , S ^k ₃₀ ,
S ^k ₃₅ ) = (0,1,1,0,1,0,1,0) (S ^k ₁ , S ^k ₇ , S ^k ₁₁ , S ^k ₁₆ , S ^k ₂₁ , S ^k ₂₆ , S ^k ₃₁ ，
S ^k ₃₆ ) = (0,0,0,0,0,1,0,1) (S ^k ₂ , S ^k ₆ , S ^k ₁₂ , S ^k ₁₈ , S ^k ₂₂ , S ^k ₂₇ , S ^k ₃₂ ，
S ^k ₃₇ ) = (1,1,1,1,1,0,1,0) (S ^k ₃ , S ^k ₈ , S ^k ₁₃ , S ^k ₁₇ , S ^k ₂₃ , S ^k ₂₈ , S ^k ₃₃ ，
S ^k ₃₈ ) = (1,1,0,0,0,1,0,1) (S ^k ₄ , S ^k ₉ , S ^k ₁₄ , S ^k ₁₉ , S ^k ₂₄ , S ^k ₂₉ , S ^k ₃₄ ，
If S ^k ₃₉ ) = (1,0,1,0,1,0,1,0) is selected, it is possible to realize the error monitoring insensitive to the idle set and having the minimum bit interval W of "4". General nB / mB
It can be easily understood from the above description that the case where the coding rule is used can be similarly realized.

Further, in the above description, the number of data series n'is n '<
Although the case of setting m is described as an example, the present invention is not limited to this, and n ′ is set to m or more, and n ′ (n ′ bits) parity is set to a plurality of bits having a number of bits smaller than m bits. It may be divided into count values, and these count values may be converted into m-bit codewords and transmitted. For example, in the case of the above-mentioned fiber channel standard, 10 (n '= 10
= M) data sequences are (S ^k ₀ , S ^k ₁₁ , S ^k ₂₀ , S ^k ₃₀ ) = ( ₀ , ₀ , ₁ , ₁ ) (S ^k ₁ , S ^k ₁₀ , S ^k ₂₂ , S ^k ) ^k ₃₁ ) = (0,1,1,0) (S ^k ₂ , S ^k ₁₂ , S ^k ₂₁ , S ^k ₃₃ ) = (1, 1, 0,0) (S ^k ₃ , S ^k ₁₃ , S ^k ₂₃ , S ^k ₃₂ ) = ( ¹ , 0, 0, ¹ ) (S ^k ₄ , S ^k ₁₄ , S ^k ₂₄ , S ^k ₃₄ ) = (1, 1, 1, 1) (S ^k ₅ , S ^k ₁₅ , S ^k ₂₅ , S ^k ₃₅ ) = (1,0,0,1) (S ^k ₆ , S ^k ₁₆ , S ^k ₂₆ , S ^k ₃₆ ) = (1,0,1,0) (S ^k ₇ , S ^k ₁₇ , S ^k ₂₇ , S ^k ₃₇ ) = (0,0,0,0) (S ^k ₈ , S ^k ₁₈ , S ^k ₂₈ , S ^k ₃₈ ) = (1,1,1, 1) Select (S ^k ₉ , S ^k ₁₉ , S ^k ₂₉ , S ^k ₃₉ ) = (0,0,0,0).

The 10-bit parity counted from these 10 data series is divided into 5 bits (5 <m) corresponding to the 5 data series and divided into 5B /
One of these count values is set to the first using the 10B code conversion rule.
To the second parity code word, and the other count value is converted to the second parity code word. Further, any two codewords of the 2nd to 4th codewords of the 40-bit length idle set may be replaced with these parity codewords and transmitted as a 40-bit length parity set.

It is not necessary to divide the n'bit (m <n ') parity into equal bit count values. For example, 10
The bit validity may be divided into 6-bit and 4-bit count values. In this case, the code conversion rule corresponding to the number of bits of each count value, that is, the 6-bit count value is 6
The B / 10B code conversion rule is used, and the 4-bit count value is 4B
/ 10 / B may be used to generate 10-bit validity codewords.

The Fiber Channel standard is equivalent to a predetermined 32B / 40B data packet for every 32 bits.
Converts to a 40-bit code word according to the code conversion rule, and inserts one or more idle sets consisting of one 40-bit code word between the data packets converted to the code word, and serializes this as a continuous data string. It can also be regarded as a transmission system. Therefore, also in the case of m ≦ n ′, it can be said that the case of dividing the parity information into a plurality of parity codewords and encoding the same is the same as the case of n ′ <m, as described above.

Next, a data transmission error monitoring system according to a second embodiment of the present invention will be described with reference to FIG. FIG. 9 is a block diagram for explaining a data transmission error monitoring system according to claim 7 of the present invention. 5 differs from the embodiment described with reference to FIG. 5 in that a data transmission device 11a is provided after the encoder 111 of the transmitter 310, and a data transmission device 11b is provided in the relay 320 instead of the data transmission device 11. That the data receiving device 13a is provided instead of the data receiving device 13 in the receiver 330.
b and 13b are deployed.

The data transmitting device 11a in the transmitter 310 and the data transmitting device 11b in the repeater 320 are different in the functions of the data transmitting device 11 described with reference to FIG. This replacing device 4 is provided with two consecutive idle sets (/-K28.5/+D16.2/-K28.5/+D) of the Gigabit Ethernet standard from the continuous data string input from the input terminal 1.
16./) is detected, and the second and fourth codewords (+ D16.2) of the two consecutive idle sets are detected at approximately constant intervals,
The predetermined identification code word P'and the parity code word P are replaced.

That is, the second code word (+ D16.
2) is replaced with the identification code word P ', for example,'00a'A'b'B'c'C'd'D'', and the second code word (+ D16.2) is output from the encoder 3. Replace with the parity code word P '00aAbBcCdD'. As a result, two consecutive idle sets have a parity set of 40 bits / −K28.5 /
P '/-K28.5 / P /. Figure 10
Idle set replaced by (a) and (b) and replacer 4
The format of the data string output by is shown.

On the other hand, the data receiving device 1 in the receiver 330
3ab and 13b are different in the functions of the data receiving device 13, the detector 6 and the parity counter 8 described with reference to FIG. The detector 6 detects a 40-bit parity set from the data string, and the second codeword P'of the parity set is a predetermined "00a'A'b"B'c'C'd'D. ″ Is notified to the parity counter 8, and the fourth parity code word P (10 bits) is guided to the decoder 7.

The parity counter 8 suspends counting when a parity set detection notification is sent from the detector 6, and when the second codeword P ′ of the second parity set has a predetermined value, the count value up to immediately before the parity set. Is output to the comparator 9, and the counter is initialized to perform cumulative counting again from the codeword immediately after the parity set.

According to the embodiment using FIG. 9, in the data transmitters 11a and 11b, a predetermined identification code word P'which is different from the parity code word can be included in the parity set to be replaced and inserted, and Data receiving device 13
In a and 13b, the parity comparison is performed on the parity set including the specific identification code word P ', so that it is possible to identify and detect the occurrence of a data transmission error in each of a plurality of sections. For example, the data transmitters 11a and 11b are respectively assigned identification code words '0001010101' (+ D23.2) '0001010110' (+ D23.6) and the data receivers 13ab and 13b compare the parity sets including the respective identification code words. Should be done.

As a result, the data receiving device 13b can detect the presence or absence of an error in the optical fiber transmission line 5b, and the data receiving device 13ab can detect the presence or absence of an error in both the optical fiber transmission lines 5a and 5b. It can be identified by the receiver 330. Note that FIG. 9 shows an example in which a plurality of data receiving devices 13ab and 13b are provided in the receiver 330, but since the functions of the detector 6, the decoder 7, the parity counter 8, and the comparator 9 are the same, One data receiving device 13
Of course, it may be configured to identify and execute two processes according to the second codeword P ′ of the second parity set.

[0076]

As is apparent from the above description, according to the data transmission error monitoring system and method of the present invention, an nB / mB encoded continuous data sequence is divided into n'data series. Since each parity is counted, the idle set is replaced, and the count value information is transferred, even if consecutive bit errors occur, a parity mismatch occurs in each data sequence, so consecutive bit errors occur. Can be reliably detected. Further, since the repeater counts the parity and replaces and inserts it, it can be specified that the bit error occurrence section is downstream of the repeater.

[Brief description of drawings]

FIG. 1 is a block diagram of a data transmission error monitoring system according to a first embodiment of the present invention.

FIG. 2 is a format diagram showing an input data string and a transmission data string of the monitoring system of FIG.

FIG. 3 is an explanatory diagram of a code conversion rule used in the monitoring system of FIG.

4 is a circuit configuration example of an encoder, a detector, and a decoder used in the monitoring system of FIG.

5 is a block diagram of a relay transmission system using the monitoring system of FIG.

FIG. 6 is a block diagram showing a partially modified example of the monitoring system of FIG.

FIG. 7 is an explanatory diagram of another code conversion rule used in the monitoring system of FIG.

8 is a circuit configuration example of an encoder, a detector, and a decoder according to the code conversion rule of FIG.

FIG. 9 is a block diagram of a data transmission error monitoring system according to a second embodiment of the present invention.

10 is a format diagram showing an input data string and a transmission data string of the monitoring system of FIG.

FIG. 11 is a block diagram of a conventional data transmission error monitoring system.

FIG. 12 is a format diagram showing a data string serialized by 8B / 10B code conversion.

FIG. 13 is an explanatory diagram of an 8B / 10B code conversion rule.

FIG. 14 is a format diagram showing a transmission data string of the monitoring system of FIG.

[Explanation of symbols]

1 ... Input terminal, 2 ... Parity counter, 3 ... Encoder, 4
... Substitution device, 5 ... Optical transmission line, 6, 6 '... Detector, 7 ... Decoder, 8 ... Parity counter, 9 ... Comparator, 10 ... Output terminal, 11, 11a, 11b ... Data transmission device, 13, 1
3ab, 13b ... Data receiving device, 110 ... Transmitter, 2
20 ... Repeater, 230 ... Receiver, 111 ... Encoder, 12
1 ... Reproducer, 129, 139 ... Decoder.

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-2-98244 (JP, A) JP-A-58-34656 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H03M 13/00-13/53 H04L 1/00

Claims (17)

(57) [Claims] 1. A data packet is converted every n bits (n is a natural number) into an m-bit code word by a predetermined nB / mB code conversion rule (m is a natural number of n <m), and the m-bit code is converted. One or a plurality of i-bit (i is a multiple of m) idle sets consisting of one or a plurality of m-bit codewords are inserted between the data packets converted into words, and these are set as a continuous data string from the data transmission device. A data transmission error monitoring system for serially transmitting to a receiving device, wherein the data transmitting device includes n ′ continuous data strings (n ′ is a factor of i <1 <
A parity counter that divides into n '<m) data sequences and counts 1-bit parity for each data sequence, and n'number (n' bits) output from the parity counter.
An encoder that converts the parity of the above into an m-bit parity code word according to a predetermined n′B / mB code conversion rule, and detects one or a plurality of idle sets from the continuous data string, and detects this from the encoder. A detector that replaces and outputs a j-bit (j is a multiple of i) parity set including a parity code word to be output, and the data receiving device detects the parity set from a reception continuous data string. A decoder for reconstructing the n'parities from the parity set, dividing the received continuous data sequence into the n'data sequences and excluding the parity set for each data sequence to determine the parity. A data transmission error monitoring system comprising: a parity counter for counting; and a comparator for comparing the output of the decoder with the output of the parity counter. 2. The data transmission error monitoring system according to claim 1, wherein the data transmission device replaces the parity counter with the continuous data string,
There is a parity counter that divides into n'-number (n 'is a factor of i, m≤n') data series and counts 1-bit parity for each data series, and replaces the encoder with the parity counter. The n '(n'-bit) parity output from the counter is divided into a plurality of count values so as to be smaller than m bits, and these count values are m-bit parity according to a predetermined code conversion rule corresponding to the number of bits. A data transmission error monitoring system having an encoder for converting each to a code word. 3. The data transmission error monitoring system according to claim 1, wherein in the n ′ data series, a bit interval on the serial continuous data string of data belonging to at least one series has a plurality of cyclic values. The data transmission error monitoring system is characterized in that the parity is selected before and after the idle set is repeated so that the parity remains unchanged. 4. The data transmission error monitoring system according to claim 3, wherein the n ′ data series are selected such that a minimum bit interval of data belonging to the same series on the serial continuous data string is maximized. Data transmission error monitoring system characterized by: 5. The data transmission error monitoring system according to claim 4, wherein (n, m, i, n ′) = (8,10,20,5) and the minimum bit interval is 4. A data transmission error monitoring system characterized in that 6. The data transmission error monitoring system according to claim 1, wherein the parity code word always matches any one of the code words generated by the nB / mB code conversion rule. Transmission error monitoring system. 7. The data transmission error monitoring system according to claim 1, wherein the parity set includes an identification codeword in addition to the parity codeword. 8. A data packet is converted into an m-bit code word for every n bits (n is a natural number) by a predetermined nB / mB code conversion rule (m is a natural number of n <m), and the m-bit code is converted. One or a plurality of i-bit (i is a multiple of m) idle sets consisting of one or a plurality of m-bit codewords are inserted between the data packets converted into words, and these are set as a continuous data string from the data transmission device. In a data transmission device used in a data transmission error monitoring system for serial transmission to a reception device, the continuous data string is n'pieces (n 'is a factor of i 1 <
A parity counter that divides into n '<m) data sequences and counts 1-bit parity for each data sequence, and n'number (n' bits) output from the parity counter.
An encoder that converts the parity of the above into an m-bit parity code word according to a predetermined n′B / mB code conversion rule, and detects one or a plurality of idle sets from the continuous data string, and detects this from the encoder. A data transmitter, comprising: a replacer that replaces and outputs a j-bit (j is a multiple of i) parity set including a parity code word to be output. 9. The data transmitting apparatus according to claim 8, wherein the continuous data string is replaced with the parity counter,
There is a parity counter that divides into n'-number (n 'is a factor of i, m≤n') data series and counts 1-bit parity for each data series, and replaces the encoder with the parity counter. The n '(n'-bit) parity output from the counter is divided into a plurality of count values so as to be smaller than m bits, and these count values are m-bit parity according to a predetermined code conversion rule corresponding to the number of bits. A data transmission device comprising an encoder for converting each into a code word. 10. The data transmission device according to claim 8, wherein in the n ′ data series, a bit interval of data belonging to at least one series on the serial continuous data string repeats a plurality of values periodically. , A data transmission device characterized in that the parity is selected before and after the idle set. 11. The data transmitting apparatus according to claim 10, wherein the n ′ data series are selected such that a minimum bit interval of data belonging to the same series on the serial continuous data string is maximized. And a data transmission device. 12. The data transmitting apparatus according to claim 11, wherein (n, m, i, n ′) = (8,10,20,5) and the minimum bit interval is 4. Characteristic data transmission device. 13. The data transmission device according to claim 8, wherein the parity code word always matches any one of the code words generated by the nB / mB code conversion rule. . 14. The data transmitting apparatus according to claim 8, wherein the parity set includes an identification codeword in addition to the parity codeword. 15. A data packet is converted into an m-bit code word for every n bits (n is a natural number) by a predetermined nB / mB code conversion rule (m is a natural number of n <m), and the m-bit code is converted. One or a plurality of i-bit (i is a multiple of m) idle sets consisting of one or a plurality of m-bit codewords are inserted between the data packets converted into words, and these are set as a continuous data string from the data transmission device. In a data receiving device used in a data transmission error monitoring system for serially transmitting to a receiving device, a detector for detecting a predetermined parity set inserted in place of the idle set from a reception continuous data string, and n from the parity set. A decoder for recovering the 'parities, and dividing the received continuous data sequence into n'series of data sequences, and for each data sequence A parity counter for counting the parity exclude Tisetto, data receiving apparatus, characterized in that it comprises a comparator for comparing the parity counter output and the decoder output. 16. A data packet is converted into an m-bit code word for every n bits (n is a natural number) by a predetermined nB / mB code conversion rule (m is a natural number of n <m) and the m-bit code is converted. One or more i-bit (i is a multiple of m) idle set consisting of one or more m-bit code words are inserted between the data packets converted into words, and these are set as a continuous data string from the transmission side to the reception side. A data transmission error monitoring method of serially transmitting data to the transmission side, wherein the transmission side includes n ′ lines (n ′ is a factor of i, 1 <
n '<m) data sequences are divided and 1-bit parity is counted for each data sequence, and the counted n'(n'-bit) parities are converted according to a predetermined n'B / mB code conversion rule. It is converted into an m-bit parity code word, one or a plurality of consecutive idle sets is detected from the continuous data string, and this is included in the parity code word containing the parity code word.
The data is replaced with a parity set of bits (j is a multiple of i) and transmitted, and the receiving side detects the parity set from the reception continuous data string, restores the n ′ number of parity from the parity set, Data that divides a data string into the n ′ data series, excludes the parity set for each data series, counts the parity, and compares the restored parity with the counted parity. Transmission error monitoring method. 17. The data transmission error monitoring method according to claim 16, wherein at the transmitting side, there are n ′ continuous data strings (where n ′ is a factor of i and m ≦).
It is divided into n ') data sequences and 1-bit parity is counted for each data sequence, and the counted n'(n'-bit) parities are divided into multiple count values so as to be smaller than m bits. Then, the data transmission error monitoring method is characterized in that each of these count values is converted into an m-bit parity code word by a predetermined code conversion rule corresponding to the number of bits.