JPS5997248A - Communication system - Google Patents

Abstract

PURPOSE:To guarantee to make even a lower-level node succeed in packet transfer within a certain time by postponing the transmission start of a next packet to give an opportunity to a node having a lower priority level when collision is detected. CONSTITUTION:The node succeeding in packet transfer is informed by the output of a collision detector 5 that at least one inferior node is requested to transmit and the next transmission request is postponed to give the time of packet transfer to the inferior node; and in this case, the transmission start of the inferior node is conducted at the time point when a carrier wave detected in an auxiliary receiving part Ra out of the carrier waves detected in duplication in a receiving part R and the auxiliary receiving part Ra by a carrier wave detector is dissipated. Thus, a packet interval is held down to a certain time shorter than a propagation delay time.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a data communication system, and specifically to a data communication system for transmitting variable length data packets between nodes of a local communication network. This is related to the Komie Nikeshiron e system.

[Technical background of the invention and its problems]

C8MA/5URVIVA using C8MA method in local communication network
T.

There is a method. In this method, as shown in Fig. 1, directional couplers Ct and Cr for transmission and reception are provided on a transmission line with one end turned back, and these are connected to the transmission section T and reception section R of each node. It is used for communication with N. In order to ensure that one of the packets survives if a collision occurs, an auxiliary receive combiner C is installed before the transmit combiner.
ra and an auxiliary receiving section Ra are provided. With the network configured in this way, each node 1) does not transmit a packet while a carrier wave is detected on R and Ra, and 2) even after starting to send a packet, a carrier wave is detected on Ra. If so, stop sending.

In this way, if any node i detects that there is no received carrier wave and starts sending out a packet, due to the functions 1) and 2) above, nodes lower than the node will not be able to send out the packet. Even if it is, it will stop. Therefore, if the upper nodes do not start sending out packets before the carrier wave sent by the node reaches R of each upper node, the node can complete the transmission. In this way, when a plurality of nodes detect an empty channel and start sending out packets, the highest node always survives, and the line utilization rate is improved compared to the similar C8MA system or C8MA/CD system.

However, if the higher node always takes priority like this,
A lower node can successfully forward a packet only if all nodes above it do not send packets. Therefore, for example, if a node that postpones or cancels packet transmission according to 11.2) above always starts packet transmission the next time no carrier waves are detected on R and Ra, lower nodes will A packet can only be forwarded when all nodes in the node have no packets to send. In this way, lower nodes are at a significant disadvantage in channel acquisition opportunities, and since they cannot dynamically change their priorities, lower nodes have no guarantee that they will successfully transfer packets within a certain amount of time. do not have.

Also, while carrier waves are detected in R and Ra, other nodes except the transmitting node do not send out packets, so the time between two consecutive packets is the worst case as shown in Figure 2. In this case, the round trip delay time of the transmission line is about the same. Furthermore, if multiple nodes detect an empty channel after packet transfer and start sending packets, the highest node among them will be the last to know that no carrier is detected in R, so the time between two packets will be The packet with the longest interval will survive. This means that when the packet length is short, the packet interval time increases compared to the packet transfer time, leading to a decrease in line utilization.

[Purpose of the invention]

The present invention improves the above-mentioned drawbacks of the prior art, and aims to ensure that even lower nodes can successfully transfer packets within a certain period of time, and to shorten the packet interval time and improve line utilization. do.

[Summary of the invention]

In the network described above, each node can receive all the signals on the network, so a) If the packet transfer is successful, it receives the packet it sent and determines whether it caused a collision or not. A mechanism for determining is provided to the receiving unit R of each node. Let CF be a flag indicating the result of this determination, and let it be as follows. Furthermore, b) a counter is provided that is incremented if CF=On and cleared (becomes 0) if CF=Off when the packet transfer is successful. Let this counter be BC. The BC0 value has an upper limit, and when it continues to increase and exceeds the upper limit, it returns to 0 or a relatively small value, and changes cyclically. In addition, along with BC, c) a counter is provided which is counted every time the value of BC increases, with the new value as an initial value, and which thereafter continues to decrease as time elapses until it reaches 0. Let this counter be DC.

In addition, the circuit that detects the carrier wave as follows is constructed by R and Ra.
Connect to. The output of this circuit is the flag SF, which changes as follows.

d) When 118F=off, set 5F=nn when a new carrier wave is detected in R or Ra.

21 If the carrier wave is not detected in Ra even when SF=on, and the carrier wave that has been detected only in R is no longer detected, 5F=off is set at that point.

If a carrier wave is detected in both Ra and R after 313F=on, 5F=off is set when a carrier wave is no longer detected in Ra.

Furthermore, the flag RF is set to e) a flag that is on while a carrier wave is detected in Ra and is off otherwise, and the operation of each node is determined as follows.

A node that has a packet to send is 1) D
If C>0, do not transmit until DC=0.

2) Even if DC=O, if 5F=on, it will not be transmitted, and 5F
Wait until = off and start transmission.

3) Even after starting transmission, if RF=on, transmission is stopped. At this time, 5F=on, so SF=off again.
Wait until then to start sending.

4) If the transfer is successful and the packet you sent is received, CF = oni, increase BC and set the value to DC.
Set to . If CF=off, clear BC.

The key point of the present invention is that the collision detection mechanism in a) above allows a node that has successfully transferred a packet to know that at least one lower mate has a transmission request, postpones the next transmission request, and sends a request to the lower node. The above d
) is carried out at the point when there is no carrier detected in Ra out of the carrier waves detected redundantly in R and Ra, thereby suppressing the packet interval to a short constant time compared to the propagation delay time. In particular. however,
Since a higher-order node may continue to start packet transfer, the lower-order node repeats the operation 3) above until the packet transfer is successful.

Furthermore, in order to guarantee packet transfer within a certain period of time as will be described later, it is necessary to pay attention to the upper limit of the value of the counter BC and how to increase it. That is, in order to guarantee that all nodes can successfully transfer packets within a certain amount of time (which may vary from node to node) even under extreme conditions, the upper limit of the BCO value is ((number of nodes - 1) x (
The maximum packet transfer time must be greater than or equal to
On the other hand, if the upper limit of the BC value is too large or increases too rapidly, a situation will arise in which the upper node is at a disadvantage.

〔Effect of the invention〕

The present invention ensures that any node on the network can transfer packets within a certain amount of time. However, the time usually differs from node to node. By improving the carrier wave detection function and shortening the packet interval time when multiple nodes have packets to transmit (when the packets are consecutive), the line utilization rate improves.

[Embodiments of the invention]

FIG. 3 shows an example of a hardware configuration for implementing the present invention. 1, 2.3 are transceivers including directional couplers. The 4 carrier wave detector performs the function d) above using Ra and R as inputs, and the 7 access controller receives S.
Outputs F and RF'. The collision detector 5 examines the contents of the reception buffer based on the notification from the transmission/reception controller 6, and outputs a CF to the access control 7.

The 6 transmission/reception controller performs a transmission operation in response to a request from the 7 access controller, and notifies the 7 access controller of the end of a reception operation. The access controller 7 includes counters BC and DC, and performs the functions 1) to 4) above. 8 and 9 are buffers for reception and transmission, respectively, and communications with 3 receivers, 2 transmitters, and 10 computers are controlled by a transmission/reception controller 6. Controls communication with 10 Combi coaters.

Ten computers are connected to this network node. This is a device that performs communication using variable length data packets.

In such a configuration, if the value of the counter BC changes as the number of nodes transfer time O → 1 → 2 → 3 → 1 → 2 → 3 →..., which node is the number of nodes 4? FIG. 4 schematically shows the situation in which nokkerato is sent out when it always has a packet.

In the figure, ■] indicates packet transmission, and 1/l indicates a packet transmission standby state when DC>0. Even though the upper nodes always have packets, the lowest node 4 also succeeds in sending packets. In the case of Figure 4, the transmission is successful after waiting 8 times the packet transfer time, but in this case, node 4 will not wait longer than 12 times the packet transfer time. You can know. As shown in FIGS. 5 and 6, when node 1 takes a waiting time that is three times the packet transfer time, node 4 cannot transfer during that waiting time because node 2,
This is a case where one of the packets 3 transfers 2 packets and the other transfers 1 packet. In this case, the node that transferred two packets would have waited for more than three packets to transfer just before that. Then, the waiting time overlaps with the waiting time of node 1, and two packet transfer time channels become available.If one node, node 2 or 3, sends two packets in succession, node 4 always receives a transmission request. This is not possible because the node has the following information, and node 4 succeeds in forwarding the packet. Therefore, node 4 successfully transfers the packet after the transmission request is generated and before node 1 finishes waiting for the next three packet transfer times. In this case, the lowest node 4 is still at a disadvantage in terms of packet transfer success rate (bandwidth usage rate), but if you increase the upper limit of BC or increase the rate of increase, the difference decreases, and in extreme cases In some cases, higher nodes are at a disadvantage.

In implementation, it is important that there is a method for recognizing packet breaks on the receiving side, since lower nodes abort their own transmissions and give priority to packets from higher nodes. This also relates to collision detection upon successful packet transfer, which is a basic element of the present invention. The following method can be used to ensure packet spacing.

Prior to packet transfer, a constant pulse (train) is sent, and then Kerat transmission is started after the shortest packet interval time has elapsed. Transmission is assumed to have started from the moment the pulse train was sent out, and when the pulse train arrives from the upper node, the subsequent transmission operation is immediately stopped. Using this method, the occurrence of a collision can be detected by the fact that the pulse train and immediately before the packet are not normal keratotransfers (for example, only the pulse train), or the pulse train is destroyed.

[Brief explanation of the drawing]

Figure 1 shows the conventional technology CSMA/S URV I.
Conceptual diagram of VAL system, Figure 2 is C8MA/5URVIV
In the AL system, the packet interval time when the two nodes furthest from the turning point send packets one after another is
Figure 3 is a block diagram of an embodiment of the present invention, and Figure 4 is a diagram showing that the delay time is close to the round trip delay time of the bus, and Figure 4 is a diagram showing a concrete change pattern of the BC value, making it possible to transfer packets even at the lowest node. 5 and 6 are explanatory diagrams used to show that this can be achieved within a certain period of time. 2: a transmitter including a directional coupler; 3: a receiver including a two-way coupler; 4: a carrier wave detector; 5: a collision detector; 6: a transmission/reception controller; 7: access controller, 8: reception buffer, 9: transmission buffer, 10: computer, 11:
Folded transmission line.

Claims (1)

[Claims] (II) In order to achieve asynchronous communication by interconnecting a plurality of nodes by a common bus with one end loopback, packets from nodes far from the loopback point are prioritized, and each node has a certain amount of time. A mechanism that detects whether a successful packet transfer collides with the packet transmission of a node near the turnaround point (lower priority) by also receiving the packet, in order to provide an opportunity for packet transfer within a short period of time. and a mechanism that postpones the start of sending the next packet when a collision is detected, giving a node with a lower priority an opportunity to forward the packet. (2) One end. In order to achieve asynchronous communication by interconnecting multiple nodes through a common loopback bus, priority is given to packets from nodes far from the loopback point, and each node is given an opportunity to transfer packets within a certain amount of time. In addition, when a packet transfer is successful, there is a mechanism that detects whether it collides with the packet transmission of a node near the turnaround point (lower priority 11) by also receiving the packet, and if a collision is detected. , a mechanism that postpones the start of sending the next packet to give a node with a lower priority an opportunity to forward the packet, and a loopback transmission path to shorten the packet interval time when packets are sent consecutively on the transmission path. A communication method that includes a mechanism that allows transmission when a carrier wave is detected simultaneously on both transmission lines (transmission line and reception line) after an appropriate packet interval time has elapsed from the time when carrier waves are no longer detected on the transmission line. .