JPH0993276A - Ring type network access control method and ring type network system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simplify a master node or to eliminate the need for the master node by providing a cyclic delay adjustment buffer so as to warrant a worst- case delay. SOLUTION: The content of the payload of a cell received by a master node 211 is copied to a transmission queue corresponding to the flow of a cyclic delay adjustment buffer 651 subjected to transmission queue for each flow. Then to which flow each cell in a frame is allocated is judged and communication data stored in a transmission queue corresponding to the cyclic delay adjustment buffer 651 are inserted into the payload of the cell and the resulting cell is transmitted (621). The cell allocated to each flow is arranged uniformly in the frame as to each flow.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an access control method for a ring network in which a plurality of nodes are connected in a ring by links, and a ring network system using the access control method.

[0002]

2. Description of the Related Art Communication such as voice, which is severely restricted in transfer delay time, has been conventionally supported by a circuit switching service. In the circuit switching service, the bandwidth of the divided communication link is divided into individual flows (traffic flows).
Limited to. Therefore, each flow can surely use the allocated band regardless of the data generation status of other flows, so that the transfer delay time can be guaranteed in advance. On the other hand, in the packet switching service, since a plurality of flows dynamically share the bandwidth of the communication link, it is difficult to guarantee the transfer delay time because a complicated mechanism is required.

Time-division multiple access (hereinafter referred to as TDMA) is well known as an access control method applied to provide a circuit switching service in a ring network. As an example of this TDMA type ring network, the first of "Optical LAN-Basics and Applications-" (supervised by Opto-Industrial Technology Promotion Association, February 25, 1988)
On pages 77 to 179, a ring type optical LAN in which a fixed length frame is circulated on a ring is introduced. Further, as a standard TDMA type ring type optical LAN, FDD is used.
I-II (Fiber Distributed Data Interface 2) is known, and its description "FDDI and FDD"
I-II: Architecture, Protocols, and Performance
317 (of ARTECH HOUSE, INC. In 1994)
~ Introduced on page 327. In particular, FIG. 7.2 on page 321, shows well the features of the TDMA method. These ring-type optical LANs provide combined support for circuit switching and packet switching, but in the following, the points related to circuit switching services will be mainly described with reference to the drawings.

FIG. 15 is a conceptual diagram for explaining a ring network based on the conventional TDMA method. In the figure, 21 is a master node, 22 is a slave node, and 3 is these master node 21 and slave node 22.
The link 3 connects between the two in a ring shape.

FIG. 16 is an explanatory diagram showing the format of a fixed-length frame generated by the master node 21. In the figure, 49 is the frame, and 59 is a fixed-length slot obtained by dividing the frame 49 into a plurality of slots. One or more specific slots 59 are assigned to each flow generated by each slave node 22. The frame 49 may be called a cycle, but in the following, it will be unified as a frame. Each of these slots 59 is identified by an offset position from the beginning of the frame 49. Therefore, although not shown here, some marker indicating the beginning of the frame 49 is required for that purpose. In the following description, the length of the frame 49 (frame length: the number of slots in one frame) is N [slot].

FIG. 17 is a block diagram showing the internal structure of the master node 21. In the figure, 31 is an incoming link from the ring, 32 is an outgoing link to the ring, 61 is a receiving unit that receives the frame 49 or the slot 59 from the incoming link 31, and 62 is a frame 49 or the slot 59 is transmitted to the outgoing link 32. A transmitting unit, 64 is a frame generating unit that continuously generates fixed-length frames 49, and 65 is a round delay adjustment buffer for keeping the round delay of the ring at an integral multiple of the frame length. In order to continuously and periodically generate the frames 49, it is necessary to set the round delay of the ring to an integral multiple of the frame length.
5 is necessary for this. The loop delay adjustment buffer 65 is a LAB (Latency Adjustment Buff).
er) is sometimes called, but in the following will be unified into the expression of the round delay adjustment buffer.

FIG. 18 is a block diagram showing the internal structure of the slave node 22. In the figure, 31 is an incoming link from the ring, 32 is an outgoing link to the ring, 71 is a receiving unit for receiving the frame 49 or the slot 59 from the incoming link 31, and 72 is transmitting the frame 49 or the slot 59 to the outgoing link 32. The transmission unit 73 is a transmission queue that is provided for each flow and stores each communication data.

Next, the operation of the master node 21 will be described with reference to FIG. The master node 21 periodically generates a frame 49 in the frame generation unit 64. At the time of network initialization, the loop delay adjustment buffer 65 is empty, so the transmitting unit 62 transmits the frame 49 while leaving each slot 59 empty. When the frame 49 goes around the ring once, the receiver 61 receives the received frame 4
The contents of each slot 59 of 9 are stored in the round delay adjustment buffer 6
Copy to 5 in order. The transmitter 62 is a frame generator 64
When the timing of the head slot of the frame 49, that is, the slot 59 of # 1 is given, the reading of the round delay adjustment buffer 65 is started and the data at the head (data corresponding to the slot 59 of # 1) Is inserted into the slot 59 of # 1 and transmitted to the outgoing link 32. It should be noted that the delay in the round delay adjustment buffer 65 is a maximum of N slots, and the round delay adjustment buffer 65 has 1 delay.
A buffer capacity for frames is required.

Next, the operation of the slave node 22 is shown in FIG.
This will be described with reference to FIG. Here, the slave node 22 operates in the following three ways depending on the attribute of the received slot 59.

When the slot 59 assigned to the flow originating from the own node is received from the incoming link 31, the receiving section 71 empties the contents of the slot 59 and passes it to the transmitting section 72. The transmission unit 72 inserts the communication data stored in the transmission queue 73 of the flow into the received empty slot 59 and transmits the communication data to the outgoing link 32. When communication data is not stored in the transmission queue 73 of the flow, the slot 59 is transmitted to the outgoing link 32 as it is.

When the slot 59 assigned to the flow addressed to the own node is received from the incoming link 31, the receiving section 71 copies the content of the slot 59 and then passes it to the transmitting section 72 as it is. The transmitter 72 transmits the received slot 59 to the outgoing link 32. If the content of the received slot 59 is empty, the receiving unit 71 does not copy the content.

When another slot 59 is received from the incoming link 31, the receiving section 71 passes the slot 59 to the transmitting section 72 as it is, and the transmitting section 72 transmits the received slot 59 to the outgoing link 32.

In any case, the time from the reception of a certain slot 59 to the transmission of the slot 59, that is, the node relay delay of the slot 59 is constant.

The delay time required for data transfer in this TDMA type ring network is as follows. In the following description, the link transfer rate of 1
The slot time is a unit of time, and the amount of data transferred in one slot is a unit of data amount.

Here, allocating n time slots to a certain flow means allocating a band w represented by the following equation [1].

W = n / N [slot / slot time] ... [1]

For the flow allocated by the bandwidth w, the total delay time for data corresponding to B [slot] generated at an arbitrary time to reach the receiving node from the source node is 1) from data generation. In addition to the "queuing delay" until the slot for transmitting the final data passes through the transmission node, 2) the "adjustment delay" added by the loop delay adjustment buffer 65 of the master node 21, and 3) the "adjustment delay", the final Is the total "propagation delay" required to propagate the data from the source node to the destination node.

If there is no untransmitted communication data in the transmission queue 73 at the time of data generation, data of n [slot] can be transmitted by 1 frame time at the latest.
After that, since n [slots] can be transmitted every frame time, the queuing delay is less than or equal to the value expressed by the following equation [2].

(B / n) × N [slot time] ... [2]

The adjustment delay is maximum N [slot time] as described above. Further, the propagation delay is fixed depending on the system configuration such as the number of nodes and the distance between the nodes, but the propagation delay between any two nodes is smaller than the propagation delay around the ring (this is Dfix). Therefore, if the amount of data generated per traffic is B [slot] or less and the generation interval is longer than B / w [slot time], then w [slot / slot time] By allocating the band, the worst delay shown in the following formula [3] can be guaranteed.

(B / n + 1) × N + Dfix [slot time] ... [3]

On the contrary, for a flow in which the data of B [slot] must be transferred within the requested worst delay Dreq [slot time], a slot having a value equal to or larger than the value given by the following expression [4] is set per frame. It is sufficient to allocate the number n.

N: B / {(Dreq-Dfix-N) / N} [slot] ... [4]

Here, {(Dreq -D in the above equation [4]
fix-N) / N} [slot time] indicates how many frames the time allowed for the queuing delay corresponds to. The allocated bandwidth w at this time is as shown in the following expression [5].

W: B / (Dreq −Dfix −N) [slot / slot time] ... [5]

By the way, a ring network in which the slots 59 in the TDMA system ring network are replaced with cells can be easily analogized. This is a network in which cells, which are fixed-length packets, are used as a transfer unit, and the network will be called a cell network below.

FIG. 19 is an explanatory diagram showing a frame format used in such a ring type cell network. In the figure, 4 is a frame, 5 is the cell, 51 is the header of the cell 5, 52 is the same payload, and 512 is a flow identifier of the header 51. The cell 5 has a fixed length like the slot 59, but is composed of a header 51 and a payload 52 as shown in the drawing.
Has a flow identifier 512. Since the cell 5 can identify to which flow the cell 5 is assigned by this flow identifier 512, a marker corresponding to the head of the frame is unnecessary. For example, in a standard cell format defined by the Asynchronous Transfer Mode (ATM) forum and the like, a virtual path identifier (VPI) and a virtual channel identifier (VC) are used.
The field of I) corresponds to this flow identifier 512.

Next, the operation of the master node 21 in the case of a ring type cell network will be described with reference to FIG. The frame generator 64 periodically generates a frame 4 of fixed length (N cells). At that time, the flow identifier 512 of each cell 5 in the frame 4 specifies which flow is assigned. Since the round trip delay adjustment buffer 65 is empty at the time of network initialization, the transmission unit 62 transmits the frame 4 while leaving the payload 52 of each cell 5 empty. When the frame 4 goes around the ring once, the receiver 61 receives the payload 5 of each cell 5 of the received frame 4.
2 are sequentially copied to the loop delay adjustment buffer 65. When receiving the timing of the beginning of the frame 4 from the frame generation unit 64, the transmission unit 62 receives the round delay adjustment buffer 65.
Reading is started, and thereafter the loop delay adjustment buffer 65
While inserting the read data into the payload 52 of the cell 5, the cell 5 is sequentially transmitted to the outgoing link 32.

Next, the operation of the slave node 22 in the case of a ring type cell network will be described with reference to FIG. Also in this case, the slave node 22 operates in the following three ways depending on the attributes of the received cell 5.

When the cell 5 whose flow identifier 512 indicates the flow originating from the own node is received from the incoming link 31, the receiving unit 7
1 empties its payload 52 and passes it to the transmitter 72.
The transmitter 72 inserts the communication data stored in the transmission queue 73 of the flow into the payload 52 and transmits the cell 5 to the outgoing link 32. If communication data is not stored in the transmission queue 73 of the flow,
The cell 5 with the payload 52 left empty is transmitted to the outgoing link 32.

When the cell 5 whose flow identifier 512 indicates the flow addressed to its own node is received from the incoming link 31, the receiving unit 7
1 copies the content of the payload 52 and then passes it to the transmitting unit 72 as it is. The transmitter 72 transmits the received cell 5 to the outgoing link 32. If the payload 52 of the received cell 5 is empty, the receiving unit 71 does not copy the content.

When the other cells 5 are received from the incoming link 31, the receiving unit 71 passes the cells 5 to the transmitting unit 72 as they are, and the transmitting unit 72 transmits the received cell 5 to the outgoing link 32.

Regarding the ring type cell network, the discussion regarding the delay time is exactly the same as in the case of the TDMA type ring type network.

The conventional ring type network described above and the ring type cell network which is a modification thereof have the following four characteristics. 1) The master node 21 always exists. 2) The master node 21 strictly specifies the service timing. Here, the service timing is a slot position or cell position (timing at which communication data can be inserted) that can be used by a certain flow. 3) Service timing is provided periodically in a cycle common to all flows. The frame length corresponds to the common cycle. For this purpose, a loop delay adjustment buffer is required. 4) The service timing is specified for each individual flow.

[0035]

Since the conventional ring network is configured as described above, it is possible to achieve the object of providing a communication service that guarantees the worst delay for traffic satisfying predetermined characteristics. However, there is a problem that the master node 21 is required, the master node 21 needs the loop delay adjustment buffer 65 for one frame, and further, an adjustment delay of maximum one frame is uniformly added to any flow. It was

The present invention has been made in order to solve the above problems, and it is possible to simplify the master node or eliminate the need for it while providing the service that guarantees the worst delay. The purpose is to obtain a type network.

[0037]

According to a first aspect of the present invention, there is provided a ring network access control method, wherein a master node divides the contents of a received cell payload into transmission queues for each flow, and a round delay adjustment buffer. Of the communication data stored in the corresponding transmission queue of the round trip delay adjustment buffer in the payload of that cell by determining to which flow each cell in the frame is assigned The cells to be inserted and transmitted and the cells to be assigned to each flow are arranged evenly in the frame for each flow.

In the ring network access control method according to the second aspect of the present invention, the flow is classified according to the required worst delay time, and the master node transmits the contents of the payload of the received cell for each class. The round trip delay adjustment buffer is copied to the queue-separated round trip delay adjustment buffer that corresponds to the class to which the flow belongs, the class to which each cell in the frame is assigned is determined, and the round trip delay adjustment buffer is added to the payload of the cell. The communication data stored in the corresponding transmission queue of (1) are inserted and transmitted, and the cells assigned to each class are evenly arranged in the frame for each class.

In the ring network access control method according to the third aspect of the present invention, the reset instruction is issued from the master node at a constant cycle, and the slave nodes each transmit queues storing communication data,
After the number of transmission cells reaches a predetermined number of transmission permitted cells, the transmission from the transmission queue is stopped, and when the reset instruction is received, the number of transmission permitted cells is reset.

In the ring network access control method according to the present invention, the master node generates the number of cells permitted to be transmitted for each flow together with the reset instruction.

The access control method of the ring network according to the fifth aspect of the present invention is such that the number of cells permitted to be transmitted for each flow is held in the slave node.

In the ring network access control method according to the present invention, all nodes form a ring network, and each node transmits a frame period common to the entire ring network. The communication data stored in each transmission queue is transmitted at a constant cycle divided by the number of transmission-permitted cells for each queue.

In the ring type network system according to the invention of claim 7, the master node and one or more slave nodes are connected in a ring by a link, and in the master node, each flow is equalized in the frame. A frame consisting of a fixed number of cells to which cells are allocated is continuously generated from the frame generation unit, and the cells received by the reception unit are transferred to the flow of the cell of the circular delay adjustment buffer having the transmission queue for each flow. The data is copied to the corresponding transmission queue, and the transmission unit determines to which flow the cell of the frame generated by the frame generation unit is assigned, and the communication data of the corresponding transmission queue of the round trip delay adjustment buffer is used as its payload. In the slave node, the payload is emptied when the cell of the flow originating from its own node is received by the receiving unit. , When the cell of the flow addressed to the own node is received, the contents of the payload are copied, and the other cells are passed to the transmitter as they are, and when the cell of the flow originated from the own node in which this payload is emptied is received. The transmission unit reads communication data from the corresponding one of the transmission queues provided for each flow, inserts it into its payload, and transmits it.

In the ring type network system according to the present invention, a master node and one or more slave nodes are connected in a ring by a link, and each master node classifies each flow according to the worst delay time required. The divided classes are assigned cells so that they are even within the frame. A frame consisting of a fixed number of cells is generated consecutively from the frame generator, and the cells received by the receiver are Copy to the transmission queue corresponding to the class of the cell of the round trip delay adjustment buffer that has a transmission queue for each, and the transmission unit determines to which class the cell of the frame generated by the frame generation unit is assigned. Then, the communication data of the corresponding transmission queue of the round trip delay adjustment buffer is inserted into its payload and transmitted, and the slave node receives it. When a cell of a flow originated from its own node is received in the section, its payload is emptied, and when a cell of the flow addressed to its own node is received, the content of the payload is copied, and other cells are sent to the transmission section as they are. When a cell with this empty payload is received, the transmitter reads communication data from the transmission queue corresponding to the flow that belongs to the class to which the cell is assigned, inserts it into the payload, and transmits it. is there.

In the ring type network system according to the invention of claim 9, the master node and one or more slave nodes are connected in a ring shape by a link, and the master node indicates a reset instruction and the number of transmission-permitted cells of each flow. A reset cell including the reset cell is arranged at a predetermined position, and a frame composed of a certain number of cells is continuously generated from the frame generation unit,
The cells other than the reset cell received by the receiving unit are sequentially copied to a buffer having a capacity of one cell, the contents of the buffer are transmitted by the transmitting unit according to the boundary timing given by the frame generating unit, and the slave node , When the receiving unit receives a cell of a flow originating from its own node, the payload is emptied, and when a cell of a flow addressed to its own node is received, the content of the payload is copied and other cells are sent as they are. When the transmission unit receives a cell whose payload is empty, the transmission unit reads communication data from the transmission queue only until the corresponding transmission cell counter counts the number of transmission-permitted cells. Is inserted in the payload of the cell and transmitted,
When the reset cell is received by the receiving unit, the reset cell monitor resets the transmission cell counter based on the number of transmission-permitted cells of the reset cell.

According to a tenth aspect of the present invention, in a ring type network system, a master node and one or more slave nodes are connected in a ring by a link, and the master node resets at cell boundary timing and frame cycle. Is generated from the reset generation unit, the reception unit turns off the reset indicator of each received cell and passes it to the transmission unit, and when the reset generation unit generates the reset timing, the transmission unit receives from the reception unit. Transmitting with the cell reset indicator turned on, the slave node holds the number of cells permitted to be transmitted in the transmission queue corresponding to the number of transmission permitted cells register, and the receiving unit receives the cells of the flow originating from the own node. Then, the payload is emptied, and when the cell of the flow addressed to the own node is received, the content of the payload is copied and The other cells are passed as they are to the transmission unit, and when the transmission unit receives a cell whose payload is empty, the transmission queue is removed from the transmission queue only until the corresponding transmission cell counter counts the number of transmission permitted cells. After reading the communication data, inserting it into the payload of the cell and transmitting it, and when the receiving section receives the cell with the reset indicator turned on, the reset cell monitor resets the transmission cell counter, This is performed based on the number of transmission permitted cells stored in the transmission permitted cell number register.

In the ring type network system according to the invention of claim 11, when a plurality of nodes are connected by a link in a ring shape and each node receives a cell of a flow originating from its own node, the payload thereof is sent. When the cell of the flow addressed to the own node is emptied and copied, the contents of the payload are copied, the other cells are passed to the transmitter as they are, and the transmitter receives the empty cell payload from the receiver. If so, the framer, which is defined in common for the entire ring network, is divided by the number of transmission-enabled cells for each transmission queue and read from the corresponding transmission queue by the shaper, and the communication data bundled by the global transmission queue is collected. , Is inserted in the payload of the cell and transmitted.

[0048]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below. Embodiment 1. 1 is a block diagram showing the configuration of a master node in a ring network system according to a first embodiment of the present invention. In the figure, 31 is an incoming link (link), 32 is an outgoing link (link), 61
1 is a receiving unit for receiving the cell 5 from the incoming link 31;
Is a transmission unit that transmits the cell 5 to the outgoing link 32, 641 is a frame generation unit that continuously generates fixed-length frames, 6
Reference numeral 51 is a round delay adjustment buffer provided with a transmission queue having a capacity of one cell for each flow. Reference numeral 211 denotes these receiving unit 611, transmitting unit 621, and frame generating unit 64.
1 and a round delay adjustment buffer 651.

FIG. 2 is an explanatory diagram showing an example of cell allocation to each flow in the first embodiment. In the figure, 41 is a frame generated by the frame generation unit 641 and 5 is a cell. According to the illustrated example, the frame length of the frame 41 is a fixed length of 8 cells, and four cells 5 per frame are used for the first flow (flow (1)) and the second flow (flow (2) )), Two cells 5 are allocated, and one cell 5 is allocated to the third flow (flow (3)) and the fourth flow (flow (4)). Here, the cells 5 assigned to each flow are arranged evenly in the frame 41 for each flow. In this case, the flow (1) has a 2-cell cycle #
Each cell 5 of 1, # 3, # 5, and # 7, each cell 5 of # 2 and # 6 with a cycle of 4 cells for flow (2), and eight cells for flow (3) and flow (4). Cycle # 4 cell 5 or #
Eight cells 5 are assigned respectively. Below,
These cycles are called cell allocation cycles for each flow.
In addition, in each node, which flow is to be generated and which flow is to be received is set in advance, and the setting method may be manual setting or automatic setting.

Next, the operation of the master node 211 according to the first embodiment will be described. Frame generator 64
1 periodically generates the frame 41 shown in FIG.
At that time, the flow identifier 5 of each cell 5 in the frame 41
In 12, the flow to which the flow is assigned is designated. At the time of network initialization, all the transmission queues of the round trip delay adjustment buffer 651 are empty, so the transmission unit 621 determines that each cell 5
The frame 41 is transmitted to the outgoing link 32 while leaving the payload 52 of the frame empty. When the receiving unit 611 receives the frame 41 that has made one round in the ring, it receives the flow identifier 5 of each cell 5.
12, the flow to which the cell 5 is assigned is determined, and the payload 52 of the received cell 5 is determined.
The content of is copied to the transmission queue of the corresponding flow of the round delay adjustment buffer 651. When the transmission timing of the cell 5 is given from the frame generation unit 641 to the transmission unit 621, the transmission unit 621 determines to which flow it is assigned based on the flow identifier 512, and transmits the corresponding flow of the round delay adjustment buffer 651. Read the data from the queue, insert it into the payload 52, and exit link 3
Send to 2.

Here, regarding the flow (1), since the cell allocation period is 2 cell times, the transmission timing is given once every 2 cell times, and the flow (1) of the round delay adjustment buffer 651 receives the transmission timing. When the corresponding transmission queue is empty, the data copied to the transmission queue is transmitted by 2 cell time at the latest. In addition, since the receiving unit 611 receives the data of the flow (1) from the incoming link 31 every two cell times, the next data is copied to the transmission queue of the flow (1) after it becomes empty. Will be. Therefore, the delay in the round delay adjustment buffer 651 of the flow (1) is 2 cell time at maximum,
That is, it becomes less than the cell allocation period. The same applies to the other flows (2) to (4), and the maximum delay in the round delay adjustment buffer 651 is equal to each cell allocation cycle.

The configuration of the slave node used in the first embodiment is similar to that of the slave node 22 of the conventional ring network shown in FIG. 18, and its operation is the same as in the conventional ring cell network. Is the same as.

That is, the receiving unit 71 uses the flow identifier 51.
When 2 receives the cell 5 indicating the flow originating from the own node from the incoming link 31, the payload 52 of the cell 5 is emptied and passed to the transmitting unit 72. The transmitter 72 inserts the communication data stored in the transmission queue 73 of the flow into the payload 52 and transmits the cell 5 to the outgoing link 32. If no communication data is stored in the transmission queue 73 of the flow, the cell 5 with the empty payload 52 is transmitted to the outgoing link 32.

When the flow identifier 512 receives the cell 5 indicating the flow addressed to the own node from the incoming link 31, the receiving unit 71 copies the contents of the payload 52, and then
It is delivered to the transmission unit 72 as it is. The transmitter 72 transmits the received cell 5 to the outgoing link 32. If the payload 52 of the received cell 5 is empty, the receiving unit 71
Does not copy its contents.

Further, the other cells 5 are entered into the link 31.
When receiving from the cell 5 received by the receiving unit 71, the receiving unit 71 passes the cell 5 to the transmitting unit 72 as it is.
To the outgoing link 32.

Here, the delay time required for data transfer in the ring type cell network according to the first embodiment is as follows. That is, the bandwidth w to be allocated to each flow can be calculated by the following equation [6] when the cell allocation period is L, and therefore the queuing delay is the following equation [7].

W = 1 / L = n / N [cell / cell time] ... [6] B / w = B × L [cell time] ... [7]

The adjustment delay is L [cell time],
Since it is 1 frame time or less in the case of the conventional ring network, the worst delay is a value represented by the following expression [8].

(B + 1) × L + Dfix [cell time] ... [8]

Further, the cell allocation period L for the flow in which the data of B [cell] must be transferred within the requested worst delay Dreq [cell time] is as shown in the following expression [9].

L: (Dreq −Dfix) / (B + 1) [cell time] ... [9]

The allocated bandwidth w at this time is calculated by the following equation [10].
become that way.

W: (B + 1) / (Dreq −Dfix) [cell / cell time] ... [10]

Allocation bandwidth w shown in equation [10]:
(B + 1) / (Dreq3-Dfix) is the conventional allocated bandwidth w shown in equation [5]: B / (Dreq-Dfix-N)
The following condition is given by the following expression [11].

Expression [5] −Expression [10] = {B / (Dreq −Dfix −N)} − {(B + 1) / (Dreq −Dfix)} ≧ 0 (B + 1) / (Dreq −Dfix) ≧ 1 / N ... [11]

This expression [11] indicates that the allocation band expressed by the expression [10] is one cell or more in one frame. In that case, the allocation band w in the first embodiment is
Indicates that less is required. Here, even in the conventional ring cell network, it is impossible to allocate less than one cell to one frame, so in any case, the allocated bandwidth w is smaller in the case of the first embodiment in any case. It will be.

Here, even in the conventional ring-type cell network, it is impossible to allocate less than one cell to one frame. Therefore, in any case, the allocated bandwidth w is smaller in the first embodiment. It will be few.

Embodiment 2 The second embodiment is basically the same as the first embodiment, but is different in that the control is performed not for each flow but for each class. Here, a class is a group of a plurality of flows that require the same worst delay.

Master node 2 in the second embodiment
The configuration of 11 is the same as that of FIG. 1 shown in the first embodiment. However, the round trip delay adjustment buffer 651 in the master node 211 of the second embodiment is provided with a transmission queue having a capacity of one cell for each class, not for each flow. Is different from that.

The configuration of the slave node according to the second embodiment is similar to that of the first embodiment as shown in FIG.
It has the same configuration as the slave node 22 in the conventional ring cell network shown in FIG.

FIG. 3 is an explanatory diagram showing a cell format used in the second embodiment. In the figure, 5a is a cell, 51 is a header of the cell 5a, 52 is also a payload, and 511 is a class identifier provided in the header 51 for identifying the class.
A flow identifier 12 is also provided in the header 51 for identifying a flow. This header 51 has a class identifier 511 in addition to the flow identifier 512, and is different from the header 51 of the cell 5 used in the conventional ring cell network shown in FIG. 19 and the ring cell network of the first embodiment. Is different.

FIG. 4 is an explanatory diagram showing an example of cell allocation to each class in the second embodiment. In the figure, 42 is a frame generated by the frame generation unit 641 and 5a is a cell. In the illustrated example, the frame length is 8
It has a fixed length of cells, 4 cells per frame for the first class (class (1)), 2 cells for the second class (class (2)), and 2 cells for the second class (class (3). ))
One cell is assigned to each of the fourth class (class (4)). Here, the cells 5a assigned to each class are evenly arranged in the frame 42 for each class. In this case, the class (1)
The cell 5a is assigned to the class 5 with a period of 2 cells, the class (2) with a period of 4 cells, and the classes (3) and (4) with a period of 8 cells. Note that these are the cell allocation periods for each class.

Master node 2 in the second embodiment
The operation of 11 is basically the same as that of the first embodiment, but the class identifier 51 is used instead of the flow identifier 512.
1 is set / referenced and the flow identifier 512 is not touched.
That is, the frame generation unit 641 generates the generated frame 41
Only the class identifier 511 is set for each cell 5a. Circular delay adjustment buffer 651 in receiving unit 611
To the loop delay adjustment buffer 6 in the transmission unit 621
Reading from 51 is performed in units of classes with reference to the class identifier 511. By operating in this manner, the round delay adjustment buffer 6 of the cell 5a of each class is
The maximum delay at 51 is equal to each cell allocation period.

Next, the slave node 22 according to the second embodiment operates as follows according to the attribute of the received cell 5a. That is, the reception unit 71 of the slave node 22
When the cell 5a having the flow identifier 512 of the flow originating from the own node is received from the incoming link 31, the receiving unit 71
Releases the flow identifier 512 of the cell 5a to empty the cell content, and then passes the cell content to the transmission unit 72. Further, when the cell 5a having the flow identifier 512 of the flow addressed to the own node is received from the incoming link 31, the receiving unit 71 makes the cell 5a
The contents of a are copied and then passed directly to the transmission unit 72. When the other cell 5a is received from the incoming link 31, the receiving unit 71 passes the received cell 5a to the transmitting unit 72 without doing anything.

On the other hand, the transmitter 72 of the slave node 22
Refers to the flow identifier 512 of the received cell 5a and determines whether or not the cell 5a is empty. If the cell 5a is empty, the communication data is fetched from the transmission queue 73 of the flow belonging to the class indicated by the class identifier 511 and inserted into the payload 52, and the flow identifier 5
After setting a value indicating the flow to 12 in the cell 5a
To the outgoing link 32. Which applicable send queue 7
If there is no communication data in the cell 3, the cell 5a is transmitted to the outgoing link 32 in an empty state. If the cell 5a is not empty, it is transmitted to the outgoing link 32 as it is.

Next, the delay time required for data transfer in the ring type cell network according to the second embodiment will be considered. Here, for example, it is assumed that the # 1 and # 2 slave nodes 22 forming the ring-type cell network generate the # 1 flow and the # 2 flow, respectively. ], The flow of # 2 generates data of B2 [cell] each time, and the request delays are the same.

For a while, for the sake of simplifying the explanation,
Focus only on queuing delays. Here, it is assumed that the queuing delays allowed for the # 1 flow and the # 2 flow are both Dq [cell time]. If bandwidth is allocated for each flow, w1 = B1 /
The band of Dq will be allocated to the flow of # 2 with the band of w2 = B2 / Dq.

Now, let us consider a case where the flow of # 1 and the flow of # 2 are in the same class, and the bands of w1 + w2 are allotted to this class. Now, consider a worst case in which both data are generated at the same time, the # 1 slave node 22 transmits all the data first, waits for the completion thereof, and then the # 2 slave node 22 starts transmission. At this time, the queuing delay of the # 2 flow that is kept waiting is the time during which the # 1 slave node 22 is transmitting the data of the # 1 flow, that is, B1 / (w1 +).
w2) and the time during which the slave node 22 of # 2 is transmitting data of the flow of # 2, that is, B2 / (w1 + w2)
). Therefore, the worst queuing delay is given by the following equation [12], which is the same as when bandwidth is allocated for each flow.

{B1 / (w1 + w2)} + {B2 / (w1 + w2)} = Dq ... [12]

Generalizing this discussion, there are m flows #i (i = 1, 2, ..., M) in a class whose allowed queuing delay is Dq [cell time], and When the amount of generated data is Bi [cell], the bandwidth w to be assigned to this class is the following expression [13].

W: Σ ^m Bi / Dq ... [13]

Next, let us return to the discussion including adjustment delay and propagation delay. The above-mentioned worst delay of the class to which the m flows belong is given by the following formula [14] when the cell allocation period is L [cell time].

Σ ^m Bi / w + L + Dfix = (Σ ^m Bi +1) × L + Dfix [cell time] ... [14]

Therefore, the required worst delay Dreq [cell time]
The cell allocation period L for the class to which m flows to which the data of Bi [cell] (i = 1, 2, ..., M) must be transferred is as shown in the following formula [15]. Is.

L: (Dreq −Dfix) / (Σ ^m Bi +1) [cell time] ... [15]

The allocated bandwidth w at this time is calculated by the following equation [1
6].

W: (Σ ^m Bi +1) / (Dreq −Dfix) [cell / cell time] ... [16]

Here, the first embodiment in which each flow is controlled
In the ring network system, the total bandwidth allocated to these m flows is given by the following equation [17], and it can be understood that the effective use of bandwidth further progresses in this second embodiment.

Σ ^m wi = Σ ^m {(Bi +1) / (Dreq −Dfix)} = (Σ ^m Bi + m) / (Dreq −Dfix) ... [17]

Since the transmission queue of the round trip delay adjustment buffer 651 of the master node 211 has a capacity of one cell for each class, the total capacity can be reduced by limiting the number of classes. Further, since the number of transmission queues is limited, the configuration of the master node becomes simple.

Third Embodiment 5 is a block diagram showing a configuration of a master node in a ring type network system according to a third embodiment of the present invention. In the figure,
Reference numeral 31 is an incoming link, 32 is an outgoing link, 613 is a receiving unit that receives the cell 5 and a reset cell 53 described later from the incoming link 31, 623 is a transmitting unit that transmits the cell 5 and the reset cell 53 to the outgoing link 32, 643 is a frame generation unit for continuously generating fixed-length frames 43, and 65.
3 is a buffer having a capacity of one cell. 213 is a receiving unit 613, a transmitting unit 623, and a frame generating unit 64.
3 and the buffer 653, and does not include the round trip delay adjustment buffer 651 in the first and second embodiments.

FIG. 6 is an explanatory diagram showing the frame structure of the frame 43 used in the third embodiment. In the figure, 43 is a frame having a fixed length of N cells generated by the frame generation unit 641, 5 is the cell, 51 is the header of this cell 5, 52 is the same payload, 512 is 512 in the header 51. This is a flow identifier. Reference numeral 53 is a special cell called a reset cell, which is added to a predetermined position (the head in this case) of the frame 43. Reference numeral 531 is a header of the reset cell 53, which is 5
32 is the payload of the reset cell 53. A value indicating that the cell is the reset cell 53 is set in the header 531 of the reset cell 53, whereby the reset cell 53 is distinguished from other normal cells 5. In addition, the payload 532 of the reset cell 53,
A list of the number of transmission-permitted cells assigned to each flow is shown. The number of cells permitted to be transmitted is the number of cells that each flow can use for transmission during one frame from the reset cell 53 to the next reset cell 53.

FIG. 7 is a block diagram showing the configuration of the slave node according to the third embodiment. In the figure, 31 is an incoming link, 32 is an outgoing link, 713 is a receiving unit that receives the cell 5 and the reset cell 53 from the incoming link 31, and 723 is a transmitting unit that sends the cell 5 and the reset cell 53 to the outgoing link 32. , 733 is a transmission queue provided for each flow and storing each communication data,
743 is associated with each transmission queue 733, and the transmission queue 73
Transmission cell counter for counting the number of cells transmitted from 3, 7
53 detects the reset cell 53 and detects the transmission cell counter 7
43 is a reset cell monitor for controlling 43. 223 is a receiving unit 713, a transmitting unit 723, a transmission queue 733,
Transmission cell counter 743 and reset cell monitor 7
It is a slave node provided with 53.

Next, the operation will be described. Here, first, the operation of the master node 213 will be described. The master node 213 uses the frame generation unit 643 shown in FIG.
The frame 43 having a fixed length of N cells shown in (4) is periodically generated. At that time, the frame generation unit 643 sets the contents of the header 531 and the payload 532 of the reset cell 53, but the header 51 and the payload 52 of other normal cells 5 are left empty, and the timing of the boundary of the cell 5 is set. Only occurs. Since the buffer 653 is empty at the time of network initialization, the transmission unit 623 leaves the header 51 and payload 52 of each cell 5 empty, and
3 is transmitted to the outgoing link 32.

The frame 43 transmitted to the outgoing link 32 in this way has the reset cell 53 at the head and the ring 1
It goes around and returns to the incoming link 31. The reception unit 613 discards the reset cell 53 of the received frame 43, and copies the contents of the header 51 and the payload 52 of the subsequent normal cell 5 to the buffer 653 in order. When the contents of the cell 5 are stored in the buffer 653, the transmission unit 623 reads both the contents of the header 51 and the payload 52 of the cell 5 from the buffer 653, and outputs the contents as they are to the boundary provided by the frame generation unit 643. It is inserted into the header 51 and the payload 52 of the corresponding cell 5 according to the timing and transmitted to the outgoing link 32.

Here, when the frame generation unit 643 generates the reset cell 53, the transmission unit 623 transmits the reset cell 53 to the outgoing link 32 and makes the cell 5 of the buffer 653 wait instead. At that time, the queue length of the buffer 653 increases by one cell, but when the reset cell 53 is received by the receiving unit 613 next, it is discarded and contracts by one cell, so there is no problem.

In the conventional ring type network system, the flow in which the cell 5 can be used is determined by the header 5 of each cell 5.
However, the third embodiment is different in that only the number of transmission-permitted cells is designated in the reset cell 53.

Next, the operation of the slave node 223 will be described. Here, the operation of the receiving unit 713 of the slave node 223 is as follows. Upon receiving the reset cell 53 from the incoming link 31, the receiving unit 713 passes the reset cell 53 as it is to the transmitting unit 723 and also to the reset cell monitor 753. Reset cell monitor 7
When the reset cell 53 is passed from the receiving unit 713, the 53 searches for the number of transmission-permitted cells of the flow originated from the own node according to the list of the number of transmission-permitted cells described in the payload 532 of the reset cell 53, and the corresponding transmission cell The contents of the counter 743 are reset to the number of transmission-permitted cells.

When the cell 5 received by the receiving unit 713 from the incoming link 31 is the cell 5 having the flow identifier 512 of the flow originating from the own node, the receiving unit 713 releases the flow identifier 512 of the cell 5. By doing so, the payload 52 is emptied, and then it is passed to the transmitting unit 723. When the cell 5 received from the incoming link 31 is the cell 5 having the flow identifier 512 of the flow addressed to the own node,
The receiving unit 713 copies the content of the payload 52 of the cell 5 and then passes it as it is to the transmitting unit 723. When receiving the other cell 5 from the incoming link 31, the receiving unit 713 passes the cell 5 to the transmitting unit 723 as it is.

The operation of the transmitter 723 of the slave node 223 is as follows. When receiving the reset cell 53 from the receiving unit 713, the transmitting unit 723 transmits it to the outgoing link 32 as it is. Further, when the normal cell 5 is passed, the transmitting unit 723 refers to the flow identifier 512 and determines whether the payload 52 of the cell 5 is empty.

When the payload 52 is empty, the communication data is read from any of the transmission queues 733 and inserted into the payload 52, and the flow identifier 512 is set to a value indicating the flow and transmitted to the outgoing link 32. To do.
At this time, the transmission cell counter 743 associated with the transmission queue 733 is decremented. When the transmission cell counter 743 becomes zero, the communication data stored in the transmission queue 733 corresponding to the transmission cell counter 743 is not transmitted thereafter. Corresponding transmission cell counter 74
When there is no communication data in any of the transmission queues 733 in which 3 is greater than zero, the transmission unit 723 transmits the cell 5 to the outgoing link 32 while leaving the payload 52 empty.

The cell 5 received from the receiving unit 713
If the payload 52 of No. 1 is not empty, the transmission unit 723 transmits the cell 5 as it is to the outgoing link 32.

Next, the delay time required for data transfer in the ring type cell network according to the third embodiment will be considered. Here, in the third embodiment, the number of cells that can be transmitted per frame is promised for each flow, but it is uncertain which cell position in the frame is actually usable. Therefore, in the worst case, an extra delay of one frame is added to the queuing delay.

FIG. 8 is a timing chart for explaining this phenomenon. For the flow (2) shown in the payload 532 of the reset cell 53 of FIG.
This is an example of a case where communication data for cells is generated. In this example, it is assumed that data is generated immediately after the transmission opportunity at the beginning of the first frame is missed, and the transmission opportunity is given at the end of the frame in the last frame.

As described above, the queuing delay in the third embodiment is equal to or less than the following expression [18] when the number of cells permitted to be transmitted is n [cell].

B / (n / N) + N = (B / n + 1) × N [cell time] ... [18]

Since the third embodiment does not have the loop delay adjustment buffer, the adjustment delay is zero. Therefore,
The worst delay shown in the following equation [19] can be guaranteed.

(B / n + 1) × N + Dfix [cell time] ... [19]

That is, according to the third embodiment, it is possible to obtain the same performance as that of the conventional ring network system while omitting the round trip delay adjustment buffer of the master node 213.

Embodiment 4 Here, the third embodiment
Then, the number of transmission permitted cells of each flow is set to the master node 213.
However, in the fourth embodiment, the number of transmission-permitted cells is set to the slave node 223.
, And the master node 213 supplies only the reset timing.

FIG. 9 is an explanatory diagram showing the data structure of a cell used in the ring network system according to the fourth embodiment of the present invention. In the figure, 5 is a cell,
Reference numeral 51 is its header, and 52 is its payload. Also, 5
12 is a flow identifier in the header 51, and 513
Is a reset indicator in the header 51. In addition, in the frame used in this Embodiment 4, Embodiment 3
The reset cell 53 used in 1 is not used.

FIG. 10 is a block diagram showing the structure of the master node according to the fourth embodiment. In the figure, 3
1 is an incoming link, 32 is an outgoing link, 614 is a receiving unit that receives the cell 5 from the incoming link 31, 624 is a transmitting unit that sends the cell 5 to the outgoing link 32, and 644 is the timing and frame of the boundary of the cell 5. It is a reset generation unit that generates reset timing in a cycle. Reference numeral 214 denotes a master node including the reception unit 614, the transmission unit 624, and the reset generation unit 644, and the round delay adjustment buffer 651 in the first and second embodiments or the third embodiment.
The buffer 653 is not provided.

FIG. 11 is a block diagram showing the structure of the slave node according to the fourth embodiment. In the figure,
31 is an incoming link, 32 is an outgoing link, 713 is a receiving unit, 7
23 is a transmission unit, 733 is a transmission queue, 743 is a transmission cell counter, 754 is a reset monitor that detects the cell 5 in which the reset indicator 513 in the header 51 is on and controls the transmission cell counter 743, and 764 is It is a transmission permission cell number register which is associated with each transmission cell counter 743 and stores the number of transmission permission cells for the corresponding flow. Reference numeral 224 denotes a slave node including the receiving unit 713, the transmitting unit 723, the transmission queue 733, the transmission cell counter 743, the reset monitor 754, and the transmission permission cell number register 764, and the reset monitor 754 replaces the reset cell monitor 753. , Each cell counter 7
The slave node 223 according to the third embodiment is different from the slave node 223 in that the transmission permission cell number register 764 is associated with each of the 43.
Is different from.

Next, the operation will be described. Here, first, the operation of the master node 214 will be described. The reset generation unit 644 of the master node 214 generates the timing of the boundary of the cell 5 and further generates the reset timing in the frame cycle. At the time of network initialization, since there is no cell 5 from the receiving unit 614, the transmitting unit 624 transmits each cell 5 to the outgoing link 32 while leaving the header 51 and the payload 52 empty. When the reset generation unit 644 gives a reset timing, the transmission unit 624 turns on the reset indicator 513 in the header 51 of the cell 5 to be transmitted only at that time.

On the other hand, the receiving unit 614 turns off the reset indicator 513 of the header 51 of all the cells 5 to be received, but otherwise, the header 51 and the payload 52 are passed as they are to the transmitting unit 624 in the order received. .
When the cell 5 is delivered from the receiver 614, the transmitter 624 receives the header 51 and the payload 5 of the cell 5.
Both 2 are left as they are, and are transmitted to the outgoing link 32 according to the boundary timing given from the reset generation unit 644. However, when the reset timing is given from the reset generation unit 644, only at that time, the transmission unit 624 turns on the reset indicator 513 of the cell 5 to be transmitted.

[0116] In a word, the master node 214
Only turns on the reset indicator 513 of the cell 5 every frame period.

Next, the operation of the slave node 224 will be described. The operation of this slave node 224 is also basically the same as that of the slave node 223 of the third embodiment shown in FIG. 7, and therefore the differences will be mainly described.

The reset monitor 754 monitors whether the reset indicator 513 of the cell 5 received by the receiver 713 is ON. When the reset indicator 513 detects the ON cell 5, the reset monitor 754 sends a reset signal to each transmission cell counter 743. Upon receiving this reset signal, each transmission cell counter 743 is set based on the value held by the associated transmission-permitted-cell number register 764. When the data of the transmission queue 733 is transmitted from the transmission unit 723, the transmission permission cell number register 764 associated with it is decremented, and when the content becomes zero, the communication data of the transmission queue 733 is not transmitted thereafter.

As described above, the difference from the third embodiment is that the number of transmission-permitted cells is instructed each time by the master node or held by the slave node. Therefore, in the fourth embodiment, a mechanism for the slave node 224 to grasp the number of cells permitted to be transmitted is separately required, but it can be easily realized by either of the following methods. 1) The system administrator manually sets the number of cells permitted to be transmitted to each slave node 224 in advance. 2) From the master node 214 to the slave node 224,
Each time communication is started, the number of cells permitted to be transmitted of the flow is notified prior to the communication.

Also, not the number of cells permitted to be transmitted itself,
Even in the method in which the allocated band is given to the slave node 224, the slave node 224 obtains the frame period by monitoring the reset indicator 513, and the slave node 224 calculates the number of cells permitted to be transmitted from the frame period and the allocated band. Good.

Since the discussion on the worst delay is exactly the same as in the case of the third embodiment, its explanation is omitted here.

As described above, according to the fourth embodiment as well, the round delay adjustment buffer of the master node is omitted,
The performance equivalent to that of the conventional ring network system can be obtained. Further, according to the fourth embodiment, there is also an advantage that it is not necessary to provide a special cell called a reset cell as in the third embodiment.

The technique applied to the slave nodes 223 and 224 in the third and fourth embodiments is a known technique applied to the Orwell Ring disclosed by British Telecom. For example, it is also shown in JP-A-5-227176.

Embodiment 5. FIG. Although each of the above embodiments relates to a ring type cell network in which one master node and one or more slave nodes are connected in a ring type, the fifth embodiment does not have a master node,
The present invention relates to a ring type cell network in which a plurality of nodes, which are all equal, are connected in a ring type. Therefore, the slave node corresponding to each of the above-mentioned embodiments will be simply referred to as a node. Further, although the fifth embodiment has the concept of a frame, there is no actual frame. In the fifth embodiment, the cell 5 including the header 51 having the flow identifier 512 and the payload 52 shown in FIG. 19 is used.

FIG. 12 is a block diagram showing the structure of a node in the ring network system according to the fifth embodiment of the present invention. In the figure, 31 is an incoming link from the ring, 32 is an outgoing link to the ring, 71 is a receiving unit for receiving the cell 5 from the incoming link 31, 73 is provided for each flow, and stores respective communication data. It is a transmission queue. Reference numeral 775 denotes a shaper that is associated with each transmission queue 73 and reads out the cell 5 from the transmission queue 73 at a constant cycle. Each shaper 775 has a virtual frame length (commonly defined in the ring network). N) and the number of cells permitted to be transmitted for each flow (n)
give. 785 is a global transmission queue that bundles the transmission cells from each shaper 775, and 725 is a transmission unit that transmits the cells 5 bundled by the global transmission queue 785 to the outgoing link 32. 23 is the receiving unit 7
1, transmission queue 73, transmission unit 725, shaper 775,
And a global transmission queue 785, the transmission unit 72 of the conventional slave node 22 shown in FIG. 18 is replaced by the transmission unit 725, and a shaper 775 and a global transmission queue 785 are newly provided. .

Next, the operation will be described. This node 2
Each shaper 775 of No. 3 reads out the cell 5 from the associated transmission queue 73 at a constant cycle of N / n and puts it in the global transmission queue 785. In addition, a plurality of shapers 7
When the cells 5 from 75 collide with each other, the order of inserting the cells 5 into the global transmission queue 785 is performed according to a priority order that is appropriately determined in advance.

Here, the operation of the receiver 71 is the same as that of the conventional slave node 22 and is as follows. That is, if the cell 5 received from the incoming link 31 has the flow identifier 512 of the flow originating from its own node, the receiving unit 71 releases the flow identifier 512 of the cell 5 to empty the contents of the payload 52. , Send it 72
Pass to 5. If the received cell 5 is the cell 5 having the flow identifier 512 of the flow addressed to the own node, the receiving unit 7
After copying the content of the payload 52, the cell 1 directly passes the cell 5 to the transmitting unit 725. When the other cells 5 are received from the incoming link 31, the receiving unit 71 passes the received cells 5 to the transmitting unit 725 as they are.

The operation of transmitting section 725 is as follows. That is, the transmission unit 725 refers to the flow identifier 512 of the received cell 5 and checks the payload 5 of the cell 5.
Determine if 2 is empty. If the payload 52 is empty, the communication data is inserted from the global transmission queue 785 into the payload 52 of the cell 5, and the flow identifier 51
After setting the value indicating the flow to 2, the cell 5 is transmitted to the outgoing link 32. If there is no communication data in the global transmission queue 785 at that time, the cell 5 is transmitted to the outgoing link 32 with the payload 52 left empty. on the other hand,
If the payload 52 is not empty, the transmitter 725 transmits the cell 5 as it is to the outgoing link 32.

Next, the delay time required for data transfer in the ring type cell network according to the present invention will be considered. In the fifth embodiment, as in the third and fourth embodiments, the number of cells that can be transmitted per frame time is promised for each flow, but the cell at which position in the frame is actually used. Whether it is possible is uncertain. Therefore, in the worst case, an extra delay of one frame is added to the queuing delay. This point will be described in more detail below.

First, the queuing delay in the transmission queue 73 will be described. The read cycle of the shaper 775 is N / n [cell time], and the allocated bandwidth is n / N.
[Cell / cell time]. Therefore, if the communication data size is B [cell], the queuing delay in the transmission queue 73 is the following expression [20] or less.

B / (n / N) [cell time] ... [20]

Next, the queuing delay in the global transmission queue 785 will be described. As an example, consider a network as shown in FIG. In FIG. 13, 23 is a node, and 3 is a link connecting the nodes 23 and having the same transfer rate. Here, the frame length N is set to 8 cells, 4 is given to the flow (1) transmitted from the node 23 of # 1 as the number n of transmission permitted cells, and 2 is given to the flow (2). 1 is given to the flow (3) that is transmitted from, and 1 is given to the flow (4). Now that all bands have been allocated, other nodes are not allowed to transmit.

FIG. 14 shows the case where the phases of the operations of the shapers 775 happen to match in each node 23 of the network shown in FIG.
6 is an explanatory diagram showing the movement of the cell 5 in FIG. In FIG. 14, the shaper 775 that transmits the flow (1) of the # 1 node 23 is the shaper (1), and the shaper 775 that transmits the flow (2) is the shaper (2), #.
Shaper 77 for transmitting the flow (3) of the node 23 of 2
5 is referred to as a shaper (3), and the shaper 775 for transmitting the flow (4) is referred to as a shaper (4). In the node 23 of # 1, for example, the transmission of the flow (1) is prioritized, and when the outputs of the two shapers (1) and (2) collide, the cell 5 of the flow (2) becomes the global The transmission queue 785 is kept waiting. In the node 23 of # 2, the receiving cell 5 from the node 23 of # 1 is transmitted with the highest priority, and then the transmission of the flow (3) is prioritized, in accordance with the node operation regulation. As a result, the cell 5 of the flow (4) will be transmitted one frame after being output from the shaper (4).

Although a very simple example is shown, the queuing delay in the global transmission queue 785 is generally one frame time or less if the following three conditions are satisfied. 1) All flows are cell-transmitted via the shaper 775, that is, cell-transmitted at an equal cycle according to the allocated band. 2) The transmission cycle of any flow is less than or equal to the frame length N,
That is, the minimum unit of allocated bandwidth is 1 / N. 3) The total of all flows with the number n of transmission-permitted cells is less than or equal to the frame length N.

This is because, if there is a flow in which it is not possible to transmit cells of the number n of transmission-permitted cells in any one frame time, the untransmitted cells will be transmitted in the next one frame time. However, as long as there is no such shift of untransmitted cells, as many cells as the number n of transmission-permitted cells can be transmitted in one frame time.

As described above, the queuing delay in the fifth embodiment is equal to or less than the expression [21] shown below.

B / (n / N) + N = (B / n + 1) × N [cell time] ... [21]

Here, in the fifth embodiment, since there is no round delay adjustment buffer, the adjustment delay is zero. Therefore, the worst delay shown in the following formula [22] can be guaranteed.

(B / n + 1) × N + Dfix [cell time] ... [22]

In other words, according to the fifth embodiment, it is possible to obtain the same performance as that of the conventional ring type network system, without using a special master node and using only all the equal nodes.

[0141]

As described above, according to the invention described in claim 1, in the master node, the content of the payload of the received cell is copied to the one corresponding to the flow in the transmission queue of the round trip delay adjustment buffer. Then, it determines which flow each cell in the frame is assigned to, inserts the communication data stored in the corresponding transmission queue of the round trip delay adjustment buffer into the payload of that cell, and assigns it to each flow. The cell is
Since it is configured to be evenly arranged in the frame, the cells allocated to a certain flow occur at the cycle of the frame length divided by the number of cells allocated to that flow, and the master node adjusts the cells in that flow for the round trip delay adjustment. After being received in the buffer, it is possible to transmit within a time period equal to or shorter than the period, and there is an effect that an access control method for the ring network in which the bandwidth allocated to each flow is small can be obtained.

According to the invention described in claim 2, the flow is classified according to the required worst delay time, and the content of the payload of the received cell is stored in the transmission queue of the round delay adjustment buffer in the master node. Copy to the one that corresponds to the class to which the flow belongs, determine which class each cell in the frame is assigned to,
The communication data stored in the corresponding transmission queue of the round trip delay adjustment buffer is inserted into the payload of the cell, and the cells to be assigned to each class are configured to be evenly arranged in the frame for each class.
The cells allocated to a class occur at a cycle that is equal to the frame length divided by the number of cells allocated to that class, and the master node receives the cells of that class in the round trip delay adjustment buffer within the time period less than that cycle. Access method of ring network that can be sent to
If the number of classes is limited, the total capacity of the round trip delay adjustment buffer can be reduced, and the number of transmission queues is also limited, which has the effect of simplifying the configuration of the master node.

According to the third aspect of the present invention, the master node issues a reset instruction at a constant cycle, and the slave node permits transmission with a predetermined number of transmission cells for each transmission queue storing communication data. After the number of cells is reached, the transmission from the transmission queue is stopped, and when the reset instruction is received, the transmission cell number is reset, so that the master node issues a reset instruction.
During the cycle, the number of cells transmitted from the transmission queue of each slave node becomes equal to or less than a predetermined number of cells to be permitted for transmission, and the cells are not transmitted beyond this number. Therefore, the master node issues a reset instruction. During the cycle, each slave node can send as many cells as the number of cells permitted to send, and it is equivalent to the access control method of the conventional ring network without providing the loop delay adjustment buffer in the master node. There is an effect that the performance of can be obtained.

According to the fourth aspect of the present invention, the master node generates the number of transmission-permitted cells of each flow together with the reset instruction. Therefore, the number of transmission-permitted cells of each flow is held in the slave node. There is no need to store the information, and the configuration of the slave node can be simplified.

According to the fifth aspect of the present invention, the number of transmission-permitted cells of each flow is held in the slave node. Therefore, a special cell called a reset cell is arranged at a predetermined position in the frame. There is an effect that the processing of the master node can be simplified because there is no need to do so.

According to the sixth aspect of the present invention, the ring network is formed by all nodes which are equal to each other, and the communication data stored in each transmission queue by these nodes is shared by the frames common to the entire ring network. Since it is configured to transmit at a fixed period divided by the number of cells permitted to be transmitted for each transmission queue, it is not necessary to provide a special master node, and the cells transmitted from each transmission queue are set to the frame period. Evenly distributed, the individual transmission cycle from each transmission queue does not exceed the frame cycle even at the maximum. Therefore, as long as the total number of cells allowed to be transmitted is less than or equal to the frame cycle, when a certain frame cycle is taken in an arbitrary phase. , The total number of transmission cells from the transmission queues contained in it does not exceed the frame period, and during each arbitrary frame period, The effect of access control method of ring network can be over de always sends the cell to the number of transmission permit cell is obtained.

According to the seventh aspect of the invention, the frame generating section continuously generates a frame consisting of a fixed number of cells, and the cells are allocated so that each flow is even in the frame, and the receiving section Copy the cell received in the above into the transmission queue corresponding to the flow of the cell in the round trip delay adjustment buffer, and in the transmitter, determine which flow the cell of the frame generated by the frame generator is assigned. , The master node that inserts the communication data of the corresponding transmission queue of the round trip delay adjustment buffer into its payload and sends it, and if the cell received by the receiving unit is assigned to the flow originating from its own node, the payload Empty, copy the contents of the payload if it is assigned to the flow addressed to the own node, and if it is another cell, leave that cell as it is. , Each of which has received the cell of the flow originating from its own node whose payload has been emptied, the transmitting section reads communication data from the corresponding one of the transmission queues provided for each flow and Since one or more slave nodes that are inserted in the payload and transmitted are configured to be connected in a ring by a link,
The cells allocated to a certain flow occur in a cycle that is the frame length divided by the number of cells allocated to that flow, and the master node receives the cells of that flow in the round trip delay adjustment buffer, and then Therefore, it is possible to obtain a ring network system in which the bandwidth allocated to each flow can be reduced.

According to the eighth aspect of the present invention, the frame generating section continuously generates frames each consisting of a fixed number of cells, and each flow is classified by the worst delay time required. Allocate the cells so that they are even within each other, copy the cells received by the reception unit to the transmission queue corresponding to the class of the cell in the round trip delay adjustment buffer, and the frames generated by the frame generation unit at the transmission unit. The master node that determines which class the cell of is assigned to and inserts the communication data of the corresponding transmission queue of the round trip delay adjustment buffer into its payload, and the cell received by the receiving unit, If used for a flow originating from the own node, empty the payload, and if used for a flow addressed to the own node, the contents of that payload If the cell is copied and is another cell, that cell is passed to the transmitter as it is, and the transmitter that received the cell whose payload is empty is the transmission queue corresponding to the flow that belongs to the class to which the cell is assigned. More than one slave node that reads communication data, inserts it into the payload of that cell, and transmits it is configured so that it is connected in a ring by a link. Occurs in the cycle divided by the number of cells allocated to the master node, the master node receives the cells of that class in the round trip delay adjustment buffer, and then a ring network system that can transmit within the time less than the cycle is obtained. By limiting the number of classes, the total capacity of the round trip delay adjustment buffer can be reduced, and the number of transmission queues is also limited. There is an effect that the configuration of the soil becomes simple.

According to the ninth aspect of the present invention, the frame generation section continuously generates frames consisting of a fixed number of cells, and a reset cell including a reset instruction and the number of cells permitted to be transmitted for each flow is set as a reset cell of the frame. When the cell other than the reset cell is received at the predetermined position and the receiving section receives it,
A master node that sequentially copies the contents of the buffer to the buffer with the capacity of cells and sends the contents of the buffer from the transmitter according to the boundary timing given by the frame generator, and the cell received by the receiver If it is used for the above, the payload is emptied, if it is used for the flow destined for the own node, the content of the payload is copied. The transmission unit that receives the cell whose payload is empty and reads the communication data from the transmission queue and waits until the transmission cell counter attached to the corresponding transmission queue counts the number of cells to be transmitted. It is inserted in the payload of the cell and transmitted, and the reset of the transmission cell counter is Since one or more slave nodes based on the number of permitted cells are connected in a ring by a link, the master node does not have a loop delay adjustment buffer and is equivalent to a conventional ring network system. Since the performance can be obtained and it is not necessary to hold the number of cells permitted to be transmitted for each flow in the slave node, the configuration of the slave node can be simplified.

According to the tenth aspect of the invention, the reset generation unit generates the timing of the cell boundary and the reset timing in the frame period, and when the reception unit receives the cell, the reset indicator is turned off and transmission is performed. When a reset timing is generated by the reset generation unit, the transmission unit transmits the master node that turns on the reset indicator of the cell received from the reception unit and transmits, and the number of cells to which transmission is permitted in the corresponding transmission queue. If the cell received in the receiving unit is used for the flow originating from the own node, the payload is emptied and is used for the flow addressed to the own node. Copy the contents of the payload, and if there are other cells, pass the cells as they are to the transmitter, and leave the payload empty. The transmission unit that receives a cell reads the communication data from the transmission queue and inserts it into the payload of that cell only until the transmission cell counter attached to the corresponding transmission queue counts the number of cells to be transmitted. Link one or more slave nodes that perform transmission and reset the transmission cell counter based on the number of transmission-permitted cells held in the transmission-permission-cell number register when a cell whose reset indicator is on is received. Since it is configured so that the master node does not have a loop delay adjustment buffer, it is possible to obtain the same performance as a conventional ring network system and to reset cells at a predetermined position in the frame. Since there is no need to arrange a special cell such as, there is an effect that the processing of the master node becomes easy.

According to the eleventh aspect of the invention, if the cell received by the receiving unit is one used for the flow originating from the own node, the payload is emptied and used for the flow addressed to the own node. If it is, the content of the payload is copied, and if it is another cell, it is passed to the transmitter as it is, and the transmitter that received the cell in which this payload is emptied is commonly set for the entire ring network. The specified frame period is divided by the number of cells permitted to be transmitted for each transmission queue, and the communication data read by the shaper from the corresponding transmission queue and bundled by the global transmission queue is sent to the payload of that cell at fixed periods. Since it is configured to connect multiple nodes that are equivalent to each other, which are inserted and transmitted, in a ring shape with links, a special master node Kicking it is not necessary, during any frame period, capable ring network system that each node transmits always cells to the number of transmission permitted cell the effect obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a master node in a ring network system according to first and second embodiments of the present invention.

FIG. 2 is an explanatory diagram showing an example of cell allocation to each flow according to the first embodiment of the present invention.

FIG. 3 is an explanatory diagram showing a cell format used in the second embodiment of the present invention.

FIG. 4 is an explanatory diagram showing an example of cell allocation to each class according to the second embodiment of the present invention.

FIG. 5 is a block diagram showing a configuration of a master node in a ring network system according to a third embodiment of the present invention.

FIG. 6 is an explanatory diagram showing a frame configuration of a frame used in Embodiment 3 of the present invention.

FIG. 7 is a block diagram showing a configuration of a slave node according to the third embodiment of the present invention.

FIG. 8 is a timing diagram for explaining queuing delay according to the third embodiment of the present invention.

FIG. 9 is an explanatory diagram showing a cell format used in the fourth embodiment of the present invention.

FIG. 10 is a block diagram showing a configuration of a master node according to a fourth embodiment of the present invention.

FIG. 11 is a block diagram showing a configuration of a slave node according to a fourth embodiment of the present invention.

FIG. 12 is a block diagram showing a configuration of a node in a ring network system according to a fifth embodiment of the present invention.

FIG. 13 is a network configuration diagram for explaining queuing delay according to the fifth embodiment of the present invention.

FIG. 14 is a timing diagram showing movement of cells in each node according to the fifth embodiment of the present invention.

FIG. 15 is a conceptual diagram for explaining a conventional TDMA-type ring network.

FIG. 16 is an explanatory diagram showing a format of a fixed-length frame generated by a conventional master node.

FIG. 17 is a block diagram showing a configuration of a conventional master node.

FIG. 18 is a block diagram showing a configuration of a conventional slave node.

FIG. 19 is an explanatory diagram showing a frame format used in a conventional ring-type cell network.

[Explanation of symbols]

3 links, 5,5a cells, 22,223,224
Slave node, 23 node, 31 incoming link (link), 32 outgoing link (link), 41-43 frame, 52 payload, 53 reset cell, 71, 6
11, 613, 614, 713 Receiver, 72, 62
1, 623, 624, 723, 725, transmitting section, 73,
733 transmission queue, 211, 213, 214 master node, 511 class identifier, 512 flow identifier, 513 reset indicator, 641, 643 frame generation unit, 644 reset generation unit, 651 round delay adjustment buffer, 743 transmission cell counter, 753 reset Cell monitor, 754 reset monitor, 764 transmission permission cell number register, 775 shaper, 785 global transmission queue.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yukio Ushisako 2-3-3 Marunouchi, Chiyoda-ku, Tokyo Sanryo Electric Co., Ltd. (72) Inventor Kazuyuki Shima 2-chome, Marunouchi, Chiyoda-ku, Tokyo No. 3 Sanryo Electric Co., Ltd. (72) Inventor Tetsuya Yokotani 2-3-3 Marunouchi, Chiyoda-ku, Tokyo Sanryo Electric Co., Ltd. (72) Inventor Keiichi Soda 2-3-2 Marunouchi, Chiyoda-ku, Tokyo Sanryo Electric Co., Ltd. (72) Inventor Ken Murakami 2-3-3 Marunouchi, Chiyoda-ku, Tokyo Sanryo Electric Co., Ltd. (72) Katsuka Takahashi 2-3-2 Marunouchi, Chiyoda-ku, Tokyo Sanryo Electric Co., Ltd.

Claims (11)

[Claims] 1. A master node and one or more slave nodes are connected in a ring by a link to form a ring network for transferring communication data of each flow with a fixed length cell as a transfer unit, In each cell transmission to the link, in the access control method of the ring network in which the cell relay from upstream to downstream has the highest priority, and the master node continuously generates a frame composed of a certain number of cells, The master node copies the contents of the payload of the received cell to the transmission queue corresponding to the flow of the round trip delay adjustment buffer divided into transmission queues for each flow, and assigns to each flow each cell in the frame. Communication data read from the transmission queue of the corresponding flow of the round trip delay adjustment buffer Is inserted into the payload of the cell, and the cells to be assigned to each flow are evenly arranged in the frame for each flow. 2. A ring-type network is formed in which one master node and one or more slave nodes are connected in a ring by a link, and a fixed-length cell is used as a transfer unit to transfer communication data of each flow, In each cell transmission to the link, in the access control method of the ring network in which the cell relay from upstream to downstream has the highest priority, and the master node continuously generates a frame composed of a certain number of cells, The flow is classified according to the required worst delay time, and the master node copies the content of the payload of the received cell to the transmission queue corresponding to the class of the round trip delay adjustment buffer divided into transmission queues for each class. , Determining which class each cell in the frame is assigned to, and adjusting the round trip delay A ring type network characterized in that the communication data read from the transmission queue of the corresponding class of the buffer is inserted into the payload of the cell, and the cells assigned to each class are evenly arranged in the frame for each class. Access control method. 3. A ring type network is formed, in which one master node and one or more slave nodes are connected in a ring by a link, and a fixed length cell is used as a transfer unit to transfer communication data of each flow. Each node in the cell transmission to the link, in the access control method of the ring network that gives the highest priority to cell relay from upstream to downstream, the reset instruction is generated from the master node in a constant cycle, the slave node, Transmission from the transmission queue that stores communication data for each flow is stopped after the number of transmission cells in the transmission queue reaches a predetermined number of cells permitted to be transmitted, and the transmission cell is transmitted by the reset instruction received from the upstream. An access control method for a ring-type network, which comprises resetting the number. 4. The access control method for a ring network according to claim 3, wherein the master node generates the number of cells permitted to be transmitted for each flow together with a reset instruction. 5. The slave node internally holds the number of cells permitted to be transmitted for each flow.
An access control method for the ring network described. 6. A ring type network is formed in which a plurality of nodes are connected in a ring type by a link and a fixed length cell is used as a transfer unit to transfer communication data of each flow, and each node is a cell to the link. In transmission, in a ring network access control method that gives top priority to cell relay from upstream to downstream, each node transmits from a transmission queue storing communication data of each flow, the entire ring network. An access control method for a ring network, characterized in that a frame period that is defined in common with the above is divided by the number of transmission-permitted cells for each transmission queue. 7. A frame generation unit for continuously generating frames consisting of a fixed number of cells, and allocating each flow to each cell so as to be even within the frame, and a capacity for one cell for each flow. Circular delay adjustment buffer having a transmission queue having, a receiving unit that receives the frame sent from the upstream and copies the content of the payload of the cell to the corresponding transmission queue of the circular delay adjustment buffer, and the frame It is determined which flow the cell of the frame generated by the generation unit is assigned to, and the transmission unit that inserts the communication data of the corresponding transmission queue of the round trip delay adjustment buffer into the payload of that cell and transmits it downstream. A master node and a transmission queue prepared for each flow and storing each communication data, before being sent from upstream The frame is received, and if the cell is a cell assigned to the flow originating from its own node, the payload is emptied, and if it is a cell assigned to the flow addressed to its own node, the contents of the payload are copied. If a cell assigned to a flow originating from its own node is received from the receiving unit that passes it to the transmitting unit and the receiving unit from the receiving unit as it is, the communication data of the transmission queue corresponding to the flow is stored in the cell. A link that connects the master node and the slave node in a ring shape with one or more slave nodes that have a transmission unit that inserts them into an empty payload and receives other cells as they are, and transmits them as they are. A ring type network system equipped with. 8. A frame consisting of a fixed number of cells is consecutively generated, and each flow is classified according to the required worst delay time so that the class of each of the cells is even within the frame. A frame generation unit to be allocated, a round delay adjustment buffer having a transmission queue having a capacity of one cell for each class, the frame transmitted from an upstream side is received, and the contents of the payload of the cell are adjusted to the round delay. The receiving unit to be copied to the corresponding transmission queue of the buffer and the class of the cell of the frame generated by the frame generation unit are determined, and the communication data of the corresponding transmission queue of the round trip delay adjustment buffer is stored. A master node that has a transmitter that inserts it into the payload of the cell and transmits it downstream, and that is prepared for each flow A transmission queue that stores each communication data, receives the frame sent from the upstream, and if the cell is the cell used for the flow originating from the own node, empty the payload and set the flow to the own node. If it is a cell that is being used, copy the contents of the payload, and if it is another cell, leave it as it is and receive the cell that has an empty payload from the reception section and the reception section. Insert the communication data of the transmission queue corresponding to the flow belonging to the assigned class into the payload of the cell, set a value indicating the flow to the flow identifier, and when receiving other cells, leave it as it is. , One or more slave nodes having a transmission unit for transmitting downstream, and a link connecting the master node and the slave nodes in a ring shape Ring-type network system with a. 9. A frame generation unit for continuously generating a frame composed of a fixed number of cells, in which reset cells including a reset instruction and the number of transmission-permitted cells of each flow are arranged at a predetermined position, and one cell of the cells. A buffer having a capacity of 1 minute, a receiving unit that receives the frame sent from the upstream side and sequentially copies cells other than the reset cell to the buffer, and boundary timing given by the frame generating unit to the contents of the buffer. According to a master node having a transmission unit for transmitting downstream according to the above, a transmission queue prepared for each of the flows, storing each communication data, receiving the frame transmitted from the upstream, the cell is
If the cell used for the flow originated from the own node, empty the payload, if the cell used for the flow addressed to the own node, copy the contents of the payload, if it is another cell, leave it as it is, A receiving unit to be passed to the transmitting unit, prepared in association with each of the transmission queues, a transmission cell counter for counting the number of cells transmitted from the corresponding transmission queue, receiving a reset cell received by the receiving unit,
A reset cell monitor that resets the transmission cell counter based on the number of transmission-permitted cells of the reset cell, and when the payload is empty from the receiving unit, the transmission cell counter corresponding to the flow indicates the transmission-permitted cells. One or more including a transmission unit that inserts the communication data of the transmission queue into the payload of the cell and transmits the data downstream, and transmits the data as it is to the downstream when other cells are received only until the number is counted. Ring network system including a slave node and a link connecting the master node and the slave node in a ring shape. 10. A reset generation unit for generating reset timing at a cell boundary timing and a frame period,
When receiving a frame sent from the upstream side, turning off the reset indicator of each cell and passing it to the transmitting unit, and a reset timing generated from the reset generating unit, the cell received from the receiving unit A master node having a transmitter that turns on a reset indicator and transmits downstream,
Prepared for each flow, a transmission queue that stores each communication data, receives the frame sent from the upstream, and if the cell is the cell used for the flow originating from the own node, empty the payload. , If the cell is used for the flow addressed to the own node, copy the contents of the payload, and if it is another cell, pass it to the transmitter as it is, and prepare it for each of the reception unit and the transmission queue. A transmission cell counter that counts the number of cells transmitted from the corresponding transmission queue, and a transmission permission cell number register that is prepared in association with each of the transmission cell counters and that stores the number of transmission permission cells for the corresponding flow When the reset indicator of the cell received by the receiving unit is ON, the transmission cell counter stores the transmission permission cell stored in the transmission permission cell number register. When a cell whose payload is empty is received from the reset monitor that resets based on the number, and the transmission cell counter corresponding to the flow counts the number of transmission-permitted cells, the transmission is performed. One or more slave nodes having a transmission unit that inserts the communication data of the queue into the payload of the cell and transmits the data downstream, and when receiving other cells, directly transmits the data downstream, the master node and the slave node A ring-type network system that includes a link that connects to each other in a ring shape. 11. A transmission queue that is prepared for each flow and stores each communication data, and a frame cycle that is prepared in association with each of the transmission queues and that is commonly set for the entire ring network is transmitted. A shaper that reads cells from the corresponding transmission queue in a cycle divided by the number of cells permitted to be transmitted for each queue, a global transmission queue that bundles the cells that are read from the shaper at a constant cycle, and the global transmission queue that is sent from upstream Receive the frame,
If the cell is a cell used for the flow originating from the own node, the payload is emptied, and if the cell is used for the flow addressed to the own node, the contents of the payload are copied,
If the cell is another cell, the cell is received as it is, and if the cell payload received from the cell receiving section and the cell receiving section is empty, the communication data of the global transmission queue is inserted into the payload and received. A ring-type network system comprising a plurality of nodes each having a transmission unit for transmitting the payload of a cell as it is, if it is not empty, and a link connecting the nodes in a ring shape.