JPS6367942A - Transmission system for computer network - Google Patents

Abstract

PURPOSE:To detect the abnormality of a receiving port within a short period in an accumulation type transmission system by forming a counter indicating the number of receiving times of a message by respective receiving ports to the receiving ports, comparing the values of the counter obtained at different points of time and deciding the receiving ports having the same value as abnormal ports as the result of comparison. CONSTITUTION:In the accumulation type transmission system for optionally coupling plural node devices 1 provided with plural transmitting/receiving ports by the ports to constitute a computer network, transmitting the same message simultaneously from an optional node device to plural node device through previously fixed one or more transmitting ports 14a-14d and repeatedly transmitting the message from the node devices receiving the message through the receiving ports 11a-11d, a receiving counter 18 is connected correspondingly to receiving buffers 12a-12d, messages received from respective receiving buffers are stored in corresponding storage areas every extract of the message, after the passage of prescribed time, the recorded value is compared with the current counter value, and a port corresponding to the contents of the counter having the same value is decided as an abnormal port. The decision is executed prior to retransmission.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a computer network transmission system, and in particular, to a computer network transmission system that can quickly detect network failures, maintain data transmission without interruption even in the event of a failure, and This invention relates to a computer network transmission system that allows node devices to be configured economically.

[Conventional technology]

Conventionally, in storage-type networks such as packet switching networks, the method of suppressing network traffic has been 7
Then, the computers connected to each other through the network insert the payment confirmation and information into the - section of the message sent by the connected computers.
We have adopted a method of not sending the sequel 1-1 until that confirmation is received. However, one calculator for toy number y has the same l'
When transmitting an IV report, so-called broadcast communication, multiple pieces of confirmation information are required, and the procedure for inserting the confirmation information is complicated. Select the only route according to certain rules, and search for an alternative route in the event of a failure;
・11 methods are adopted. In this method, it takes time to reconfigure the two routes, and it is also difficult to send the same message to all or some of the R1 computers at the same time.

Therefore, in order to solve this problem, for example,
0-280913 was proposed. In this method, each node not only relays the received message as is, but also confirms the transmission by transmitting it in the direction in which the message arrived. If delivery is not confirmed, subsequent transmission and relaying of messages from the same source is suspended, thereby effectively confirming the delivery of broadcast communications and suppressing message traffic. In that case, specifically, a message holding number counter is provided for each transmission source, the counter value is referred to immediately before transmission, the message is sent, and the message is put on hold by returning it to the queue.

On the other hand5. The 30th open collector circuit using the trino and logic elements built into the node device is
As shown in the figure, by inputting the same signal as the input data 3.1 of the tri-stay 1-wheel Jl Soryo 31 as the output control signal, an open collector@path is constructed. In the circuit of Figure 30, when the output of the oven collector is disabled, the output rise time is reduced by the pull-up resistor RL specific to the oven collector and the load capacitance fA.
It changes based on a time constant determined by 11 and is larger than that of a normal logic element. That is, in a logic circuit in which the input signal is also used as output disable (i) and operates like a normal open collector, the rise time of the output when the output is disabled is as follows.

It wasn't taken into consideration.

[Problems that the invention attempts to solve]

In the method proposed above, when detecting a network failure, after retransmitting a certain number of times, a port to which delivery has not been confirmed is determined to be abnormal. However, no consideration was given to the fact that holding a message for retransmission has a large effect on the message 1-raw function. Furthermore, in the above method, the message to be held is
Although the message is circulated within the node device mixed with the messages sent 4rJ, no consideration has been given to the processing efficiency in this case. In addition, conventional connection between node devices is half-duplex (bus method) or full-duplex communication, except for loops, but half-duplex communication requires overhead due to contention adjustment time and the complexity of the logic. In the case of full duplex, two transmission lines are required, and two ports are required to connect one node device.

Further, regarding the oven collector circuit built into the node device, the conventional technology does not give consideration to shortening the rise time of the oven collector output. In other words, in a circuit that operates at high speed by wire-ORing multiple outputs, it is impossible to operate at high speed unless the value of the pull-up resistor RL is made as small as possible. However, there is a problem that power consumption increases. Furthermore, in a logic element with a small current capacity, the pull-up resistor R cannot be made small, so there is also the problem that the rise time of the output increases.

The purpose of the present invention is to improve these problems and (a) shorten the message transmission delay time of the network;
(b) eliminates the complexity of the confirmation information insertion procedure in network message transmission; (c) reduces the time required to reconfigure network paths; and (d) is highly tolerant to failures.

The object of the present invention is to provide a computer network transmission system that (e) enables repeat communication and (f) allows economical configuration of node devices.

[Means for solving problems]

In order to achieve the above object, the computer network transmission system according to the present invention arbitrarily connects a plurality of node devices equipped with a plurality of transmission/reception ports using the 48 transmission/reception ports,
construct a computer network 5, simultaneously transmit the same message from any node device to multiple node devices via predetermined LO or more sending ports, and receive the message via the receiving port 2. did. In an accumulation type transmission method in which a node device repeatedly transmits the message and transmits the message, a counter is provided for each receiving port to indicate the number of messages received by the receiving port, and the value of the counter is calculated at different times. The feature is that the values are compared and the three receiving ports with the same result are determined to be abnormal. In addition, a counter is provided for each receiving port to indicate the number of messages received by the receiving port, and a non-message that is the same as a message transmitted or relayed from any node device is sent to the port to which the node device sent the message. Another feature is that the sending and relaying of non-messages that are different from the above-mentioned messages that are subsequently sent from the same node device that sent the above-mentioned non-sages are suspended until the above-mentioned non-sages are received and IL is reached.

(Function) In the first embodiment of the present invention, since the reception counter operates as a monitor of the transmission port, the counter value is not updated unless non-sage reception is performed. It can also be determined whether the node device is not connected or connected, the ILC is connected, or the node device is stopped.

Since receiving a new message can be regarded as a return to normality, it is possible to detect a port abnormality or return to normality in a short time and with simple operations. That is, in the first embodiment.

A reception counter is provided corresponding to each reception buffer, and each time a message is retrieved from each reception buffer.

Store in the corresponding storage area 1. If this recorded value is compared with the current counter value at this time after a predetermined period of time has elapsed, the values are equal, and reception is performed at the port corresponding to the negative counter.
It can be determined that this is an abnormal port that does not have an error of 11. This determination may be made before retransmission.

Next, in the second embodiment, the transmission 1B control module i
Since only meso messages that need to be transmitted to -u- are transferred, the throughput of relay messages is improved as shown in FIG. That is, in the second embodiment, received messages are grouped by their source and 7. A tag (hereinafter referred to as 1-1) is added to the message to be sent. An upper limit on the number of tokens to be allocated to each group is set, non-messages within the node device are exchanged using tokens, and messages that do not have a token attached to them are removed from the transmission process, thereby suspending transmission.

In the third embodiment, the computer network is a combination of loops 71. By placing the node device at the intersection of a plurality of loops, each node device is coupled to the same number of node devices as the number of ports it has.

The data broadcast by each node device goes around a loop that includes that node device, and is then re-received by all receiving ports provided by that node device. That is, since this -reverse time is feedback information of traffic around the node device, traffic can be effectively suppressed by using this information at the time of next data transmission.

In the fourth embodiment, 1-Raischi 1-For the rising or falling edge of the input signal of the logic element.

Providing means to sufficiently delay the output control signal 8
That is, delaying the output control signal delays the timing at which the tristate logic element is disabled and enters a high impedance state. Therefore, until the tri-state logic element becomes high impedance, it operates at high speed as a normal binary logic element, so the rise of the output is caused by the pull-up resistor RL.
The speed can be increased almost regardless of the size of the moreover.

Since the value of the pull-up resistor RL can be increased, the power consumption of the transistor 11 during switching operation can be significantly reduced. Further, as a characteristic of the tri-state logic element, the output current when disabled is much smaller than that of a normal binary logic element, so that an open collector circuit with low consumption can be realized.

〔Example〕

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

FIG. 5 is a configuration diagram of a computer network system to which the first embodiment of the present invention is applied.
is a node device, and 2 is a transmission line. In this case, the connection of the node devices l is shown as being connected in a lattice pattern, but 5 other connections (for example.

Of course, loop coupling) is also acceptable.

This computer network system has a configuration in which each node device 1 is connected in a lattice shape by a transmission line 2, and each node device 1 transmits messages to C, other node devices, etc. through the transmission line 2. 1: Nodoki, node equipment [Messages from 1 are sent by broadcast method,
The message is transmitted to the first adjacent node device l connected through the transmission line 2.

FIG. 6 is a detailed configuration diagram of the node device 1 shown in FIG. In FIG. 6, 3 is a processing device module that performs control functions of the node device I, 4 is a transmission control module that controls message transmission, 21 is a receiving (H line, 22 is a sending 471 line, and 471 is a sending line). Line 22 is.

It becomes the receiving fR line 21 in the connected node device 1.

Hereinafter, transmission (reception) between the transmission control module 4 and the processing device module 3 will be referred to as local transmission (reception).
The sending port (receiving) between node devices l will be referred to as global sending (receiving).

7 is a detailed configuration diagram of the transmission control module 4 in FIG. 6. In FIG. 7, 5 is a receiver (1 control module,
6 is a transmission control module, 7 is a message management table, and 8 is a port management table.

9 is a relay queue to which the message to be sent is connected to the node device 11, 10 is a retransmission queue 511 is a receiving (n port (subscripts a-d are port IDs for different models), 12
is the receive buffer (the subscript is the port I as in 11)
D and below are the same), 13 is a sending port (n buffer 7.14 is a sending port, 15 is a local sending (n queue) that combines messages to be sent to the processing device module 3, 1G is a dividing buffer for storing messages) ]t memory space (hereinafter referred to as baren E), manages 5 empty palettes o(
Pallet management module 17 is the one that is not in use.
18 is a reception counter that indicates the number of messages received at each port. FIG. 9 is a diagram showing the configuration of the reception fa counter 18 in FIG. 7. The receiving (a counter 18 is composed of five counters a to e corresponding to the receiving 4Q buffer 12,
I have X. Each counter is an unsigned integer counter, and each time a message is retrieved from the receive buffer 12, the value of the corresponding counter is incremented by one. In order to store messages in the balen 1-, the balen 1-management module 1G combines a pallet with each receiving buffer 12 upon start-up of the node device 1-. Customer receiver (i-port 11 is connected to the 5-corresponding receiver 473 buffer 12 and the like,
The reception control module 5 that stores messages arriving from the reception line A 21 takes out the pallet in which the message is stored from the reception buffer 12, and determines whether the message requires relaying or not. If the token number is not registered in the message management table 7, obtain the [--kun from the token management module 17 and register that number in the message management table 7, and combine the [--kun and palette]. and connects 1-1 to relay queue 9. However, it requires relay,
The number 1-kun is already in the message management table 7.
-8sa1+.

If so, the address of the palette is only registered in the message management table 7.

Here, the structure of the token and message management table 7 will be explained. FIG. 8 is a diagram showing an example of the configuration of 1-1, and FIG. 10 is a diagram showing an example of the configuration of the message management table 7.

In FIG. 8, 1011-1. The next token number (hereinafter referred to as NP), 102 is the previous token number (hereinafter referred to as BP).
103 is an address voice. The address pointer 103 contains the address value of the connected baren]-. If 0′C-aJl, BP102 such as NPIOl, which indicates that valene 1- is not bonded, is
Since it refers to the previous and next tokens in the e-queue that is transmitted in the relay queue 9 or retransmission queue 10,
By copying the values of its own NPIOI and BP103 to the previous and subsequent 1-1 keys, it is possible to take out a token from the middle of the queue. A plurality of such tokens are provided in the node device 1 of l.D.2 In FIG.
Field 72 is a relay pending message registration field in which the address of baren 1 to relay pending message is entered. 73
is a message serial number field (assuming that a message has information SA indicating its source and message passing information added for each source), and 74 indicates that the message has been received. 75 is a sent flag indicating that the message has been sent, 7
6 is a retransmission counter. Here, the received flags 74 are 74 a , 74 h , 74 c , 74
It is divided into 5 and 74 e, and it is J
Reception buffer 12a, respectively.

12b, 12c, 12cl, and 120.

Similarly, the sent flag 74 is 75a, 75b.

75 c + 75 d + 75 e, 1t and J3.

Transmission buffers I3a and l3b respectively
, l 3 , : .

It corresponds to 13d and 13e. The retransmission counter 76
In this case, the message is taken out from the retransmission queue lO, connected to the relay queue 9, and added 1 to I, but when the counter value reaches a predetermined value, the unidentified port or the IL This indicates that there is an error in the adjacent node connected to the node, and the message is not returned.

FIG. 11 is a diagram showing an example of the configuration of the port management table 8 in FIG. 7. In FIG. 11, 81 is a receiver ('B port error flag) indicating an error in the receiving port, and 82 is a transmitting port error flag indicating an error in the transmitting port. When the receiving port abnormality flag 81 is compared with the receiving counter record field 77 and the receiving 4B counter 18, a value corresponding to the counter with the same value is set. When the value of the counter 76 reaches a predetermined value, the counter corresponding to the unconfirmed port is sent, and when a message is received, it is reset, and the sending port abnormality flag 82 indicates that a message is sent from the corresponding port. It is set when sending (a) is not possible. While receiving (n port abnormality flag 81 is set), delivery confirmation is not performed at the corresponding port.

FIG. 12 is a detailed diagram of the receiving 13 side of the transmission control module 4 in FIG. 7, particularly the internal configuration of the reception control module 5. The reception control module 5 includes four global reception 48 control modules 50 corresponding to the transmission ports.
a = d, a local reception control module 502 corresponding to the processing device module 3, and a retransmission control module 5
03, and a timing control module 504 that takes timing of the above modules. In Japan, global reception (i control module 501 and local reception control module 502 have the same configuration).

Only 501a is shown.1. Other descriptions are omitted.

The global reception control module 501 or the local reception control module 502 controls the corresponding reception (n buffer).
When a message is retrieved from 2, a predetermined reception counter value is incremented by 1, and the reception port abnormality flag is reset to 1. When relaying or holding the retrieved message, all five reception counter values are set in the reception counter record field 77 of the message management table 17.

When retransmission control module 503 retrieves a message from retransmission queue IO, retransmission control module 503 records reception counter record field 77.
and the value of the reception counter 18, and if they are equal, the reception port abnormality flag corresponding to the corresponding port is set to 1-. Furthermore, if the value of the retransmission counter 7G exceeds a predetermined value, the reception (i-port abnormality flag) corresponding to the unconfirmed delivery is set.The reception port abnormality flag is reset.
If there is unconfirmed delivery on the IL port, the message is connected to the relay queue 9.

FIG. 13 is a diagram showing details of the internal configuration of the transmission side of the transmission control module 4 in FIG. 7, particularly the transmission control module 6. The transmission 13 control module 6 includes a global transmission module 601 and a local transmission module 602.
It consists of

The global sending module 601 takes out the token from the relay queue 9 and sends it to the sending buffer -/13a-d.
A baren that stores messages in everything! Set the address of ~. After the transmission is completed, a predetermined transmitted flag is set and 1.-kun is connected to the retransmission queue 10. Delivery 4B
The transmission port abnormality flag in the port management table 8 corresponding to the transmission port a buffer 13 in which the transmission is not performed is set.

[1-The local transmission control module 602 controls the local transmission (
The pallet address is obtained from the a-queue 15 and set in the transmission buffer γ13e. After the transmission is completed, a predetermined sent flag in the message management table 7 is set, and the pallet is returned to the pallet management module 16. In case the transmission does not take place.

A predetermined abnormality flag in the port management table 8 is set.

A case in which message transmission is performed between node devices L and the like in No. 1 (device 1) as described above will be described in detail.

First, the operation of the reception control module 1 will be described. FIGS. 1A and 1B and FIG. 2 are flowcharts of the operation of the reception control module 4. 1A and 1B show the operation flow of the global reception control module 501 or the local reception control module 502, and FIG. 2 shows the operation flow of the retransmission control module 503.

Since messages are stored in pallets, the address of the pallet is set in the receive buffer 12 before reception.
The receiver 1a port 11 stores the received message according to the pallet address of the corresponding receiver (,7 buffer 12).

First, suppose that pallets storing messages are combined in the reception buffer 12a (step 1001).
The global receiver 43 control module 501a takes out the pallet address from the reception buffer γ12a (step 1002), clears the receiver (n port abnormality flag 81a) in the port management table 8, and sets the corresponding counter in the receiver H counter 18. Add 1 to 1 shift (step 1003).Next, add 1 to 1 shift (step 1003).
Read out the serial number area 73 of 7 and compare this with the message serial number (step 1004'), equal, kunake]1
For example, clear the received (a completed flag 7.1 and the sent (n completed flag 75), set the received flag 74a corresponding to the reception buffer 12a, and record the value of the reception counter 18 in the reception counter record in the message management table 7. It is recorded in field 77 (step 1005).On the other hand, when they are equal,
Set the received 4n completion flag 74a (step 1006
), execute step 1005, and then check the token number registration field 71 of the message management table 8. Step 1008), if it is 0, receive the token from the management module 17 and set its address pointer 103, inputs the address of baren 1- obtained from the reception buffer 12a, sets the number of 1-1-kun in the 1-1-kun number field 71, clears the retransmission counter 76 to 0, and then (Step 1o
o9), connect the t-kun to the relay fragrance 9; step 1010), call the pallet management module 16;
Binding the current to the receive buffer 12 (step t
oll). As a result of step 1008, if the contents of the token number registration field 71 are other than 0, the address number of the pallet taken out from the reception (a buffer 12a) is set in the intermediate JtHR message registration field 72 (step lo 12). Then, move to step toll.

After executing step 1006, the pallet is returned to the pallet management module 16 (step 1013').
. Next, all the received flags 74 and sent flags 75 are set (excluding those corresponding to the n port error flag 81 and the transmit port error flag 82). If it is confirmed that the token is in the queue (step 1014), remove the token from the queue it is currently in (step 1015), and set the at1 pointer 1.
The pallet pointed to by 03 is returned to the pallet management module 16 (step 101G), and the contents of the message registration field 72 are transferred to the address pointer 10.
3, relay pending message registration field 7-2
=+ − 2 is cleared to O (step 1017). Step 1
Result 5 of executing 017: If the content of the address pointer 103 of the token is not 0, execute step 101O, and if it is 0, return the token to the token management module 17 and fill in the 1.1-kun registration field. 71 is cleared to 0 (steps 1017 and 1018). Step 1
After executing step 018, step 1011 is executed.

The other global reception control modules 501b-d and local reception control module 502 are also operated in the same manner.

Next, the operation of the retransmission control module 503 will be explained in detail with reference to FIG. 2. First, the retransmission control module 503 checks whether there is a 1. , and the value of the receiver 4B counter 18 at this time are compared, and if they match.

The reception port abnormality flag 81 is set to Sera I-L ((full).

Reception counter record field 77a and reception counter record field 77a
If the values of 8a are equal, receive port/abnormal flag 81a
) (step 1102), retransmission counter 7
Check whether the value of 6 is larger than a predetermined value (step 1
103). When the value is greater than a predetermined value, the receiving port abnormality flag 81 corresponding to the port whose delivery has not been confirmed (received flag 74 is not set) is set (step 110.1), and when the value is less than the predetermined value, does not execute processing step 1104, but refers to the received flag 74 of the message management table 7 to check whether the message has been received on all ports (step 1105).

If there is a port that has not received the data, 1.-kun is connected to the relay queue 9 (step 110G), and the value of the retransmission counter 76 is incremented by 1 (step 1107).

If all ports have received it, the pallet pointed to by the address pointer 103 of the token is immediately transferred to module 1.
Returning to Step 1108), the address pointer 103 is newly filled with the contents of the relay message registration field (), tab 1109'). Result of executing step 11o9.

If the content of the address pointer 103 is not 0, connect the h-kun to the relay queue 9 and clear the relay pending message registration field to 0 (steps 1110 and 1111).
), if the content of the address pointer 103 becomes 0, the token is returned to the token management module 17 and the token number registration field is cleared to 0 (
Step 1112).

Next, a case in which a message is transmitted by the transmission control module 6 in FIG. 7 will be explained. 3 and 4 are operation flowcharts of the global transmission (a control module 6).
FIG. 4 is an operation flowchart of the a control module 601, and FIG. 4 is an operation flowchart of the local transmission control module 602.

First, in FIG. 3, when there is a token in the relay queue 9 (step L201'), 1--kun is taken out and the contents of its address pointer 103, i.e., address 1 of the pallet that stores the message (sends the response). Buffer 13
a" d (step 1202), causes the sending ports 14a to 14d to send a message, and waits for the sending to finish (step 1203). When the sending is finished, the sent flag corresponding to the port that has completed normally is set. 75, and also sets the transmission port abnormality flag 82 corresponding to the port for which abnormality has been terminated (step 1).
204). Then, 1--kun is connected to the retransmission queue 10 (step 1205), and the relay queue 9 is checked again. If there is no token in the relay queue 9, C waits.

Next, in FIG. 4, when there is a message in the 0-Cal sending queue 15 (step 1301), the local sending control module 602 stores the message from the local sending queue I5 and takes out the address of the pallet. (Step 1302).

Next, set this address in the transmission buffer 13e,
Make the processing device module perform reception (a), and wait until the completion of step 1 (step 13θ4). 2. If it ends normally, set the transmission port 1 to abnormality flag 82e to zero (step 1305). After that, the pallet is returned to the pallet management module 16 (step 1306), and the local transmission queue 15 is checked again, and if there is nothing, it waits.

In this way, in the first embodiment, by providing a reception counter and comparing the counter values at different times, it is possible to detect an abnormality in the reception (i-port) in a short time. .Detection of ports to which no node device is connected or ports that are connected but the node device is stopped is shortened.

Next, FIGS. 15 and 16 are configuration diagrams of a message management table 7 and a reception control module 6 showing a second embodiment of the present invention. That is, in the second embodiment, among the configurations of the node device shown in FIG. 7, the reception control module 6 is a feature.

In FIG. 15, 71 is a token number registration field, and 72 is a relay station message registration field, into which the pallet address of the relay pending message is entered. 73 is a message sequence number field (a message is sent if there is information SΔ indicating the source (i) and message poisoning information added for each source), and 74 is a message serial number field (if the message has been received (n)). Sent flag indicating that there is, 7
6 is a retransmission counter. Here, the received flag 74 is divided into five parts, 574d to 74e.
] corresponds to the L reception buffers 12a to 12e. Sending (n
Similarly, the completed flag 74 is also divided into five parts 75a to 75e, which correspond to transmission buffers 13a to 1.
Compatible with 3e. The retransmission counter 76 indicates that the message is taken out from the retransmission queue 10 and connected to the relay queue 9.
When the counter value reaches a predetermined value, there is an abnormality in the unconfirmed port 1 or the adjacent node connected to it, and the message is returned, indicating that there is no message. ing.

In FIG. 16, the reception control module 5.

Global reception control module 501 corresponding to ports
a to d, a local receiving 13 control module 502, a retransmission control module 503, and a timing control module 504 for timing the above modules.

FIG. 14 shows the operation flow of the retransmission control module 503 of the operation flow [1-] of the reception control module 4. Note that the operations of the global reception control module and the local reception control module are the same as those shown in FIGS. 1(a) and 1(b).

Since a message is stored in a pallet, the address of the pallet is stored in the receive buffer 12 before reception.
The received message is stored according to the palette address.

First, the retransmission control module 503 checks whether there is one hunk in the retransmission queue 10 (step 1101a).
, A] l[... 1102a), it is checked whether the value of the retransmission counter 76 is larger than a predetermined value.

If it is larger than the predetermined value, the reception port abnormality flag 81 corresponding to the port whose delivery is not confirmed (received flag 74 is not set to cell) is set to cell 1 (step 1104).
), if the value is less than the predetermined value, the fifth processing step 1104 is not executed.

Refer to the received flag 74 of the message management table 7 and check whether the message has been received on all ports (step 1105a) If there is a port that has not received the message, connect the token to the relay queue 9 (step 1106), and the value of retransmission value 76 is incremented by 1 (step 1107). If it has been received on all ports,
The pallet pointed to by the token address pointer 103 is returned to the pallet/management module 16, and (step 11
08a), new.

The contents of the relay pending message registration field are set in the address pointer 103 (step 1109a).
), if the content of the address pointer 103 is not 0 as a result of executing step 1109a, connect the token to the relay queue 9 and clear the relay pending message registration field to O (step 1lla), set L, address pointer 103 If the content of becomes O, 1--
The token is returned to the existing token management module 17 and the token number R8 field is cleared to O (step 1112).
a).

As described above, in the second embodiment, the complexity of the confirmation information insertion procedure-33-nr in network message transmission is eliminated, the time required to reconfigure the network route is shortened, and the tolerance against failures is increased. It is possible to construct a network that has a high value of 2 and can also perform broadcast communication.

Next, FIG. 17 is a basic configuration diagram of a computer network system showing a third embodiment of the present invention. In FIG. 17, 1 is a node device, 2 is a transmission line, and the direction of the arrow indicates the flow of data. In other words, a unidirectional communication system is used between two adjacent node devices, and the configuration of this embodiment is a combination of a loop formed by three node devices.

FIG. 18 is a configuration diagram of the node device 1 in FIG. 17. In FIG. 18, 41 is a processing unit (hereinafter referred to as
42 is a communication control unit (hereinafter referred to as CU), 43a=c is a reception port, and 44d-f are transmission ports. Also.

121 is a reception control unit (hereinafter referred to as RCU), 1
22 is a transmission control unit 1- (hereinafter referred to as SCU),
123 is a relay queue, 125 is a data management-3=1- table (hereinafter referred to as DMT), and 126 is a port status 5127 is a clock. Data from other node devices is received at the reception port 13 at the RCU l 2
Passed to 1, 11. Data from PU41 is from RCIJ
121 receives. The RCUI 21 transfers the received data to the relay queue 123. The 5CU 122 retrieves data from the relay queue 123, and the transmission port 44 transmits the data to another node device. Also, SC, U
122: 1) sends data to the computer 11;

FIG. 19 is a diagram showing the format of data used in FIG. 17. In Figure 19°, 131
is the data type code (hereinafter referred to as FC) area, 132
is the data source address (hereinafter referred to as SA) area 51
33 is a data serial number (hereinafter referred to as C) area, and 134 is an information area (hereinafter referred to as ■).

FIG. 21 is a configuration diagram of the RCU 121 in FIG. 19, and FIG. 20 is a diagram of the data management table (DMT).
FIG. 22 is a configuration diagram of the port status 126 in FIG. 21.

In FIG. 21, 1211 is a reception buffer (hereinafter referred to as R
1212 is a reception control unit.

1213 is a timing control unit that operates the reception control unit 1212 in a time-division manner. In addition, the subscript g is.

RB for PU41 is shown.

In Fig. 20, 1251 is the SA area 32.1252
is the serial number area, and these are the SA area 325 of the data.
Each corresponds to the serial number area 33. Further, 1253 is a received flag (a=c corresponds to the subscript of the received call h43), 1254 is a sender flag, and 1255 is a received call flag. Also.

In FIG. 22, 1261 is a reception port abnormality flag, and 1262 is a transmission port abnormality flag. a-f
corresponds to the port index.

In FIG. 21, the receiving port 43 receives received data (P
U41 is the generated data), 1 is the corresponding RB1
For example, the reception port 43a is stored in RB12.
11a) Then, the timing control section 1213 is activated. The timing control unit 1213 controls the port status 1
The reception control unit 1212 corresponding to the bit that is cleared (in other words, indicates normal) of the R port abnormality flag 1261 is started.Reception control unit 121
2, if there is no data in the RB 1211 corresponding to itself, calculates the elapsed time since the last reception, and if the value exceeds a predetermined value, sets the reception port abnormality flag 12131. Sets the corresponding bit to indicate an error on the receiving port. If there is data in RB1211, search for 0MT125 using 5A32 and C33 of that data. As a result, if the data is not yet RIII=t', write the value of the serial number field 33 to the serial number field 1252 of 0MT125) and set the bit of the received flag 1253 corresponding to the receiving port. , is connected to the relay queue 123. As a result of the search,
If the data has been registered, the bit of the received flag 1253 corresponding to the port that received the data is set and the received data is deleted. The receiving in control unit 1212 that receives data from the PU 41 sets the seven compensation completed flag 1255 instead of the received flag.

Next, the configuration and operation of the 5CU 122 will be described in detail.

FIG. 23 is a configuration diagram of the 5CU 122. 1221 is a transmission data processing unit, 1222 is a transmission-in buffer (hereinafter referred to as
1223 is an FC registration table (hereinafter referred to as FCT).

The 5CU 12'2 takes out the data from the relay queue 123 and sets it to 5B1222def. Also, if the value of the FC area 31 of that data is registered in the FCT 1223, and the t-credit flag for that data in the MoTl 25 remains clear, SB1
Set it to 222g.

However, with data equal to S A 32 of that data,
Previously sent data is received at 4'B port 43.
If not all have been received, (received flag 12)
53 and the reception port abnormality flag '1261 for each corresponding bit (in order to ignore two signals since reception is not possible at the abnormal port), and check whether the result is all 1 or not. If all 1 is Jl, all ports receive (
73 completed), connect to the relay queue 123 without setting it to 5I31222, set Seno [ to 5B1222de, f
・If it is, set the sent-in completed flag 1254 and send it to SB.
If set to 1222g, the incoming call flag 125
The eSCU 122 that sets 5 sets the data to 5B122.
If set to 2def, the transmission port 44 is activated. The transmission port 44 transfers the 5B1222def data to another node device, but if a failure is detected during the operation of the transmission port 14, the transfer is interrupted and the transmission is attempted again. However, if this number of times exceeds a predetermined value (0 or more), the data is not transferred. At this time, an indication that the port is abnormal is made by setting the bit corresponding to that port in the sending port abnormality flag 1262 (an indication that it is normal is not made by clearing that bit). (This is done by the sending port when the sending is successfully completed).

The basic configuration shown in FIG. 17 has low fault tolerance. Each node device 1 is provided with three ports for transmission and three for reception, but in FIG.・
located at the edge of The node device 1 uses only three of the ports. For example, it was located in the center. Assume that the node device 1 has stopped.

Since there are node devices that cannot transmit or receive data, connect the vacant ports (the number of transmit ports and receive ports of one node device are odd numbers, so
The number of empty transmitting ports and the number of empty receiving ports in the basic configuration are equal), and the configuration is as shown in FIG. This jt is designed so that all node devices form one loop. In this configuration, there are at least two data buses, and (1) the number of times data is relayed is one at most. In this configuration, FIG. 25 is a configuration diagram when the node device located in the center is stopped. Even with this configuration, there are at least two data buses between two node devices. Also,
All node devices (6 devices) constitute one loop.

Regarding processing of empty ports, there are several other methods, for example,
This is a method of creating the configuration shown in FIG.

Even in this configuration, there are at least two data buses. Furthermore, in the configuration shown in FIG. 26, if the node device located in the center is stopped, the configuration shown in FIG. 27 will be obtained. Even in this configuration, there are at least two data buses, but there are (1) paths that require data to be relayed at least twice.

In this way, in the third embodiment, the increase in traffic can be suppressed through simple control, and the same message can be sent to multiple routes at the same time. A network can be realized.

Also, it can be connected to the total number of node devices including receiving ports and sending ports. In other words, it is economical because the number of transmitting (receiving) ports can be less than the number of connectable node devices.

FIG. 28 shows a fourth embodiment of the present invention, and is a block diagram of an open collector circuit built in a node device. Further, FIG. 29 is a timing chart showing the operating principle of FIG. 28.

Rise of input signal of tri-state logic element, -, +1
−
Delaying the output control signal sufficiently with respect to 'lρ or falling will delay the timing at which the tristate logic element is disabled and enters a high impedance state. Therefore, until the tri-state logic element becomes high impedance, it operates as fast as a normal binary logic element, so
The rise of the output can be achieved at high speed, almost unrelated to the size of the pull-up resistor RL. Furthermore, the pull-up resistor R
Since the value of L can be increased, power consumption during 5-switching operation can be significantly reduced. Further, since the tristate logic element has a characteristic that the output current during 5-day fabrication is much smaller than that of a normal binary logic element, a low-consumption open collector circuit can be realized.

When using a tri-state logic element as an open collector circuit, an output control signal input that turns off the output and makes it high impedance is used to turn the output off to high impedance.
Method 5: For example, by providing a delay circuit, the timing at which the output of the tri-state logic element becomes a high in-bee/12- Since the timing of the output signal (g) is later than that of the open collector element, the rise of the output signal is the same as the operation of a normal logic element that is not an open collector, and the rise of the output signal (g) is faster than that of an open collector element.

On the other hand, when the input signal of a tri-state logic element is directly connected to the output control input and used as an open collector circuit, the output becomes high impedance due to the output control input. Since it is faster than variable f, it operates in the same way as a normal open collector element, and the rise of the output is determined by the time constant TL = R, which is determined by the product of pull-up resistor RL and load capacitance cL.
In order to achieve the same rise speed as a non-open collector logic element, the value of RL needs to be sufficiently small. To reduce RL.

A device with a large output current is required, resulting in high power consumption.

In the fourth embodiment of the present invention, when the input changes and the output is set to high level, the output control No. 4a of the tri-state logic element is operated with a delay from the input change, so that the output changes from low level to high level. Since the change to level 5 operates as a normal logic element, the output rise speed is hardly affected by the pull-up resistor RL, so by increasing the value of R1, power consumption can be reduced. Can be done.

Input data is input to the tri-state logic element 51,
When the output control signal 5G is ``H'', the output of the 5-tristate logic element 51 is reflected in the output data 57 and becomes erroneous. The output control signal 56 is based on the input data 54 and the delay data 55 that delays the input data 54 by a predetermined time via the delay circuit 5'2.
It is output by the output control circuit 53. The output of the tri-state logic element 51 is pulled up to the power supply via a pull-up resistor (R) 58, and is connected to IL and a load capacitance cL equivalent to the load to be driven. For simple qt, an inverter is used for the tri-state logic element 51, and a rectangular wave with a pulse width t P w shown in FIG. 29(a) is input as the human data 54. explain.

The input data 54 is delayed by a predetermined time td by the delay port vfs 52, and becomes delayed data 55 as shown in FIG. 29(b). The output control circuit 53 receives input data 54 and delay data 55 and generates an output control signal 56. The output control signal 5G may be generated by various methods, such as by calculating the logical sum of the two-manpower data 54 and the delay data 55.

The signal has a pulse width of (tpw+td) as shown in FIG. 29(c), and when the output control signal is HHH, the output is enabled and the output control signal is set to II.
When LH, the output of the tri-state logic I>'bark 51 is controlled so that the output is disabled.

That is, since the timing at which the input data changes from 'H' to 'L' is delayed by the delay time td, signal No. 46 shown in (c) is used as the output control signal 56, the input data 54 changes from 'H' to 'L'. Even during the delay time td after changing from `` to ``L'', the output control signal remains ``H''.
It is. Therefore, the tri-state logic element 51 is enabled and its output becomes the inverted signal of the input data 54, and after an element-specific propagation delay time, immediately n L 1
Changes from 1 to ``H''.

Next, after the input data 54 changes from ``H'' to ``L'' and time td elapses, the output control signal 56 becomes
As shown in (e), from "H"' to zl L II /
Because of the change in f, the output of tri-state logic element 51 is disabled and placed in a high impedance (z) state.

After this, the input data 54 is (f L 11 to I ) (
11, the tri-state logic element 51 is enabled and the output data becomes L'', which is the inverted signal of the input data.

The waveform of the above output data 57 is shown in (d).

Therefore, the rising edge of the output of tri-state logic element 5I is equal to the rise of I when tri-state logic element 51 is enabled.
It becomes equal to the rise time t T Ll from L''' to #HII.

On the other hand, in the conventional open collector circuit using the tristate logic element 31 shown in FIG.
is connected to the output control input of the tri-state logic device 31, and as soon as the input data changes from ``H'' to ``L'', the output of the tri-state logic device 31 is disabled.

Since the output is in a high impedance state, the output data 37' rises with a time constant TL'' RLCL determined by the product of the pull-up resistor RL and the equivalent load capacitance CI-.

This waveform is shown in FIG. 29(e), and its rise time tT
LH2 is a function of TL, and the output varies from 10% to 90%
The time for the change to tTLH2, that is, the rise time tTLH2, is expressed by the following equation.

t-rLHs+ =TL (12n0.9-1!nO,
l) Length 2.2 TL Therefore 1 For example, low shot key (LS) type T TL (Transister -T ra
integrated circuits (I
C: Ir+tegrated C1rcuijin)
Since the output rise time t TL H of is less than l D n S, in order to operate equivalently to this, the load capacitance CL is on the order of tens of pF, and the value of RL is made small. There must be. However, doing so increases power consumption and may not be able to drive the conventional LS type TTL.

When the output rises, the pull-up load resistance R and load capacitance C
Since the time constant of L is determined by TL = RL CL, 4y.

The output rise time is slower than that of a normal logic element that is not an open collector. If it were to obtain the same rise speed as a normal logic element, the pull-up resistor RL would have to be made smaller, and the output current would increase, resulting in an increase in the power consumption of the output current.

In the fourth embodiment of the present invention, the timing for turning off the roll signal is delayed from the timing at which the level of the input signal 4'B changes. When the signal changes, it operates as a normal logic element that is not an open collector until the output signal changes following the input, and then the output can be set to high impedance by turning off the 5 output control (No. 8 is turned off) Therefore, the time for the output to rise to a high level is the same as that of a normal logic element, and is faster than an open collector element.In addition, the off-state output current of a tri-state logic element when it becomes high impedance is equal to that of a normal logic element. Since it is one fold heavier than an open collector element, it has low power consumption.Furthermore, the value of the pull-up resistor RL has little effect on the output rise time, so a pull-up resistor with higher impedance than an open collector element is used. That is, in the fourth embodiment of the present invention, the rise time corresponds to the logic element used, so an open collector circuit that is faster and consumes less power than the conventional open collector circuit can be realized. .

〔Effect of the invention〕

As explained above, according to the present invention, an abnormality in a receiving port can be detected in a short time by comparing counter values at different times, and also it is possible to detect an abnormality in a receiving port in a short time by comparing counter values at different times. It reduces complexity, reduces the time required to reconfigure network routes, is more tolerant to failures, and allows for round-trip communication.
Furthermore, the increase in traffic can be suppressed through simple control, and the same message can be sent to multiple routes at the same time, thereby improving reliability. Further, the output rise time is shortened, and an open collector circuit with low power consumption can be realized.

[Brief explanation of the drawing]

FIG. 1 is an operation flowchart of a reception control module of a node device in a computer network system showing a first embodiment of the present invention, FIG. 2 is an operation flowchart of a retransmission control module, and FIGS. 3 and 4 are transmission flowcharts. Operation flowchart 1 of the control module, FIG. 5 is a configuration diagram of a computer network system to which the first embodiment of the present invention is applied, FIG. 6 is a detailed configuration diagram of the node device in FIG. 5,
Figure 7 is a detailed configuration diagram of the transmission control module in Figure 5, Figure 8 is a token configuration diagram, Figure 9 is a configuration diagram of the reception counter in Figure 7, and Figure 10 is the message management table in Figure 7. , FIG. 11 is a configuration diagram of the port management table in FIG. 7, FIG. 12 is a detailed configuration diagram of the reception control module in FIG. 7, and FIG. 13 is a detailed configuration diagram of the transmission control module in FIG. 7. , FIG. 14 is an operation flowchart of the reception control module showing the second embodiment of the present invention, FIG. 15 is a configuration diagram of the message management table in FIG. 7, and FIG. 16 is a flowchart of the reception control module in FIG. Configuration diagram, No. 17
Figure 18 is a diagram showing the overall configuration of a network system showing a third embodiment of the present invention; Figure 18 is a diagram showing the configuration of the node device in Figure 17; Figure 19 is a diagram showing the configuration of the data format;
FIG. 20 is a configuration diagram of the data management table in FIG. 18, FIG. 21 is a configuration diagram of the reception control unit in FIG. 18, and FIG.
The figure is a configuration diagram of the port status in Figure 21, Figure 23 is a configuration diagram of the transmission control unit in Figure 21, Figures 2 and 1 are configuration diagrams of the third embodiment, and Figure 25 is reconfiguration in the event of a failure. 26 is a diagram showing an example, and FIG. 26 is a configuration diagram of a modified embodiment of FIG. 24.
The figure shows an example of reconfiguration in the event of a failure with the configuration shown in Fig. 26, Fig. 28 is a block diagram of an open collector circuit showing the fourth embodiment of the present invention, and Fig. 29 shows a timing diagram showing the operating principle of Fig. 28. The chart, FIG. 30, is a configuration diagram of a conventional open collector circuit. l: node device, 2: transmission line, 3: processing device module, 4: transmission control module, 5: reception control module,
6: Transmission control module, 7: Message management table, 8: Port management table. 9: Relay queue, 18: Reception counter, 100; 1・-
41: Processing unit, 42: Communication (S control unit, 13: Reception port, 14: Transmission port 5121: Reception control unit, 122: Transmission control unit, 123:
Relay queue, 125: Data management table, 126: Port status, 31°51: t-Rystate theory T! J, element, 52; delay circuit, 53: output control circuit,
54: Input data, 55: Delay data, 56: Output control signal, 57°37': Output data, 58.3
8 nibble up resistance, 59, 39: load capacity. Genken Patent Attorney Ko Month 1 Katsu Ogoto 10 (e-) Start of Figure 2/-77t'g l! ρl kat-31, 3 is also -1, Fuzuνㅅ)x3+1Ld)nt71ρ2. , yyρ2 Y-ririn) 1t, yll, 3 Kun ψ Sen) ζ tanta 1j ν //7θJ・ノX,

,j;'y~
7 y' 4
ts = θrou fee
“I〃t Seme Tsuki Shijiba All Beds゛-) 9 7/p4 2Y-Kun Sosu Yu Iyu
Tokun θ7】IY /71!77 Notano 7 Rod 3 concave rod 4 figure rod 5a I finished the rod A'z 8rfJ $9TM child 10 concave rod 17 I! It No. 7z Kuchibee 13 T! 1 Flower! 4 figure potato 15 separate Z4 7 (grass 7 Otsu figure firewood! 7 figure ¥ I8 concave falcon lq mouth ¥ zO mouth Ll) C child 21 figure #zz eye spear 24 T! 1 Hayabusa z3Tj: J ¥25 Figure V-260 Rod 27 times' Figure 23

Claims (1)

[Claims] 1. A computer network is constructed by arbitrarily coupling a plurality of node devices each having a plurality of transmission/reception ports by the transmission/reception ports, and one or more predetermined transmission ports are transmitted from any node device. In the storage type transmission method, the same message is sent to multiple node devices at the same time via a receiving port, and the node device that receives the message via a receiving port repeatedly sends the message again to transmit the message. A computer network characterized in that a counter is provided to indicate the number of messages received by the receiving port, and the values of the counter at different times are compared, and receiving ports for which the results are equal are determined to be abnormal. transmission method. 2. The computer network transmission system according to claim 1, wherein the abnormal port of the node device is determined to have returned to normal when a new message is received. 3. Configure a computer network by arbitrarily connecting a plurality of node devices equipped with a plurality of transmitting/receiving ports using the transmitting/receiving ports, and connect a plurality of node devices from any node device via one or more predetermined transmitting ports. In a storage type transmission method, in which the same message is simultaneously sent to each receiving port, and a node device that receives the message via a receiving port repeatedly sends the message again to transmit the message, A counter is provided to indicate the number of messages received by the node device, and the node device transmits the above message until the same message as the message transmitted or relayed from the node device is received from all the ports to which the node device sent the message. A transmission method for a computer network characterized by suspending the transmission and relaying of messages different from the above-mentioned messages that are subsequently transmitted from the same node device that has been transmitted. 4. When suspending the transmission of the above message, tag the message to be sent, do not tag messages that are different from the above message that are subsequently sent from the same node device, and after the delivery confirmation of the above message is completed, tag the message to be sent. 4. The computer network transmission system according to claim 3, wherein message transmission is suspended and released by passing the message. 5. The above-described arbitrary node device retransmits the transmitted message anew if the transmitted message has not been received from at least one receiving port of the predetermined ports within a specified time. The computer network transmission method described in Section 3. 6. The above arbitrary node device retransmits the message a certain number of times until the message is received by all of the one or more receiving ports determined corresponding to the sending port that sent the message, and then the above arbitrary node device still sends the above message. A patent characterized in that a reception port that does not receive data is regarded as a faulty port, retransmission is terminated, and thereafter, only the condition for completion of reception from the faulty port is set as the retransmission condition without waiting for the completion of reception from the faulty port. A computer network transmission system according to claim 3. 7. The above-mentioned arbitrary node device considers a receiving port that does not receive data even after a predetermined period of time has passed as a failed port, and considers a port that has newly received a message as having returned to normal, so that the receiving port for which transmission is pending is treated as a failed port. 4. The computer network transmission system according to claim 3, wherein the completion condition is changed as appropriate. 8. The above arbitrary node device includes an open collector circuit that operates a tristate logic element equivalently to an open collector logic element using an output control signal, and the open collector circuit operates the output of the tristate logic element. 4. The computer network transmission system according to claim 3, wherein the output disable timing by the control signal is delayed from the change timing of the input signal.