JPH0630499B2 - CSMA communication system - Google Patents

Description

TECHNICAL FIELD The present invention relates to a carrier detection multiple access (CSMA) communication system.

2. Description of the Related Art In the conventional CSMA system, a transmitting station sends one packet, and then a receiving station does not receive an acknowledgment (ACK) signal within a predetermined time until the transmitting station knows that a collision has occurred. Therefore, if the length of the packet to be sent is considerably long, it will take a long time when the occurrence of collision is identified by the transmitting station. Therefore, the packet length cannot be too long. However, if the packet length is too short, the ratio of the data part such as the preamble, destination address, source address, control part, and CRC part in the packet increases, and the overhead reduces the transmission efficiency. Also, by dividing into short packets, the number of ACKs increases, the overhead due to this increases, and the transmission efficiency also decreases.

In the CSMA / CD method, useless collision packet transmission can be interrupted by detecting collision during packet transmission.
Therefore, it is possible to increase the packet length in the initial stage of packet transmission, and it is possible to increase the throughput.

CD (collision detection) is performed by comparing the received data with the transmitted data in bit units at the node that is transmitting. In the case where the communication channel is a bus type, the node which is transmitting can receive at the same time, and at this time, the data received from the own node should be received. When the two are compared and a mismatch occurs, it is determined that the data transmitted from another node interfered, that is, a collision occurred. Generally, in a local area network, S / N is good enough and the probability of random data error due to noise can be regarded as extremely small, so such collision detection is possible.

Ethernet (by DEC, Intel, Xerox) and 1EE
With the CSMA / CD baseband scheme according to the E-802 standard, CD is relatively easy. Because the connection to the coaxial cable is made by the tap, the transmission data flows from the tap to the coaxial cable in both directions, is terminated at both ends, and the signal is absorbed.
Received data is tapped. Since the CD can be received at the same time from the tap being transmitted, the received data has almost no time delay with the transmitted data. Even if there is, it will be a certain value determined from the circuit constant. Therefore, it is easy to compare transmitted and received data bit by bit.

However, in order to ensure a clear detection of collisions, it is necessary that the signals from the other taps are not significantly attenuated by cables or the like. That is, the transmission loss of the transmission line system must be sufficiently small. For this reason, high-quality cables are required, which increases the price of cables and construction costs.

On the other hand, the problem of applying CSMA / CD to a wideband system is that CD is difficult. One reason is that in the wideband method, the transmission data is sent to the headend once and then re-transmitted to all the nodes, so that for the transmission data and the reception data, the node that transmitted the transmission data Even if there is, there is a big delay in time. This delay is due to the round trip propagation delay from the tap to the headend. Moreover, since the amount of delay varies depending on the tap, a considerably complicated configuration is required to compare transmitted and received data in bit units.

Another reason is that the transmission loss from the head end to the tap (and vice versa) is large in the wideband system, but the amount of the loss varies depending on the tap. If this variation in loss is too great, CD cannot be performed reliably.

As described above, the CSMA / CD method has the advantages that the packet transfer delay time under heavy load is shorter and the usage efficiency for the line capacity is higher than that of the CSMA method. However, as mentioned above, there are various problems caused by CD, so avoid this,
It is hoped that a CSMA system will appear, which has characteristics similar to those obtained when CD is performed, while making full use of the characteristics of the CSMA system.

Object The present invention has an object to provide a CSMA communication system which does not have a defect due to a CD and has substantially the same characteristics as the CSMA / CD system.

Configuration In order to achieve the above object, the node transmitting a packet transmits a packet after confirming that the carrier wave is not on the common transmission line for a time longer than the maximum round trip propagation delay time in the common transmission line. The CSMA communication system is characterized in that the node that receives the packet immediately returns the confirmation packet, and the transmitting node immediately transmits the next packet when receiving the confirmation packet. Hereinafter, a specific description will be given based on an embodiment of the present invention.

In order to help understanding of the present invention, the CSMA method with immediate acknowledgment (ACK) function will be touched on.As in the conventional OMNINET, the CSMA method with immediate acknowledgment function sends an ACK immediately after sending a data packet. If is not returned from the destination node, it is determined that the data packet was lost due to a collision or other reasons. If the partner node is operating normally and the transmission line is also operating normally, it would be appropriate for the network to determine that no ACK is returned as a collision. That is, ACK can be used as a kind of collision detection (detection of no collision). ACK there
Immediate acknowledgment schemes such as functional Ethernet and OMNINET make ACKs have higher priority than data packets, thereby avoiding ACK collisions in advance. However, even if the data packet to be transmitted is long, there is a drawback that the presence or absence of collision cannot be determined unless the entire one packet is transmitted.

In the present invention, the feature of the immediate acknowledgment system is used to avoid the collision of ACK in advance, and the long data packet following the ACK is also given high priority to avoid the collision in advance. Also, the first packet may be a collision, so
Use short data packets to minimize packet loss due to collisions.

As shown in FIG. 2, the maximum propagation delay time on the bus is τ
Then, the packet length corresponding to the round trip propagation delay time 2τ is the minimum packet length in the CSMA / CD system. The time γ ₁ from when packet transmission is started until collision is detected and the channel becomes empty is approximately γ ₁ = 2τ + α (1). Here, α is the sum of the time (delay) required for collision detection and the time required for collision emphasis. α is generally not so large, but if α is smaller than τ and can be ignored, then γ ₁ ≈ 2τ (1a).

In CSMA / CD and CSMA, the transmission efficiency (throughput) changes depending on the packet length. It is now assumed that the packet length is fixed and the transmission time is T. In order to detect a collision with the CSMA / CD method, T ≧ 2τ (2) must be satisfied.

In the CSMA method, T / τ is set to 10 in order to reduce retransmission due to collision.
More often than not. Generally, it is set at 100th place. If T / τ is small, the throughput and hence the channel capacity is small. Further, the smaller a = τ / T (3) (thus, the larger T / τ), the larger the throughput.

In the conventional CSMA with an immediate acknowledgment function, the time γ ₂ from the start of sending a data packet to the end of receiving an ACK packet is γ ₂ = T + 2τ + T _ACK (4). Where T is the transmission time of the data packet, T
_ACK is the transmission time of the ACK packet, and 2τ is the round-trip propagation delay time.

If the length of the ACK packet is the minimum packet length described above, then T _ACK = 2τ (5), so γ ₂ = T + 4τ (4a). If the data packet is sent within time γ ₂
Since it is considered that there was a collision when ACK was not received, this γ ₂ is the same as γ ₁ in CSMA / CD, and after the packet transmission is started, the collision is (indirectly) detected and the channel is Corresponds to the time until is empty.

By the way, in a long packet, T >> 2τ (6), so γ ₂ = T + 4τ >> 2τ (7) Therefore, γ ₂ >> γ ₁ (7a). That is, conventional CSMA means that the transmission time wasted when collision occurs is much longer than that of CSMA / CD.

In the CSMA method according to the present invention, the time γ ₃ from when packet transmission is started to when a collision is (indirectly) detected and the channel becomes empty is γ ₃ = T, as in equation (4). It becomes _init + 2τ + T _ACK (8). Here, T _init is the time required to send out the first packet at the time of starting to send data, and the length of this packet is set to the above-mentioned minimum packet length. Therefore, in this case, T _init = 2τ (9). Substituting Eqs. (5) and (9) into (8), we obtain γ ₃ = 6τ (10). This is because γ ₂ >> γ ₃ > γ ₁ (11). Therefore, in the communication method according to the present invention, the transmission time wasted when a collision occurs is a little longer than that by CSMA / CD, but the conventional CSMA It means that it will be drastically reduced than that by.

If the first packet is successfully transmitted (because of the return of ACK), the ACK is followed by a long packet. The ACK packet and the long packet are, in principle, configured to be avoided from collision, as described later.

As seen from Eq. (6), since T >> 2τ for long packets, the difference between Eqs. (10) and (1) is small. Therefore, the CSMA system according to the present invention has almost the same characteristics as CSMA / CD. It only outperforms CSMA / CD under very heavy loads.

The CSMA method according to the present invention will be described with reference to FIG. In this model, a plurality of nodes including nodes A and B are connected to a plurality of taps 14 of a coaxial cable 12 terminated by a terminating resistor 10. When sending data from node A to node B, node A sends an initial packet when it detects that there is no carrier on the coaxial cable 12 for tw time. This is indicated at point 1a in FIG. Where tw> 2τ (12). This tw corresponds to a so-called basic waiting time (BWT) in Ethernet or the like.

The initial packet is transmitted from time 1a to 2a, and arrives at the node B with a delay of τ (1b to 2b).
Immediately after the initial packet is completely received by the node B, an ACK packet is returned to the node A (3a-4).
a).

The first time any node attempts to send a packet, it must make sure that there is no carrier for time tw. By doing so, the ACK packet starts to be received by the transmitting node within 2τ, so that the ACK packet has a high priority. Moreover, since this ACK packet is sent only immediately after a specific reception, it is unlikely that a collision will occur.

In the communication method according to the present invention, long data is divided into a short (minimum length) initial packet and a long second packet, and if the initial packet is successfully transmitted, the subsequent transmission of the second packet is reserved. By doing so, the collision of the second packet is avoided. Specifically, AC at node A
Immediately after receiving the K packet (4b), the node A sends the second packet to the node B (5a-).
6a).

Similar to the reception of the ACK packet by the node A, the node B starts receiving the second packet within 2τ after the transmission of the ACK packet. Therefore, any other node connected to the coaxial cable 12 transmits the initial packet. Can't start. This is because according to this method, it is specified that any other node can send the initial packet only when it detects that there is no carrier for tw (> 2τ) or more. Therefore, this second packet has a high priority and in principle cannot cause a collision.

When Node B receives the second packet (5b-6
b), immediately return the second ACK packet to node A (7a-8a). As before, this ACK packet cannot cause a collision. Node A receives this ACK packet (7b
8b) to know that a series of packet transfer has succeeded (the node B has been definitely reached).

If the transmission data is longer, the node A
A third data packet (third packet) may be transmitted subsequent to the reception of the ACK packet. As before, this third ACK packet cannot cause a collision either. The Node B sends out the third packet immediately after receiving this third packet. This ACK packet cannot cause a collision either. Similarly, the data packets may be transmitted in the same manner as the fourth, fifth, ...

The second ACK packet and the subsequent ACK packets are not indispensable in the communication system according to the present invention. If the destination node (node B in this example) guarantees that the second packet and subsequent data packets can be reliably received, these ACK packets are unnecessary.
That is, only the first ACK packet is needed. If an extremely high-quality transmission system with very few transmission errors is used as the transmission line 12, data packets after the second packet cannot collide with each other, so that the destination node can be reliably transmitted without causing an error. To reach. In this case, after the third packet, transmission is started following the immediately previous data packet (that is, at least within tw).

In the above description, the transmission data is divided into the initial packet and the second packet (the third packet if necessary), but the initial packet does not include the data as the transmission information, and the second packet (and the subsequent packets) is not included. It may be configured such that data is put only in the data packet). In this case, the initial packet will be used exclusively for the purpose of collision detection. That is, the use of the channel is reserved for the second packet by the initial packet and the first ACK packet (first ACK packet). If the first ACK packet does not come back, it means that reservation could not be made, and it is thought that the cause is mainly collision.

FIG. 3 shows an example of a device configuration of a node that realizes the CSMA communication system according to the present invention. This applies to the node A or B connected via the tap 14 to the coaxial cable 12 shown in FIG.

The host machine (computer) 100 of node A is node B
Node A when trying to send data to the host machine 100 of
The host machine 100 sets the data and command to be sent in the buffer 102, and the transmission controller 110
To send to. The transmission control unit 110 decodes the command and sets the destination address (Node B in this case) in the transmission packet destination register 112.

Initialization of each unit such as the synchronization pattern generation unit 126, the CRC generation unit 120, the collision control unit 104, and the like will finally result in transmission, but before that, it must be confirmed that the carrier wave on the cable 12 is not tw for a certain period. This carrier wave detection is performed by the transceiver 150. In order to avoid complication of the drawing, a carrier detection signal line (not shown) is connected from the transceiver 150 to the transmission controller 110. Carrier detection, tw
If we determine that the time carrier of
The transmission control unit 110 starts transmission. If a carrier is detected, the subsequent measures for re-attempting after seeing the vacancy of the channel are slightly different depending on the so-called non-persistent, 1-persistent, or P-persistent method, but the latter carrier disappears. Transmission starts when time tw or more has elapsed.

The transmission is started by generating a sync signal from the sync pattern generator 126. The synchronizing signal is the modulator 124.
The appropriate encoding is performed by. For example, Manchester encoding may be used. The encoded signal is transmitted / received by the transmitter / receiver 150.
Sent on cable 12. This example is Ethernet
It is a baseband method using a coaxial cable such as.

When the synchronization signal has been sent, the transmission control unit 110 sends the contents of the transmission packet destination register 112 to the parallel-serial conversion unit 118 and converts it into a serial bit string. This bit string signal is sent to the modulator 124 through the CRC generator 120, encoded, and sent out on the cable 12 by the transceiver 150. When the contents of the transmission packet destination register 112, that is, the destination address have been transmitted, the contents of the transmission packet source address register 114, that is, the transmission source address are similarly transmitted.

Next, in this case, the packet type pattern generator 116 sends the type of the initial packet. Next (if any), (part of) the data in buffer 102 is similarly sent. The length of the data is set in the counter 108, and the data of the length corresponding to the count value of the counter 108 is transmitted by the command. When the transmission of data is completed, the transmission controller 110 instructs the CRC generator 120 to add a CRC code to the end of the packet. When the transmission of the CRC is completed, the transmission control unit 110 causes the transceiver 150 to stop the transmission of the carrier wave and set the reception mode. Here, the node A waits for the reception of the ACK packet from the node B.

Now, while node A is transmitting, the other non-transmitting nodes are receiving the signal on cable 12. For example, in the node B, when the transceiver 150 detects a carrier wave, the reception control unit 134 is notified by a signal line (carrier wave detection) not shown.

When the carrier wave is detected, the reception control unit 134 activates the synchronization pattern detection unit 148 and initializes the CRC inspection unit 144 and the counter 132. Therefore, the synchronization pattern detection unit 148
Detects a synchronization signal, and when synchronization is established, the reception controller 1
34 starts the operation of the CRC inspection unit 144 and activates the destination detection unit 138. When the destination detection unit 138 detects its own address (node B in this example), it notifies the reception control unit 134 of the fact, and the control unit 134 starts an operation to accept the packet. If it differs from its own address, the reception of that packet is ignored. That is, the node does nothing until there are no more carriers.

Thus, if the packet is for itself, it is accepted and then its source address is received, which is placed in the received packet source register 136.

It then receives the packet type, which is placed in the packet type register. Next (if any), the data is received and stored in buffer 102. At this time, the length of the data is counted by the counter 132.

After receiving the data, the reception control unit 134 causes the CRC checking unit 144 to check the validity of the CRC code. CRC is correct if there is no transmission error including collision, CRC if there is error
Will be an error. If there is an error, the reception control unit 134
Ignores the received packet. If there is no error and the packet type register 140 determines that the packet is an initial packet, the reception control unit 134 controls the transmission so that the ACK packet is immediately transmitted as soon as there is no carrier on the transmission path 12. Instruct the unit 110.

The transmission control unit 110 of this node B transmits an ACK packet in the same manner as the transmission of the initial packet from the node A, except that the transmission of the ACK packet is started immediately when there is no carrier on the cable 12. . That is, it does not wait until the time tw has elapsed. Node B
The destination address of the ACK packet returned from the node A to the node A corresponds to the source address stored in the received packet source register 136 when the initial packet is received,
Use this. That is, in the node B, a synchronization signal is transmitted from the synchronization pattern generator 148, then the destination address of the node A is received from the reception packet transmission source register 136, and then the transmission source address of the node B is transmitted from the transmission packet transmission source register 114. Further, the packet type pattern generator 116 sends an ACK type as the packet type. At this time, the data to be sent to node A is node B
If it is in the buffer 102, the data may be continuously sent. In general, statuses such as the air-blocking information of the receiving buffer 102 and the air-blocking information of the host machine 100 are sent as needed. Finally, the CRC generator 120 adds a CRC code, and the transmission of the ACK packet ends.

Node A sends the initial packet (point 2 in FIG. 1).
a), waiting for the arrival of the ACK packet. If there is no collision or transmission error, the ACK packet should start to be received within 2τ time (3b). If the length of the ACK packet is the above-mentioned minimum packet length, as can be seen from FIG.
It should be received within τ time (4b).

The node A receives the ACK packet in the same manner as the reception of the node B. If there is an error in the CRC, the packet addressed to itself does not arrive, no packets return, and if the ACK packet is not received within a certain time ta, it is considered that there was some transmission error due to collision. To be As described above, it is presumed here that the ACK packet was not returned due to the collision. Here, 4τ <ta <tw + 2τ (13) (where tw> 2τ, from equation (12)).

If the ACK packet is not received within ta, the collision control unit 104 of the node A backlogs the initial packet and operates to retransmit the initial packet according to a backoff algorithm that reduces the retransmission probability.

When the ACK packet is normally received, the transmission controller 110
Immediately transmits the second packet when there is data to be transmitted. This transmission is after the carrier wave is gone
It starts immediately without waiting for tw. Also, except that the packet type is the second packet, it is the same as the above two packet transmissions. This second packet, like the ACK packet, does not cause a collision in principle. Only the initial packet may cause a collision in this method.

If node B receives the second packet and the CRC is complete, it immediately sends an ACK packet to node A. The node A knows that the series of data transfer has succeeded by receiving the second ACK packet. If necessary, node A will forward the third and subsequent packets as well.

If there is data to be sent by a node other than node A,
The initial packet can be transmitted only when the channel is free for time tw or more, that is, when the carrier wave on the cable 12 is not more than tw. When an ACK packet or a second packet is transferred between node A and node B,
The channel is free for less than 2τ. That is, 2τ
Since the channel will not be emptied for a longer time, and tw> 2τ, these two nodes A and B
Packets cannot be sent from other nodes.

It should be noted that although the present invention has been described with respect to particular embodiments, the invention is not limited thereto. For example, the cable 12 used herein is a coaxial cable, but may be another, for example, a twisted pair cable. Also, as explained in baseband transmission,
It may be wide band transmission using an RF modem. In this case, a tree structure having a tree structure with the head end as a route is advantageously applied. Furthermore, the present invention can be advantageously applied to an optical communication network using an optical fiber that utilizes an optical coupling circuit network such as an optical star coupler, and the present invention can also be applied to a wireless system using light or radio waves.

Effect According to the present invention, the packet is transmitted after confirming that there is no carrier on the transmission path for a predetermined time, and then the confirmation packet and, if necessary, the data packet are alternately transmitted. I am trying. Therefore, even if the collision detection as in the CSMA / CD system is not performed, the performance is basically the CSMA system and almost the same performance as the CSMA / CD system can be achieved.

[Brief description of drawings]

FIG. 1 is a communication flow diagram for explaining the principle of the CSMA communication system according to the present invention, FIG. 2 is a flow diagram showing an example of collision in a channel, and FIG. 3 is a configuration example of a node realizing the present invention. It is a block diagram. Description of symbols of main parts 12 ... Common transmission line 14 ... Tap 104 ... Collision control unit 108, 132 ... Counter 110 ... Transmission control unit 112 ... Transmission packet destination register 114 ... Transmission packet transmission source register 116 ...... Packet type pattern generator 120 ...... CRC generator 134 ...... Reception controller 136 ...... Received packet sender register 138 ...... Destination detector 140 ...... Packet type register 144 ...... CRC checker 150 ...... Transceiver A, B ... Node

Claims (1)

[Claims] 1. A CSMA communication system in which a plurality of nodes are commonly connected to a common transmission line, and the plurality of nodes send a packet to the common transmission line when there is no carrier on the common transmission line. The first node having the data packet to be transmitted among the above nodes confirms that the carrier wave is not on the common transmission line for a time of the maximum round trip propagation delay time 2τ or more on the common transmission line, and the first node corresponds to 2τ time. A first node of the plurality of nodes, which first sends an initial packet of a length and waits for the arrival of a confirmation packet, receives the initial packet, and then a second node receives the initial packet, and immediately after receiving the initial packet, the second node has a length corresponding to 2τ time. Of the confirmation packet to the first node, and the first node sends out the data packet to be transmitted immediately after receiving the confirmation packet. When not receiving the acknowledgment packet within the predetermined time, CSMA communication method, characterized by retransmitting the initial packet.