JP2004153850A - Communication system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an optical communication system capable communication having a small delay rate, a small discard rate and a high throughput by using a network which uses a wavelength division multiplexing and a time division multiplexing in combination. <P>SOLUTION: This system has a wavelength multiplexing optical network 1 having a plurality of optical transmission channels of different wavelengths, a plurality of nodes 2 interconnected via this wavelength multiplexing optical network and performing transmission/reception by using the time slots into which each of the transmission channels is divided, and a network controller 3 for centrally controlling the time slot allocation for the plurality of nodes. To ensure common reception idle time slots for all of the plurality of nodes, the network controller moves the assigned time slots when an idle time slot is newly generated or the network controller is capable of another processing. <P>COPYRIGHT: (C)2004,JPO

Description

  The present invention relates to a communication system, and more particularly to an optical communication system using a wavelength division multiplexing optical network.

  With the development of the information society, there is a demand for a backbone network capable of larger-capacity communication. As one method for increasing the capacity of a backbone network, there is wavelength division multiplexing in which a plurality of optical signals of different wavelengths, which have been independently data-modulated, are inserted into one optical fiber in an optical communication system. Several methods of configuring a backbone network using wavelength division multiplexing have been conventionally proposed (for example, Non-Patent Document 1).

  Among them, the report by Shimosaka et al. (OCS92-79) adopts a method in which an optical transmission channel of each wavelength is divided into a plurality of short time slots, and transmission data is grouped into packets for each destination and sent to an empty time slot. are doing. In other words, this is a method in which time division multiplexing is used together with wavelength division multiplexing. This method is effective in a system such as a backbone LAN where transmission requests to a large number of partners occur almost simultaneously.

  Meanwhile, the report by Shimosaka et al. Described above is a fixed reception wavelength system, in which the transmission wavelength of the optical transmitter of the transmission source is matched with the reception wavelength of the transmission destination. In this method, any number of data can be transmitted as long as there is an empty time slot in the wavelength of the transmission destination. For example, if one node A uses all the empty time slots of another node B, the node A A situation arises in which even if the intervening node C wants to send data to the node B, it cannot send it. That is, the node C has to wait for the transmission operation until the node A finishes transmitting data to the node B, resulting in a very large delay or a very high discard rate. In a backbone network for LAN-to-LAN connection, transmission requests to a number of destinations occur almost simultaneously as described above. Conversely, it can be said that a transmission request for a certain node is generated almost simultaneously from many other nodes. Therefore, the above-described network management in which the traffic is restricted by some nodes (the node A in the above case) is not suitable for the backbone network.

  In a LAN, a broadcast function of transmitting the same data to all nodes and a multicast function of transmitting the same data to a plurality of arbitrary nodes (hereinafter, broadcast and multicast are collectively referred to as broadcast / multicast) are important. is there. Since the method of Shimosaka et al. Described above is a fixed reception wavelength method, in order to realize a broadcast / multicast function, data must be copied and transmitted to all transmission destination nodes, which is inefficient.

Further, even if the reception wavelength is variable, if only one data receiver is provided in one node, all the transmission destinations of the broadcast / multicast are transmitted at the timing of the transmission time slot in the broadcast / multicast communication. The node's data receiver must be able to receive the data, that is, not receiving data from sources other than the broadcast / multicast source. In other words, if each node has one data transceiver, each receiver must have a common free time slot in order to perform broadcast / multicast communication, but such a state is realized. No consideration has been given to this. In order to solve this problem, for example, as described in Non-patent Document 2, each node may have a plurality of data transceivers, but this requires a significant increase in cost, and the system Is not preferable from the viewpoint of feasibility and cost performance.
IEICE Technical Report-Optical Communication System OCS92-77-82 IEICE Technical Report-Optical Communication System OCS92-84

  As described above, in a conventional optical communication system using a network in which time division multiplexing is used in combination with wavelength division multiplexing, traffic is limited by some nodes that transmit a large amount of data, and it is desired to perform other data transmissions. There is a problem that a node cannot transmit, and a problem that broadcast / multicast communication cannot be performed because a common free time slot cannot be obtained for all destination nodes even when an attempt is made to realize a broadcast / multicast function.

  The present invention has been made to solve such a problem, and enables communication with high throughput at a low delay and a low discard rate using a network combining wavelength division multiplexing and time division multiplexing. An object of the present invention is to provide an optical communication system capable of easily realizing a broadcast / multicast function.

  The present invention relates to a wavelength division multiplexing network in which a plurality of nodes communicate using a plurality of wavelengths time-divided into time slots, the number of time slots for realizing a lower discard rate and a higher throughput, and a time slot dividing method. And a method for distributing clocks for synchronizing data rates in the network without preparing a dedicated transceiver for clock distribution, and excessively concentrating the load on network operation on the network controller. It is intended to provide a method of assigning a path / channel identifier to prevent such a situation.

  In order to solve the above problems, the present invention connects a plurality of nodes via a wavelength division multiplexed optical network having a plurality of optical transmission channels of different wavelengths, and forms a plurality of time slots obtained by dividing the optical transmission channel of each wavelength. An optical communication system that performs transmission and reception by using a network controller that centrally controls the assignment of time slots to a plurality of nodes is provided.

  In the present invention, a control signal for notifying a transmission request (call setting request) from each node to the network controller and notifying a time slot assignment from the network controller to each node is different from a data channel for data transmission. Preferably, it is performed via a control channel.

  Further, according to the present invention, in such a basic configuration, a failure detecting means for detecting a failure of a network controller, and when the failure detecting means detects a failure of the network controller, each node assigns a predetermined time slot. And a means for switching to a pre-assignment method in which transmission and reception are performed using time slots in accordance with the above-mentioned method, or a distributed control method in which time slot assignment is determined among the nodes.

  Further, the present invention has such a basic configuration that the network controller further has a function of controlling assignment of time slots to transmission / reception nodes such that a common reception free time slot is secured for all of the plurality of nodes. Features.

  Further, when communication of connectionless data is performed by the optical communication system of the present invention, a relay node for connectionless data is connected to a wavelength-division multiplexed optical network, and a node is connected to the node by using an available bandwidth through the relay node. The transmission and reception of connectionless data may be performed between them.

  Further, according to the present invention, in an optical communication network in which a plurality of nodes communicate using a plurality of wavelengths obtained by time-dividing the frame into a plurality of time slots, the frame has a time period within one frame. An optical wavelength division multiplexing network characterized in that the number of slots is at least twice the number of nodes connected to the network.

  Further, the present invention provides an optical wavelength multiplexing network in which two or more types of time slots having different time lengths coexist in one frame in the same network.

  Further, the optical communication network has a plurality of wavelengths each time-divided into time slots, each of the time slots is used for data transmission and reception according to instructions of a network controller, and further has a control channel for network control. The network controller is connected to a public network or a device having a clock synchronized with the public network, and the network controller distributes a clock to a node connected to the optical communication network via a control channel of the optical communication network. Provide multiple networks.

  Further, an optical communication network in which light of a plurality of wavelengths is time-divided into time slots, and each time slot is used by a plurality of nodes for data transmission and reception in accordance with an instruction of a network controller, wherein a path identifier is assigned by a network controller, Provides an optical wavelength-division multiplexing network in which each node independently allocates, and further, path identifier allocation is performed independently of time slot allocation.

  In the optical communication system of the present invention configured as described above, the network controller determines how the transmitting and receiving nodes perform communication using the time slots obtained by time-dividing the optical transmission channels of the respective wavelengths by centralized control. I do. Therefore, if the time slot allocation rules programmed in the network controller are properly set, the unfairness that the other node cannot transmit due to a specific node transmitting a large amount of data as in the related art occurs. Absent.

  In this case, if a transmission request from each node to the network controller or a control signal for notifying a time slot assignment from the network controller to each node is performed through the control channel, data transmission is interrupted due to the control signal. Nothing.

  In the case where the time slot allocation is centrally controlled by the network controller as described above, in order to prevent the entire communication system from operating if the network controller fails, the present invention employs a network controller and its spare device (duplicate network controller). When the failure is detected, the time slot assignment is switched from the network controller to the pre-assignment method or the distributed control method, thereby avoiding a situation in which the communication system is stopped.

  Further, communication from a certain node to another certain node generally uses a time slot periodically at a cycle determined by a frame or the like, and it is not often possible to pack all data in one time slot and transmit it. Since there is no time slot, there is a reservation for use of the time slot not only for the currently used time slot but also for a short time before. Therefore, in order to perform broadcast / multicast communication, the position of another time slot to be used may be moved so that a common reception time slot can be allocated to all transmission destinations. In the method of starting time slot movement to create a common reception free time slot after a request is generated, it takes time to set up a call. Further, in a system configuration in which time division multiplexing is combined with wavelength division multiplexing, all of the time slots are rarely used at the same time, and are often always free to some extent.

  Focusing on this point, in the present invention, the network controller controls the assignment of time slots so that a common reception free time slot is secured for all of a plurality of nodes, so that each node has one data transceiver. Even if there is only one request, it is possible to immediately respond to the occurrence of a broadcast / multicast request.

  If connectionless data is communicated via a relay node, there is no need to issue a transmission request to the network controller for transmission and reception of connectionless data and perform a procedure. There is no need to limit the bandwidth for transmission and reception.

  In order to explain the operation of the present invention, consider an optical wavelength division multiplexing network as shown in FIG. The transmission light of each node is mixed or switched by the optical medium in the network and distributed to the receiver of each node. As shown in FIG. 37, each node selects and receives transmission light to be received by its own node for each time slot. At this time, when the number of nodes in the network is n as shown in FIG. 24, the efficiency of the network differs depending on the number of time slots in the frame (reception period) with respect to the number n of nodes. If the amount of traffic sent from one node to all other nodes is equal, or if the traffic does not change at all when the network is introduced, the number of time slots in a frame and how to separate time slots are fixed at the time of network introduction. It suffices to decide, but in reality, the traffic volume differs for each partner, and the traffic itself fluctuates with time. In such a case, the assignment of a transmission time slot from one node to another node, or the assignment of a reception time slot when a certain node receives from another node, may be performed with respect to each other node (or from each other node). For the time being, one time slot should be secured, and in addition, a time slot for absorbing traffic fluctuation or a spare time slot should be sufficient. Therefore, if the number of time slots in one frame is not larger than the number of nodes in the network, it may not be possible to sufficiently respond to a new transmission request. When a computer simulated whether or not a new transmission request could be made when a new transmission request was issued, a result as shown in FIG. 38 was obtained as an example. In the figure, the horizontal axis is a normalized value of the throughput in the network, and the vertical axis is a probability that the transmission request generated at the time of the horizontal axis is unacceptable (call loss). n is the total number of nodes. It can be seen that there is a large difference between the case where the number of time slots is the same as the number n of nodes and the case where the number is twice. Therefore, by providing the number of time slots at least twice the number of nodes, it is possible to prevent a significant call loss.

  Furthermore, since the amount of traffic differs for each communication partner node, there may be a partner whose transmission data is only short for a time slot length. Also, as described above, the probability of call blocking decreases as the number of time slots increases, but in a network such as the present invention in which a wavelength is switched for each time slot during reception, a guard time is required at the time of switching. On the other hand, if the time slots are divided too finely, the ratio occupied by the guard time increases, and the efficiency decreases. Therefore, it is desirable to reduce the number of time slots to such an extent that efficiency is not reduced and no significant call loss occurs.

  As described above, when the time slots cannot be made very thin due to the problem of efficiency, and when the time slots are all equal in size, one time slot is occupied even for a partner whose traffic amount is very small. In other words, it is uneconomic if the traffic bias is very large. Thus, if two or more types of time slots are prepared, a small one and a large one, the size of the time slot can be selected according to the amount of traffic, and the throughput is improved.

  The sixth invention relates to clock distribution. If clock speeds are not standardized in the network, when a terminal connected to one node in the network performs real-time communication (telephone, etc.) to a terminal connected to another node, a bit shift occurs due to a slight shift in clock. Is running out or is left out. Therefore, the clocks in the network need to be unified. Furthermore, the clocks must also be equal when connecting to an external device such as a public network from within the network. For this reason, a clock synchronized with the public network is usually used in a network such as the present invention. At this time, how to unify clocks becomes a problem.

  If connected directly to the public network, a clock synchronized with the public network can be obtained, but not all nodes are directly connected to the public network. In many cases, a backbone network such as the present invention itself is a gateway for connection to the public network, and if no clock is supplied via the network, other nodes are synchronized with the public network. Clocks are often not available.

  On the other hand, as a method of supplying a clock obtained by a node connected to the public network to another node, a method of providing a dedicated clock channel, and a method of extracting a clock from a transmission wave of a node connected to the public network are available. It is considered.

  However, the former method requires a transceiver dedicated to the clock for that purpose, which increases the cost. Therefore, the latter method is usually adopted. However, in the network of the present invention, since the communication is performed by switching the partner for each time slot, a node not connected to the public network receives a signal of a node supplying a clock in a certain time slot. Then, the clock must be held at the same speed until the next time slot from the same partner comes around, which is difficult depending on the frame length.

  In the above-described invention, the network controller is provided in the network for time slot allocation, and the control channel is provided for notifying the time slot allocation. In this case, the network controller transmits the control channel using the control channel. The percentage of time spent is very large and frequent, and the transmissions of the network controller are always received by all nodes. Therefore, the transmission speed of the control channel is set to a speed compatible with the hierarchy of the public network, the network controller is connected to the public network, the public network clock is received, and the network controller conforms to the public network when transmitting on the control channel. If the transmission is performed using the clock thus set, the other nodes can extract a clock suitable for the public network. As described above, since the network controller frequently performs transmission on the control channel, it is possible to hold a clock for a short time between transmissions.

  The network gives the call an identifier to identify the call. Identifiers usually include a path identifier that identifies a path that is a large delimiter and a channel identifier that identifies a call within the path. In a network such as the present invention, it is necessary to make the time slot the same as or cooperate with the path. It is a straightforward method.

  On the other hand, in the network of the present invention, the traffic fluctuates when transmitting and receiving the same partner over a plurality of time slots, and when the number of time slots can be reduced, the time slots are reduced and released. . If the time slot and the path are linked, the identifier must be reissued when the traffic fluctuates and the number of time slots is reduced. At this time, by making the physical time slot and the virtual path completely independent, even when the number of time slots is reduced due to traffic fluctuation, the assignment of the path identifier does not need to be changed, and the channel identifier is further reduced. Also get better without changing.

  In a normal switching system, such an identifier is issued by an exchange. When this is applied to the present invention, the network controller performs the processing. However, the network controller performs time slot management as its main task, and if it also issues an identifier, the processing amount becomes extremely large. Since the path identifier must be centrally managed on a network basis, the network controller performs the operation.However, unless the same path identifier is assigned to a different transmission-reception pair, the method of using the channel in the path is determined by that path. Can be delegated to the transmission / reception pair that uses, and the issue of the channel identifier is delegated to each node, thereby reducing the load on the network controller.

  According to the present invention, when the transmitting node transmits to the receiving node using each time slot obtained by time-dividing the optical transmission channel of each wavelength, the network controller intensively determines which node uses which time slot. By programming appropriate time slot allocation rules in the network controller for control and decision, time slots can be allocated without inequity and quickly, enabling low-drop rate and high-throughput communication. It becomes.

  In addition, if the network controller and its spare unit also fail, if the failure is detected and switched to the pre-assignment method or the distributed control method, such a failure may cause the entire communication system to stop. can avoid.

  Further, the network controller performs time slot allocation control in which the time slot allocation is always or whenever necessary so that the reception free time slot becomes a common time slot in each node. Even in the case where there is only one each time, a common reception time slot can be set to all the other nodes immediately when a request for broadcast / multicast communication occurs.

  Further, by setting the number of time slots at least twice the number of nodes and preparing at least two types of time slots in one frame, a lower discard rate and higher throughput can be realized. Furthermore, the clock distribution for synchronizing the data rates in the network is performed by the network controller, thereby eliminating the need for a dedicated clock channel and a dedicated clock receiver. The network controller manages / assigns the path identifier, and manages / assigns the channel identifier by the general node so that the load related to the network operation is not excessively concentrated on the network controller. Furthermore, by allocating path identifiers independently of time slot allocation, time slot arrangement and management are facilitated.

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

  FIGS. 1A and 1B are diagrams illustrating a schematic configuration of an optical communication system according to an embodiment of the present invention. In FIG. 1A, a wavelength multiplexing optical network 1 is connected to general nodes (n1 to nn) 2 for communication and a network controller (NWC) 3. The wavelength multiplexing network 1 includes, for example, a passive star type network including a star coupler 4 and an optical fiber 5 as shown in FIG. 2A, a ring type network using a ring fiber 6 as shown in FIG. Alternatively, it is constituted by another type of network. The network controller 3 may be a dedicated node as shown in FIG. 1A, or may be a configuration which also serves as any general node as shown in FIG. 1B.

  When a certain node transmits data to another node, the network controller 3 time-divides a plurality of time slots Sij obtained by time-dividing the optical transmission channels of wavelengths λ1 to λn used in the wavelength multiplexing optical network 1 as shown in FIG. Which is to be used is determined by centralized control.

FIG. 4 is a block diagram illustrating a specific configuration example of the network controller 3, which mainly includes a control signal receiver 11, a control signal transmitter 12, a signal processing unit 13, a CPU 14, and a memory (RAM) 15. The control signal receiver 11 receives a control signal such as a transmission request generation notification from each node 2 to the network controller 3. The control signal transmitter 12 transmits a control signal such as a time slot assignment notification from the network controller 3 to each node 2.

  The signal processing unit 13 classifies the control signal received by the control signal receiver 11, converts the control signal into data in a form that can be handled by the CPU 14, and transmits data such as time slot allocation from the CPU 14 by the control signal transmitter 12. The main processing is to convert the signal into a signal suitable for the following. The signal processing unit 13 also receives an alarm signal and an error signal, which will be described later, and transfers them to a display unit (not shown) for notifying them or notifying the administrator.

  The memory 15 stores a time slot assignment table for the node 2. When receiving the data received by the control signal receiver 11 from the signal processing unit 13, the CPU 14 determines the time slot assignment with reference to the memory 15 and controls the information representing the determined time slot via the signal processing unit 13. The signal is sent to the signal transmitter 12. The control signal transmitter 12 transmits a control signal for time slot assignment notification based on information from the signal processing unit 13.

  The transmission of the control signal of the time slot setting request (transmission request generation) from the node 2 to the network controller 3 and the control signal of the time slot allocation notification from the network controller 3 to each node 2 are different from the data channel for transmitting data. Through a control channel. By using the control channel in this manner, the transmission of the control signal does not hinder the data transmission. As the control channel, an optical transmission channel of a specific wavelength among a plurality of wavelengths provided for wavelength division multiplexing may be used, or an optical transmission path or an electric transmission path may be prepared exclusively for the control channel. . When the information amount of the control signal is small, a specific time slot in a transmission channel originally used for data transmission may be used as the control channel.

  FIG. 5 is an example of the configuration of the general node 2 in FIG. 1, and includes a data transmitter 21, a control signal transmitter 22, a data receiver 23, a control signal receiver 24, a WDM (wavelength division multiplexing) multiplexer 25, a WDM It has a duplexer 26 and an optical switch 27. The data transmitter 21 and the data receiver 23 transmit and receive data through a data channel that is wavelength division multiplexed in, for example, a 1.5 μm band. The control signal transmitter 22 and the control signal receiver 24 transmit and receive a control signal using, for example, a 1.3 μm wavelength-division multiplexed control channel. The WDM multiplexer 25 and the WDM demultiplexer 26 multiplex and demultiplex the data channel and the control channel in order to transmit the data and the control signal through the same optical fiber 5. The optical switch 27 may or may not be necessary, and will be described in detail later. Although not shown, a control unit is provided in the node, and controls each of the transceivers 21 to 24. Further, the data transceivers 21 and 23 have data input and output lines.

  Next, an outline of time slot allocation control by the network controller 3 in the present embodiment will be described with reference to a flowchart shown in FIG. First, a control signal notifying that a network controller has issued a slot allocation request from a general node (transmitting node) to another general node (receiving node) is transmitted from the control signal transmitter 22 and received by the control signal receiver 11. Then (S11), the received data is input to the CPU 14 via the signal processing unit 13, and the CPU 14 determines whether or not there is a time slot that is common to the transmitting node and the receiving node (referred to as a common free time slot). 15 (S12). If there is a common vacant time slot, information indicating that the common vacant time slot is allocated to data transmission / reception is output from the CPU 14 to the control signal transmitter 12 via the signal processing unit 13, and the control signal transmitter 12 A control signal for notifying that a common free time slot is allocated to transmission and reception is transmitted to the transmitting node and the receiving node (S14). On the other hand, if there is no common free time slot, the time slot assignment of the transmitting node or the receiving node is changed (S13), and the process proceeds to S14. If the assignment is not possible in S12 and S13, the transmission node is notified of the slot assignment failure. In the transmitting node and the receiving node, when the control signal receiver 24 receives the control signal for notifying the time slot allocation transmitted from the network controller 3, the data transmitter 21 and the data receiver 23 use the time slot. Send and receive data.

  7 and 8 show examples of specific time schedules for transmission and reception, where (a) shows the time slot assignment of the transmitter and (b) shows the time slot assignment of the receiver. S1 to Sm are time slots, and a time slot with a circle represents a time slot used for transmission or reception. Here, the transmission wavelength of each node is fixed and set to a value unique to each node, and the receiver side switches the reception wavelength for each transmitter to receive the signal. However, the transmission and reception wavelengths are both variable. May be. However, when changing the transmission wavelength, an optical switch 27 is provided between the data transmitter 21 and the WDM multiplexer 25, for example, as shown in FIG. Until the value is stabilized at a correct value, it is necessary to prevent light from being transmitted to the wavelength division multiplexing network 1.

  Here, it is assumed that a unit consisting of m continuous time slots is a frame, and that transmission and reception are performed by repeating the frame. However, it is also possible to determine that transmission to a certain party is performed every 10 time slots and transmission to a different party is performed every 20 time slots, so that communication can be performed without particularly determining a frame.

  Now, suppose that a new transmission request is generated from the node n1 to the node n3, for example, in the state shown in FIGS. In this case, the node n1 notifies the network controller 3 via the control channel of a control signal indicating which node and how many slots are required as a transmission request. The network controller 3 having received this notification searches for a common portion between the transmission free time slot of the node n1 and the reception free time slot of the node n3, that is, a common free time slot. Here, assuming that the transmission request from the node n1 is one time slot / one frame (one time slot per frame), the third time slot S3 from the top in the state of FIGS. Since it is free, the network controller 3 notifies the node n1 on the control channel to transmit data to the node n3 using the time slot S3 from the next frame, and notifies the node n3 from the node n1. The control channel informs that data transmission is to be performed by the time slot S3 from the next frame. Also, if the transmission request has a severe delay, if possible, consider allocating the end of the frame so that transmission from the current frame is possible instead of the next frame. Can be

  By deciding the time slot allocation in this way, even if one node ni issues a transmission request to another node nj to transmit data using a very large number of time slots, If a transmission request to the node nj is issued or expected to be transmitted from the node nj, the network controller 3 can also instruct the node ni to reduce the traffic of the transmission request of the node ni. Is not confused for some nodes.

  On the other hand, when a common free time slot is not found, a common free time slot is created by moving another time slot that can be moved. For example, in FIGS. 7A and 7B, it is assumed that the node n2 wants to transmit to the node n4 in one time slot / one frame, but a common time slot cannot be found between the transmission of the node n2 and the reception of the node n4. Here, paying attention to the transmission from the node n2 to the node n1, since S3 is used as the transmission time slot, the transmission in the time slot S3 moves to the time slot S2 as shown by the arrow. As a result, as shown in FIGS. 8A and 8B, the transmission time slot S3 of the node n2 and the reception time slot S3 of the node n4 are both vacant, and the time slot S3 allows transmission from the node n2 to the node n4. Becomes possible. Therefore, the network controller 3 notifies the corresponding nodes that the transmission time slot from the node n2 to the node n1 is moved to S2 and that the time slot S3 is the transmission time slot from the node n2 to the node n4. By collecting things without confusion.

  Here, if the same time slot allocation is to be performed by distributed control, which is determined through discussions between the nodes through a control channel, for example, first, node n2 transmits to node n4 the number of its own idle time slot and the number required for communication. The number of time slots is notified, and the node n4 compares it with its own reception free time slot number, and notifies the node n2 that a common time slot cannot be obtained and its own reception free time slot number. Then, the node n2 examines the reception free time slot of the node n4, finds a time slot candidate that can be taken in common with the node n4 if it moves to any of its own transmission time slots, and sets it as a target to be moved. It requires a very complicated series of procedures, such as notifying the other party transmitting in the time slot of his or her own transmission free time slot number and asking whether it can be moved somewhere.

  On the other hand, if the network controller 3 is provided with a time slot assignment table for all the general nodes 2 as in the present embodiment and the time slot assignment is centrally controlled based on the table, There is no need to perform the complicated procedure as described above, and time slot assignment can be performed extremely easily.

  It is also possible to prepare a time slot allocation table for all the general nodes 2 and to negotiate the time slot allocation based on this table. If the time slot assignment is changed or moved, there is a possibility that a transmission or change request may be issued by mistake before the notification is made, and the negotiation may be confused. In order to avoid confusion, it is only necessary that all the general nodes 2 have the same level of function as the network controller 3 and operate with a common and deterministic algorithm. Become higher. Therefore, a method in which one network controller 3 determines the transmission and reception time slot allocation of all the general nodes 2 as in the present embodiment is advantageous.

  As another solution when a common free time slot is not found, a method as in the following embodiment can be adopted. FIG. 9 is a schematic configuration diagram of an optical communication system according to another embodiment of the present invention, in which one (nr) of the general nodes 2 is used as a bucket brigade node. In the present embodiment, when the network controller 3 cannot find a common free time slot, it transmits the transmission data from the transmission node to the bucket relay node nr once as shown in FIG. Take the method of sending to the target node. FIG. 10 shows an example in which transmission from the node n2 to the node n4 is desired to be performed in one time slot / one frame as in the example of FIG. 7, but there is no common time slot between the transmission of the node n2 and the reception of the node n4. . In this case, after transmitting the transmission data from the transmission node n2 to the bucket relay node nr using the time slot S2, the bucket relay node nr transmits the transmission data to the reception node n4 using the time slot S3 so as to relay the transmission data. ing.

  In FIG. 10, one of the general nodes 2 also serves as the bucket relay node nr. However, it may be prepared as a dedicated bucket relay node, or a bucket relay function may be provided in the network controller 3 in advance. Good.

  In addition, when transmission from a certain node ni to another certain node nj is completed, or when the number of time slots in use decreases and the time slots used so far become free, the time slot becomes free. When the time is known, the node ni informs the network controller 3 of which time slot becomes empty and when. Alternatively, when the transmission is completed, that is, when the time slot is vacant, the network controller 3 may be notified that the time slot is vacant. In this way, the network controller 3 keeps track of the time slot usage. The range of processing by the network controller 3 may be a method of managing on a call basis. However, since the processing amount increases, transmission requests and the like are made in slot units, and how to use the slots is determined by each general node 2. It is desirable to do. This is because, when a new call is generated, if there is no problem with the remaining bandwidth of the already set slot, it is not necessary to make a new slot request.

  When the optical transmission channel of each wavelength is divided into a plurality of time slots and used as in the optical communication system of the present invention, the timing is commonly used so that the timing of the time slot does not differ for each wavelength as follows. There is a need. That is, in the case where the wavelength multiplexing optical network 1 is of a passive star type as shown in FIG. 2A, if the timing is matched at the time when the optical signal arrives at the star coupler 4, the waveforms shown in FIGS. Optical signals from each node can be mixed by the star coupler 4 at the timing shown. To this end, if the time ti for the light to travel from its own node to the star coupler 4 is known, one of them may provide a timing signal and adjust it accordingly.

  For example, if the network controller 3 gives a timing signal via the control channel, the time slot transmitted by the own node is set at the start of the star coupler 4 for ti seconds until the timing signal arrives at the node ni after passing through the star coupler 4. It also takes ti seconds to arrive at. Therefore, if the node ni transmits the beginning of the time slot 2 ti seconds before the timing signal reaches the node ni, the start of the time slot transmitted by the node ni by the star coupler 4 and the timing signal are mixed at the same timing. become. If the timing signal is given periodically, such as once in one time slot, once in one frame, or once in several time slots, the node ni transmits the beginning of the time slot two ti seconds before. It is also possible to do.

  The time ti may be obtained by calculation based on the cable length measured in advance when the optical fiber 5 is laid, or may be obtained by actually measuring the time required for the signal to reciprocate. The timing signal may be given through the control channel as described above, or, for example, the network controller 3 may transmit data on the data channel once per frame and obtain the timing signal based on the data.

  Similarly, when the wavelength multiplexing optical network 1 is a ring network as shown in FIG. 2B, the network controller 3 periodically supplies a timing signal via a control channel, and each node 2 transmits data in accordance with the timing signal. do it.

  The control channel is a simpler transmission / reception system than the data channel. Therefore, the control channel itself does not basically have a plurality of channels due to wavelength division multiplexing or the like, and even if it has a plurality of channels, the wavelength is changed between the upstream and downstream as shown in FIG. And it is only about using a separate cable in the downstream. In this case, since all nodes transmit and receive control signals on the control channel using the same wavelength, it is necessary to determine a protocol such as the order of transmission and reception. For example, as shown in FIG. 11, when an uplink, that is, when the general node 2 transmits a control signal, the control channel is time-divided into a plurality of time slots and the transmission order of each node 2 is determined in advance by TDMA (time division When the network controller 3 transmits a control signal, a method such as TDM (time division multiplexing) for sequentially transmitting a control signal to each node 2 may be adopted. . Further, the frame length of the control channel and the frame length of the data channel may not be equal, and it is preferable that one data frame length has a plurality (preferably an integer) of control channel frames for more flexible management.

  Next, a description will be given of measures to be taken when the network controller 3 fails or a spare network controller described later fails. In such a case, the network controller 3 or its spare device, or the general node 2 is provided with a failure detection function of the network controller 3, and when a failure of the network controller 3 is detected, the communication throughput is deteriorated. Then, it is assumed that the method is switched to the pre-assignment method or the distributed control method as shown in FIG.

  The pre-assignment method is a method of allocating time slots according to a predetermined time slot allocation. In the example of FIG. 12, a time slot is allocated so that data is transmitted evenly from one node to another node, and a time slot is allocated during reception. Data from the other nodes in order. As another method, the traffic may be declared in advance, and more time slots may be allocated to the node pairs having a large traffic based on the traffic. Further, a method of not changing the time slot allocation at the time when the network controller 3 fails may be adopted. Although this method reduces the discard rate and the throughput, it is very simple, and is effective when it is expected that the repair of the network controller 3 will be completed relatively quickly.

  The distributed control method is a method in which the individual general nodes 2 determine the time slot allocation by discussing through the control channel as described above. If there are few empty time slots, the negotiation becomes very complicated as described above. Takes time. In addition, when the change in the communication amount is large, it is possible to speak to a plurality of parties at the same time, and there is a possibility that replies may be overlapped and confusion may occur. Therefore, when the distributed control is performed, it takes a long time to perform the control for collecting the situation, and as a result, the discard increases, and the throughput decreases. However, if the transmission / reception time slot of each node has a relatively large amount of free space and the change in the communication amount is gradual, the distributed control method can sufficiently cope with this. Further, the distributed control method has a smaller discard rate than the pre-assignment method, and the decrease in throughput is small.

  In this embodiment, a basic failure can be detected by the network controller 3 and the general node 2 so that the time slot allocation method can be switched to the pre-assignment method or the distributed control method when the network controller 3 fails. Have a self-diagnosis function. Since a failure that cannot be detected by the self-diagnosis may occur in the network controller 3, in order to detect such a failure of the network controller 3, for example, the network controller 3 is connected to the network controller 3 as shown in FIG. A monitoring device 7 having an independent transceiver (at least a transceiver for control signals) is arranged. Further, in consideration of the possibility of a power failure or other power failure, the monitoring device 7 is arranged at a different position as shown in FIGS. 14B and 14C without being placed at the same location as the network controller 3. You may do so. The monitoring device 7 may be an independent dedicated device, or may be one of the general nodes 2.

  Further, since there is a possibility that the monitoring device 7 itself may fail, it is desirable to provide a plurality of monitoring devices 7 in a system having a sufficient cost. Specifically, for example, a plurality of dedicated monitoring devices 7 may be prepared, or a spare network controller may have a monitoring function, or some general nodes 2 that also have the function of the monitoring device 7 may be prepared. .

  When the monitoring device 7 or the general node 2 having the monitoring function detects an abnormality of a device or node other than its own, the self-diagnosis function determines whether the abnormality is due to its own failure. An operation that has a large effect on the operation of the system when it is not its own failure, for example, an operation such as interrupting the transmission of a control signal from the network controller 3 through a control channel as described later or stopping the network controller 3 Since the self-diagnosis function may be out of order, the device which has found the fault, other devices other than the node, and whether or not the node has also detected the fault are checked through the control channel. For example, if you find a failure other than yourself, you only need to monitor the control channel if you always want to report on the control channel, otherwise check through the control channel What should I do? After confirming the failure, the operation that greatly affects the operation of the system is performed for the first time. If the failure cannot be confirmed, there is a possibility of the failure, and the fact is notified to the network controller 3 or the network administrator.

  The monitoring device 7 monitors the transmission / reception contents of the network controller 3 for the presence or absence of an abnormality (failure). When the abnormality is confirmed, the monitoring device 7 communicates with any or all of the network controller 3, the network administrator, and the entire system through a control channel. Issue an alarm signal. If the monitoring device 7 is directly connected to the network controller 3 as shown in FIGS. 14A and 14B, the alarm signal to be sent to the network controller 3 or the network administrator can be sent using the direct connection. If it is not directly connected, send an alarm signal through a different external network such as a public network, or prepare a dedicated wavelength for the alarm signal, or use a dedicated control channel for the alarm signal. For example, there is a method of preparing a time slot.

  When detecting an abnormality in the network controller 3, the monitoring device 7 first notifies the network controller 3 of the abnormality as described above, and causes the self-diagnosis function of the network controller 3 to perform a self-diagnosis. At this time, as shown in FIG. 15B, an alarm signal as short as not to hinder the function of the communication system may be output to the entire communication system. Also, no improvement is seen with this alarm, and if the failure is not a failure of the failure detection function itself as described above and the degree of abnormality of the network controller 3 is so large that the communication system cannot operate, When an alarm signal is output to cover the control signal as shown in FIG. 15 (a), and when the degree of abnormality is small and does not significantly affect the communication system, as shown in FIG. The alarm signal may be output in a pulse shape during a guard time or the like. In this case, it is preferable that the type of the pulse constituting the alarm signal is determined in advance as shown in FIG. 15C according to the content of the abnormality, so that the content of the abnormality can be understood by the alarm. .

  Furthermore, when the degree of abnormality of the network controller 3 is large and cannot be repaired immediately, it is preferable to give authority to stop the network controller 3. In addition, when there is a spare network controller 9 as shown in FIG. 21, the spare network controller 9 may be switched to the spare machine 9 and a dedicated monitoring device may operate only to detect a failure of the spare machine 9. The operation of stopping the network controller 3 can be easily realized when the monitoring device 7 is directly connected to the network controller 3 as shown in FIGS. 14A and 14B, for example, and is directly connected as shown in FIG. If not, the network administrator may be notified and stopped. Further, a function of stopping the network controller 3 when a special signal is received on the control channel (a function of starting the operation of the spare device 9 if there is the spare device 9) is provided. The monitoring device 7 may output it.

  Further, when the network controller 3 including the standby unit 9 stops or is stopped artificially, the network controller 3 instructs each node through the control channel from what timing and from what protocol (described above) to perform communication. . For example, the monitoring device 7 may transmit such contents at a time when the network controller 3 transmits the control signal on the control channel or at a scheduled transmission time. The general node 2 starts monitoring the network controller 3 from the time when a serious abnormality occurs in the network controller 3, that is, when the monitoring device 7 notifies the occurrence of the abnormality, or when the monitoring device 7 outputs a signal for stopping the network controller 3. It is advisable to take measures such as not changing the time slot assignment until an instruction is received from.

  It is not always necessary to specially prepare the above-described monitoring device 7. The general node 2 has at least an abnormal operation of the network controller 3 related to the own node, for example, a transmission time slot is allocated even though the own node has not issued a transmission request, or a new time slot allocation is overlapped with a used one. Abnormalities such as the presence of an error can be detected. In such a case, the node 2 first informs the network controller 3 via the control channel that the operation is abnormal. If the abnormality of the network controller 3 still does not improve, if the transmission state of the control signal on the control channel is in the form as shown in FIG. A certain node (for example, a node that rarely shuts down, such as a node serving as a public network gateway) should always listen to the transmissions of other nodes, and The detected node notifies the specific node of the abnormality of the network controller 3 and its degree in its assigned time slot of the control channel. The specific node that has received the notification notifies the network administrator via the public network, the hotline, or the wavelength multiplexing optical network 1.

  When the network controller 3 causes a more serious abnormality, when the specific node signals through the control channel, the network controller 3 switches to the standby unit 9 or the standby unit 9 breaks down, or the standby unit 9 If not, switch to the backup protocol described above. The judgment may be made by a specific node or may be made by a network administrator, and differs depending on the system.

  This method can be applied not only to an abnormality in the network controller 3 but also to an abnormality in another node. If a node that communicates with or is to communicate with its own node is not communicating correctly, it can be detected immediately because the data is not correctly received. If this detection result is notified to the network controller 3, the network controller 3 can instruct a node that is performing abnormal communication to improve it.

  As a simple countermeasure against the abnormality of the network controller 3, a method of coping with the abnormality only when the network controller 3 has failed to such an extent that any node can detect the abnormality can be considered. For example, the network controller 3 does not transmit on the control channel, the transmission power on the control channel is abnormally reduced, or the timing of transmission on the control channel is obviously wrong. In addition, this includes the case where the network controller 3 detects its own abnormality and notifies that it cannot operate.

  In these cases, when each node 2 first detects an abnormality, the node 2 notifies the network controller 3 or another node of the abnormality through a control channel. If the restoration is impossible, it is sufficient if the network controller 3 itself can instruct the switching of the protocol of the above-described method. If the instruction does not come, for example, a specific node is determined, and the determination by the node is performed. Instruct each node to switch. Any one of the specific nodes may be determined in advance. However, in a LAN (local area network), particularly a wavelength division multiplexing LAN, the nodes may be inactive. So that the first node gives an instruction first, and when the first node does not give an instruction after waiting a predetermined time, the second node gives an instruction, and so on. Is also good. The node that issues the instruction may determine that the failure of the network controller 3 is actually the failure of its own node. Is preferable.

  Further, when it is found that the own node is out of order, the network controller 3 is notified of the failure, and if there is an administrator of the own node, it is notified. The network controller 3 notifies the network administrator and takes action according to the degree of failure of the node. The contents of the countermeasures are, for example, that the node does not change the time slot assignment, stop the communication, etc., and notifies other nodes of the abnormality of the node and the degree of the abnormality.

  Further, in the above-described optical communication system, the control signal may not only determine a time slot assignment but also play a role of a timing signal indicating a data transmission timing. Therefore, in order to maintain the timing even if the network controller 3 fails, for example, the monitoring device 7 or the general node instructing the switching of the protocol may give the timing in place of the network controller 3, It is good to give instructions and give timing.

  Next, a broadcast / multicast function in the optical communication system of the present invention will be described. In a LAN, the broadcast / multicast function is important. However, if this is to be realized in an optical communication system as in the present invention, there is a possibility that a plurality of transmission destinations may not be able to take common reception free time slots. For this reason, the network controller 3 manages and arranges the time slot allocation for transmission and reception of each node so that it can immediately respond to a broadcast / multicast request.

  Specifically, for example, some time slots at the end of the frame are always free for a time when a broadcast / multicast is requested. Therefore, when there is a normal transmission request other than the broadcast / multicast, a time slot that can be transmitted is searched for as early as possible in the frame. If data cannot be transmitted without using some time slots reserved for broadcast / multicast, data transmission is started using those time slots, and the time slot allocation method described above is used. As described above, a common empty time slot for transmission and reception is created by moving one of the earlier time slots that can be moved.

  Also, in order to prevent such a situation from occurring as much as possible, for example, the network controller 3 that transmits and receives data using a time slot near the end of the frame is shifted as far forward as possible by the network controller 3. Preferably, control is performed. For example, a list of transmission / reception time slots that should be moved is listed up based on the usage status of time slots, etc., and when one of the time slots becomes a new free time slot, calculation is started, and It is checked whether or not the listed set can be moved using the time slot. If the processing capacity of the network controller 3 has sufficient capacity, it may be checked whether or not efficient time slot allocation packed forward can be performed even when a new free time slot is generated. These calculations may be performed by providing a dedicated processor in the network controller 3 according to the capability of the network controller 3, or may be performed during normal control.

  Also, here, the time slots are described as being packed in front of the frame as much as possible.However, the time slots may be packed up as much as possible, or the time slot numbers and priorities to be vacated for broadcast / multicast are determined. Then, similar control may be performed.

  By the way, in an optical communication system using wavelength division multiplexing and time division multiplexing as in the present invention, one time slot is much longer (10 times as large as a cell which is an information transfer unit in ATM (asynchronous transfer mode) exchange). (About 100 times). When the optical communication system of the present invention is used as a backbone LAN, data from a plurality of terminals of different lower networks connected to the transmitting side are put together in one time slot and are connected to the receiving side. Will be transmitted to a plurality of different terminals in the lower network.

  However, in the case of performing the broadcast / multicast communication, the transmission data from one node used for the broadcast / multicast communication only needs to be an amount that can always fill one time slot. For example, when the terminal is only one terminal of the lower network, the time slot may be almost empty even though one time slot is allocated. Furthermore, a one-way broadcast / multicast communication such as a broadcast may be used. However, in a system in which each node performs transmission, such as a video conference system in which a plurality of locations are connected, the number of times when a time slot is almost empty is reduced. Is also occupied, which is inefficient.

  Therefore, it is preferable that the time slots to be left open for broadcast / multicast communication be designed to have a shorter time slot length than other time slots as shown in FIG. Of course, when a time slot for normal communication cannot be obtained in other parts, normal transmission and reception may be performed using this short time slot.

  In the case of performing the broadcast / multicast communication as in the TV conference system described above, a specific communication amount may be applied to each node, and a new time slot may be newly divided at a time corresponding thereto. . FIG. 13B shows this state, for example, one normal time slot is finely divided according to the amount of communication information. If the amount of communication information does not fit in one time slot, it is sufficient to use as many time slots as fit.

  Further, another information may be carried in the empty portion of the time slot. For example, when a plurality of broadcast / multicast communication data are transmitted from one node, the communication data may be collected into one time slot, or data for a specific partner may be put in an empty portion of the time slot. You may send it.

  In an application such as a TV conference system in which there are a large number of broadcasters and the delay may be large, data from another broadcaster of the application is sent to a certain node by normal transmission and reception. Therefore, it may be transmitted once again to one time slot.

  Here, if a common reception free time slot cannot be obtained for all broadcast partners, the following method may be used. For example, even if one common free time slot cannot be obtained, a broadcast partner is divided into groups in which a common free time slot can be obtained within a group, and the sender transmits the same information by the number of the groups. At this time, if the sender of the broadcast does not have a transmission time slot in those time slots, the relay is performed by a bucket brigade to a group in which the sender cannot transmit. This relay may be performed by the above-described bucket relay dedicated node, or may be performed by the network controller 3 or another general node. Also, if a node that can transmit in a common free time slot of a group that a broadcast sender cannot transmit is in a group that a broadcast sender can transmit, that node may relay. Further, when the general node 2 does not have such a copy function, the relay may be performed by the bucket relay dedicated node.

  Next, a case where the wavelength division multiplexing optical network 1 is extended in the present invention will be considered. When the wavelength multiplexing optical network 1 is a passive star network as shown in FIG. 2A, it is difficult to extend the network 1 beyond the number of ports of the star coupler 4. Although it is possible to connect two or more nodes to one port via a coupler, these nodes have a problem that the transmission / reception light intensity decreases and the reception sensitivity deteriorates. If this deterioration is a problem, it is necessary to take measures such as inserting an amplifier into the optical transmission line or preparing a transceiver having a large transmission power and a good reception sensitivity, which increases the cost. On the other hand, in the transmission wavelength fixing method, one wavelength is allocated to one node, and therefore, the total number of wavelengths, that is, the number of nodes is limited due to the wavelength variable range of the receiver.

  FIG. 16 shows an embodiment in which the wavelength division multiplexing optical network 1 is extended beyond such a limit, and a plurality of wavelength division multiplexing optical networks 1 are connected by a gateway 8. The gateway 8 is configured, for example, as shown in FIGS.

  The gateway 8 shown in FIG. 17 is configured by cascade-connecting a gateway receiver 31, a signal processing unit 32, and a gateway transmitter 33, and converts an optical signal addressed to a wavelength-division multiplexing optical network from a transmitting side to a wavelength-division multiplexing optical network on a receiving side. The signal is converted into an electric signal by the gateway receiver 31, subjected to predetermined processing by the signal processing unit 32, converted to an optical signal by the gateway transmitter 33, and transmitted again. One or a plurality of gateway receivers 31, signal processing units 32, and gateway transmitters 33 are prepared according to traffic. It is assumed that the reception wavelength of the gateway receiver 31 and the transmission wavelength of the gateway transmitter 33 follow the method of the wavelength multiplexing optical network connected to each.

The gateway 8 shown in FIG. 18 includes a wavelength selecting device 41 and a wavelength converting device 42. The wavelength selecting device 41 selects a time slot on the transmission side instructed by the network controller 3, and the time slot is designated by the wavelength converting device 42. The signal is converted into an optical signal of a wavelength designated by the wavelength multiplexing optical network on the receiving side and output.

  More specifically, as shown in FIG. 19, an optical signal from a wavelength-division multiplexing optical network on the transmission side is divided for each wavelength by a wavelength demultiplexer 51, and this is input to an optical switch 52, and wavelength division multiplexing on the reception side is performed. A time slot destined for the optical network is selected, the wavelength is further converted by the wavelength converter 53, multiplexed by the wavelength multiplexer 54, and output. As another configuration method, as shown in FIG. 20, the coupler 61 and the tunable filter 62 have the functions of the wavelength demultiplexer 51 and the optical switch 52 in FIG. 19, and the optical signal output from the tunable filter 62 is The wavelength may be converted by the wavelength converter 63 and multiplexed by the coupler 64 for output.

  Next, a method of supporting connectionless data (hereinafter, CL data) in the optical communication system of the present invention will be described. As a method for supporting CL data, for example, there is the following method. First, as shown in FIG. 22, a relay node 10 for relaying CL data is connected to the wavelength division multiplexing optical network 1. The CL data is transmitted once from the general node 2 that transmits the CL data to the relay node 10, and is transferred from the relay node 10 to the destination node. The transmission and reception of data is performed by previously determining a node that communicates with the relay node 10 for each frame. In a network in which the concept of a frame is not introduced, the same operation is performed by dividing the time slot sequence into several slots.

  When there is data to be transmitted at the time when the order of the own node arrives according to a predetermined transmission order, that is, at the time when a frame in which the own node can transmit CL data arrives, the general node 2 has an available transmission space of the own node. Insert CL data into the time slot. Further, in a network with a lot of traffic where it is frequently expected that there is no free time slot even though there is CL data to be transmitted, a free band, that is, the remaining band of the slot being used may be used. At this time, it is preferable to attach a header so that the CL data can be distinguished as normal data.

  For example, information indicating that fact may be written in the header of the data so that the relay node 10 knows that the content of the time slot is CL data, but it is easy to determine at the hardware level. You may keep it.

  For example, if the wavelength multiplexing optical network 1 is a variable transmission wavelength network, a dedicated wavelength for CL data is determined, and the relay node 10 tunes the reception wavelength of the receiver to the transmission wavelength of CL data. Only CL data can be received. When the wavelength multiplexing optical network 1 is a fixed transmission wavelength network, there is a method of transmitting a specific pulse train at the beginning of a time slot. Further, the relay node 10 previously communicates with the network controller 3 or monitors the control channel to grasp the transmission / reception time slot allocation of the general node 2 by, for example, having a time slot allocation table therein. If there is data in a transmission free time slot of the node, there is a method of receiving the data.

  The receiver in the relay node 10 receives transmission data of a specific node in a certain frame in accordance with the transmission order of CL data of the general node 2, extracts only CL data from the transmission data, and stores it in a memory or the like. At this time, it is preferable to sort the data for each destination by looking at the destination of the CL data. Specifically, for example, when the general node 2 transmits the CL data to the relay node 10, the general node 2 may transmit the CL data in a different time slot for each destination, or the general node 2 may care about the destination at all. It is also possible to simply transmit the message without relying on the relay node 10 and leave the sorting to the relay node 10. These are different for each network.

  Further, in the memory in the relay node 10, for example, a memory area may be divided in advance for each destination, and when data of the destination comes, the data may be stored in that area, or the destination is immediately determined in the data. A tag that can be attached may be added.

  Transfer of CL data from the relay node 10 to the general node 2 is performed in the same manner. That is, the general node 2 determines the order of CL data reception, and tunes to the transmission from the relay node 10 in the reception free time slot of the own node when a frame in which the own node receives CL data comes. . The relay node 10 contacts the network controller 3 in advance, or grasps the time slot allocation by monitoring the control channel, and transmits the data at the timing of the reception free time slot of the partner node. At this time, the relay node 10 extracts and transmits the data destined for the partner node at that time among the data stored in the memory.

  In this way, by providing the relay node 10 for CL data in the network and transmitting and receiving CL data using an empty time slot, it is necessary to issue a request to the network controller 3 for CL data transmission and reception and perform a procedure. In addition, the band for transmitting and receiving other normal data due to CL data is not limited. Of course, some of the CL data has a severe delay, so such data should be sent in the normal procedure, or if a time slot is set with the partner node, the remaining bandwidth will be used. Just send it. At this time, the remaining band in the time slot generated by the statistical multiplexing effect may be used.

  Next, details of the above-described optical wavelength division multiplexing network will be described.

  FIG. 23 shows an example of a method of dividing a time slot of a wavelength division multiplexing optical network having n nodes as shown in FIG. In FIG. 23A, there are 2n time slots in one frame. As described in the operation, if the number of time slots is twice or more than the number of nodes, remarkable call loss hardly occurs. If the system has a short guard time for wavelength switching, the throughput can be further increased by setting the number of time slots to be larger than 2n as shown in FIG.

  In a network such as the present invention, the number of nodes 2 actually connected to the network often differs from time to time and from time to time. However, since the maximum number of nodes 2 that can be connected to the network is almost always determined, the number of time slots may be determined in advance based on the maximum number. If the number, length, or frame length of time slots can be changed at a slow speed (on a daily or monthly basis), change the number of nodes when the number of nodes changes. Is also good. Alternatively, even if the number of nodes does not change, the number of time slots is too large or too small relative to the number of nodes, and the efficiency or the total throughput of the network 1 becomes unsatisfactory (or likely). It may be changed at the time.

  FIG. 25 is an example in which two or more types of time slot sizes are prepared to cope with traffic bias. Even in this case, of course, the total number of time slots in one frame is preferably at least twice the number of nodes. The size of the time slots and the number of the time slots to be prepared depend on the traffic conditions of the network.

  For example, in a network where the bias of the partner node is relatively small, a time slot for transmitting several to several tens of telephones for transferring small files such as telephones and e-mails is prepared for the number of nodes. In preparation for transmission of large burst data such as CAD data, it is conceivable to use a large time slot such that image data passes in one time slot. In a highly biased network, prepare a small time slot smaller than the number of nodes. After that, a large time slot and a medium time slot may be prepared as shown in FIG.

FIG. 25A shows an example in which there are two types, a fine time slot and a large time slot. (B) shows another example in which three types of time slots (M _{1 to} M _M ) are prepared in combination with the narrow time slots (M _{1 to} M _M ) for broadcasting described in the above embodiment. (C) shows an example in which there are three types of time slots, large, medium and small. (D) divides the frame into very thin time slots, and uses several thin time slots in succession according to the traffic volume (no guard time is taken between them) to equivalently reduce the time. An example in which the length of the slot is changed is shown. In the case of (d), since the reception wavelength is actually switched only at the portion indicated by the arrow, the number of times of the guard time does not increase too much and the efficiency does not deteriorate. Furthermore, instead of being divided into thin time slots as in (e), it can be divided at an arbitrary position. However, in the wavelength-division multiplexing network of the present invention, since the wavelength is switched for each partner as shown in FIG. Complex algorithms may be required.

  FIG. 26 shows a configuration example of a system in which the network controller 3 extracts a clock from the public network and distributes the clock to general nodes through a control channel in order to synchronize the clock of the wavelength division multiplexing network of the present invention with the clock of the public network. The network controller 3 is connected to the public network 20 and serves as a gateway to the public network 20. Alternatively, even if it is not connected to the public network 20 (even if it cannot be connected to the public line), it may be connected to a line from which a clock synchronized with the public network 20 can be obtained. FIG. 27 shows an example in which the gateway to the public network and the network controller 3 are different nodes. Here, the network controller does not undertake connection to the public line.

  FIG. 28 shows an example in which only a clock is sent from the gateway node to the public network 20 to the network controller 3 by a hot line or the like.

  In this way, the network controller 3 obtains a clock synchronized with the public network, transmits the clock at the speed of the clock when the network controller 3 transmits on the control channel, and each node 2 receives the clock and extracts the clock. By doing so, each node obtains a clock synchronized with the public network 20.

  FIG. 29 is a configuration example of a node when a control channel is multiplexed by WDM. In this configuration, the WDM multiplexer / demultiplexer 71-1 connected to the optical fiber 72 is coupled to the data transmitter 76 and the control signal transmitter 77, and the WDM multiplexer / demultiplexer 77-2 is connected to the data receiver 78 and the control signal Connected to receiver 79. Data transmitter 76, control signal transmitter 77, data receiver 78 and control signal receiver 79 are coupled to node controller 80. In this configuration, since the wavelength of the control channel is shared by all the nodes, the network controller and each of the general nodes transmit and receive all the transmitted control information in a time-sharing manner as in a normal LAN. become. One example of the time division method is shown in FIG. FIG. 30A shows that information from the network controller 3 is continuously transmitted, and transmission from each general node is performed by TDMA. This method is effective when the time during which the NWC is not transmitting is short, such as when the number of nodes is relatively small or the amount of information from a general node is short, and each node can hold the clock correctly during that time. Otherwise, the general node and the network controller are transmitted alternately as in (b), or the general nodes are divided into several groups as in (c) and what is similar to (a) is performed. It is preferable that the time during which the network controller 3 is not transmitting is kept within the time that the clock can hold by taking a method such as making the cycle one or more times.

  As the clock transmission form, in addition to the method of setting the bit rate of the control channel to a bit rate synchronized with the public network 20 as described above, an integer multiple or an integer of any bit rate of the hierarchy of the public network is used. There is also a method of transmitting at an easy-to-understand bit rate such as 1/100. Further, it is possible to arbitrarily set the bit rate of the control channel and simultaneously deviate the microwaves of the clock frequency synchronized with the public network to the transmission light of the control channel as shown in FIG. As shown in FIG. 31 (b), the control information may be modulated into a microwave having a clock frequency, and the modulated wave may be used to modulate the transmission light of the control channel (subcarrier modulation). FIG. 31 shows the transmission spectrum at the electric stage on the control channel of the network controller.

  As shown in FIG. 31A, the network controller can continue to transmit the clock even when another node transmits on the control channel, and the general node can always be supplied with the clock. At this time, if the control channel is used in a time-division manner so as to occupy a certain optical wavelength, if there are simultaneous transmissions from different nodes in that wavelength band, the optical carriers of those transmission lights will interfere with each other. Together, beat noise occurs. Therefore, when the network controller transmits only the clock at the time of transmitting the control information of another node, the wavelength of the control channel transmitter of the network controller 3 and the other nodes are reduced to such an extent that the beat noise does not enter the reception band. It is good to keep the wavelength of the transmitter for the control channel away.

  Also, if the method as shown in FIG. 31 is adopted, it seems that the band of the control channel transceiver must be expanded. In the example of FIG. 31 (b), the band must be expanded. In (a), since only the clock is in the high-frequency portion, there is almost no need to worry about the receiving sensitivity, and it is only necessary to pass the sine wave of the clock frequency. Need to be changed, but there is no need to replace the transceiver with a higher bit rate one.

  FIG. 32 is an explanatory diagram showing paths in the network. In a network like the present invention, it can be considered that a line is substantially formed for each transmission-reception pair as shown in FIG. The "network" provides one or more path identifiers for the transmission-reception pair (for one line in FIG. 32 (b)). In the case of the present invention, the “network” is specifically the network controller 3. However, if there is a dedicated node, either the dedicated node or the general node 2 may serve as the network controller 3.

  Basically, one node issues a path identifier in the network 1. Also, node i transmission-node j reception and node j transmission-node i reception are different and treated as independent paths.

  One or more path identifiers given to one transmission-reception pair are given according to the amount of traffic in the pair. At this time, it is preferable to use the path identifier independently of a time slot which is a physical division of a transmission / reception band. . That is, as shown in FIG. 33 (a), there is a case where different time slots are sent to the same partner, but these time slots sent to the same partner are treated as a group as shown in FIG. Is treated completely independently of the time slot delimiter. The group of time slots in FIG. 33B is a group of calls as shown in FIG. 33C, and each call has a path for identifying a call having a header as shown in FIG. An identifier and a channel identifier are described. Even in calls within the same time slot, the path identifier may be different. If the time slot and the path are taken independently in such a manner, for example, as shown in FIG. 34, when one of the calls is terminated and the number of time slots to be sent to the other party can be reduced, the path identifier and the channel identifier must be reset. No need to issue. Of course, if the path is no longer used due to the termination of the call, the path identifier may be returned to the network.

  In addition, the buffer configuration in the transmitter hardware of each node 2 is configured to buffer for each transmission partner as shown in FIG. 35 (not for each time slot), whereby each node can perform such buffering. It is possible to respond promptly to a decrease in time slots. That is, the input switch 116 is connected to the output switch 114 via a plurality of buffers (B1 to Bn-1) 115. The output terminal of the output switch is connected to the transmitter 113. The transmitter 113, the switches 114 and 116, and the buffer 115 are controlled by the node controller 110.

  Usually, the network also issues a channel identifier that identifies the channel in the path. Since a network such as the present invention has a very high throughput, a small call size may have a bandwidth equivalent to that of a telephone. In such a case, the number of calls existing in the network becomes very large, and it is a great amount of processing to collectively process all of them in one node. The network controller (or another node) manages the path identifiers in a unified manner on the network, but if the channel identifiers are also issued at the same time, the size of the network controller becomes very large. In addition, the amount of information exchanged therefor in a control channel or the like becomes very large, resulting in poor efficiency. Therefore, in a network such as the present invention, it is preferable that each general node 2 constituting each transmission-reception pair issues a channel identifier independently.

  As described above, since the number of calls existing in the network is enormous, the number of bits used for the path identifier and the channel identifier corresponding to the number of calls must also increase. Since they reduce the effective data rate, it is desirable to reduce the number of bits as much as possible. If the number of channels included in a single path varies and the number of paths that include only a small number of channels is large, the number of bits used for identifiers must be increased after all, and the number of bits used for identifiers must be increased. It gets bigger. Therefore, it is desirable that the path identifiers are allocated according to the number of calls in the pair. However, if the network has a very high throughput and a small decrease in data rate is acceptable, a method of allocating a path identifier corresponding to a lower network may be considered. FIG. 36 is an explanatory diagram of this, in which (a) is a logical configuration diagram of the network of this embodiment, which also shows the lower-level networks connected to each node. (B) is an extract of only the transmission from the node 1 to the node 2 from (a), and different path identifiers are given to the lower network 17, that is, the transmission from the network LNW1 to the LNW3 and the transmission from the LNW2 to the LNW3. An example is shown. This is an example in which the data is divided for each lower network of the transmission source. In addition to this, the data may be divided for each lower network of the transmission destination, or may be divided for each pair.

  Requests for these path identifiers are made by the general node as to whether a new path identifier is required. When a new transmission request from the lower network comes to the node 2, the node checks the use status of the transmission band of the own node (whether there is room for the call to be newly entered) and sends a time slot to the network controller. It is good to decide whether to make a request and a path identifier request. In other words, if there is no room for the requested call, do not make a request to the network controller, and do not notify the network controller if the call can be set up without rejecting it. It is advisable to reduce the load on the network controller as much as possible. Also, how to handle the call that could not be set (whether to wait or give up) may be determined by each node, and it is not particularly necessary to unify them in the network.

However, a time slot request may be made to the network controller and the time slot may not be set, or a time slot request may be made from another node at the same time and only one of them may be set. In that case, the concept of grading may be introduced into the network as a factor in determining to what extent the request can wait, or which one has priority. In other words, the time that can be taken from the time slot request is made until the time slot is set, and the degree of importance of the call are determined according to the grade. Priority is given to things, and things that can be waited for when they cannot be set are put in a queue.

  In this case, it is assumed that a new time slot request occurs when there is a new call setup request from the lower network, but the call setup request is not limited to only one, and simultaneously to the same partner node. It is possible that more than one call will be transmitted and that the calls will not be of the same grade. In this case, the node selects a stricter priority and selects a longer waiting time until the call setting time, and the node requests the network controller. In the latter case, if a time slot request is queued, the node will voluntarily send a call loss notification to the lower network when the shorter allowable waiting time expires. Further, if the number of requested time slots changes due to this, the fact is notified to the network controller. Also, when a time slot request is queued at the network controller, there is a possibility that a new call setting request to the same partner node comes from the lower network. At this time, in the same manner, the node selects and determines the importance, the waiting time, and the like, and if the number of requested time slots changes, notifies the network controller of the change (if any) in addition to the change.

  Furthermore, if a quality class after call setup is introduced, the total throughput can be improved by the statistical multiplexing effect. In a system that does not tolerate any information loss or delay jitter in a call, when a call setup request is issued from a lower network, the node allocates a time slot band so that its maximum bit rate is always guaranteed, and Request must be made. In this way, it is always possible to cope with any band fluctuation within the initially requested band, but for calls that do not always communicate at the maximum bit rate, most of the time the band remains. In addition, other calls cannot use the surplus bandwidth, which is inefficient. On the other hand, in the future direction of information communication, ATMs and the like are widely used, band compression technology is also advanced, and rather than transmitting at a fixed bit rate, information that does not need to be transmitted is not transmitted as much as possible and the total amount of information is reduced. It is thought that it will proceed in the direction of reduction. Therefore, in order to efficiently respond to such a call, it is necessary to use the statistical multiplexing effect as in the ATM system.

  In the network of the present invention, there is a slight delay even if the time slot is set smoothly after the time slot request is made until the time slot is set, and the time slot request is not always passed. Therefore, if the time slots are simply filled based on the time-average bandwidth of each call in anticipation of statistical multiplexing, it is not possible to immediately cope with a case in which the data is out of one time slot due to band fluctuation. For this reason, if the delay priority of the call and the permissible discard rate are determined, a new time slot for discarding a call having a high permissible discard rate even when there is a band change and it is likely to instantly protrude from one time slot. This can be handled by, for example, issuing a request and making a call with a low delay priority wait until it is set.

  However, the network of the present invention is originally a network with a very high throughput, and it is conceivable that most of the bandwidth is used in an unused state. In such a case, statistical multiplexing is forcibly performed and discarded. No need to wake up. Therefore, in such statistical multiplexing, only the quality class is determined in advance, and when there is sufficient network bandwidth, the time slot application at the maximum bit rate is performed as described above, and the network is congested, and the call is called. If loss is likely to occur, it is good to allocate bandwidth using the effect of statistical multiplexing. In this case, it is also possible to simply reduce the average bit rate, but call that never allows discarding It is advisable to secure a band at the maximum bit rate as before, and perform statistical multiplexing for calls that do not.

  In addition, when the statistical multiplexing is performed in this manner, the bandwidth may be sufficiently reduced from the average in the same manner as the bandwidth may exceed the time slot for one time slot. If a class is provided within the quality class to keep the drop rate low by allowing retransmissions with little regard to delay, when the time slot bandwidth is insufficient, the contents of calls of that class are stored in a buffer and the bandwidth is There is also a way to send when you are.

  By the way, in a normal non-broadcast time slot setting procedure, an example of a more detailed procedure of time slot arrangement and movement when a common empty time slot for transmission and reception is not found will be described with reference to FIGS. .

  First, the number of free time slots for transmission and reception is checked, and if either one has no free time slots, setting of a new time slot is impossible, so the time slot setting fails, and the network controller requests this. To the node that issued the.

  Next, if there is an empty time slot in either of the transmitting node (T1) and the receiving node (R1), whichever may be earlier, for example, attention is paid to an empty time slot Si where the transmitting node is located. The receiving node examines the party that is transmitting. If this is Ti, it is checked whether there is a common free time slot for R1 and Ti. If there is, if Si is moved to Sj (this is Sj), Si of R1 is vacant, so that a common free time slot is created between T1 and R1. The network controller notifies R1 and Ti to transfer the communication performed by Si to Sj, and notifies T1 and R1 to perform communication by Si. If no common empty time slot is found in R1 and Ti, the same procedure is performed in another empty time slot in T1. If a common empty time slot for T1 and R1 cannot be created even if this procedure is performed for all empty time slots of T1, a similar procedure is performed for empty time slots of R1. If a common free time slot cannot be created, the time slot setting has failed, and the node that has issued the time slot setting request is notified to that effect.

  In the above example, the sorting is performed by focusing on the empty time slots of the transmitting or receiving node. If the network controller of the system has sufficient capacity, or if the priority of the call that should be sent in the case of success is high, if the above procedure could not be set up, move to another place. It is also possible to create a common free time slot. In the above example, for example, it is checked in a similar procedure whether or not a common empty time slot of R1 and Ti can be created. For example, check to see if you can move to a time slot.

  Also, if the system reserves a common free time slot for broadcasting, even if a time slot for broadcasting is vacant, it will be organized as if there is no time slot, and as a result, the settings will be incorrect. Only when possible, send / receive using the broadcast time slot temporarily, and move from the broadcast time slot as soon as possible by the following procedure for organizing the broadcast time slot. And good.

  Next, an example of a detailed procedure of time slot arrangement and movement for broadcasting will be described. A procedure is performed in which a transmission / reception pair set in a time slot to be kept free for broadcasting is moved to another time slot. However, when the usage status of the time slot changes at least from the time when the normal transmission / reception pair is set to the broadcast time slot, or the procedure for organizing the broadcast is the normal call setting procedure There is no effect unless the degree of examination is broader than that.

  First, the transmission / reception pair is checked for a common free time slot. If not, a common free time slot is created in addition to the broadcast time slot in the same manner as in the normal time slot setting described above, and the pair is created. Make the move.

  In addition, since the calculation amount of the network controller differs depending on how far the time slot is checked, the degree of the check may be changed depending on the load status of the system or the processing capacity of the network controller. For example, when only searching for a common free time slot cannot be obtained, and when there is no common free time slot, it is necessary to focus on one of the free time slots and search until the other cannot be moved. For example, when the user moves to a location other than the above to search for a common empty time slot, the user can use the information appropriately.

  For a portion for performing processing such as time slot assignment in the network controller, a method of preparing a dedicated logic circuit or a method of incorporating a ready-made processor or a computer can be considered. If you choose the latter method, you need software to run the calculator. At this time, several types of software are prepared depending on how much processing (time slot arrangement, etc.) is performed. It is also possible to make it possible for a user who has introduced the system to change the system after the introduction. Alternatively, a dedicated logic circuit and a processor can be combined in a hybrid manner. For example, only a part that requires high speed in processing that is performed very frequently is made into a logic circuit, and the other part is processed by software using a processor. Also in this case, the software can be made changeable.

  Finally, a brief description will be given of a method of protecting data security in the optical communication system of the present invention. For example, the network controller 3 outputs a method of encrypting data and a control signal for switching received data given to a receiver. For example, a method of preventing each receiver from switching channels without permission can be used. In addition, the present invention can be implemented with various modifications.

1 is a schematic configuration diagram of an optical communication system according to one embodiment of the present invention. FIG. 1 is a diagram showing a configuration of a passive star network and a ring network used for a wavelength division multiplexing optical network according to the present invention. FIG. 4 is a diagram illustrating a state in which each wavelength used in the wavelength division multiplexing optical network is time-divided into a plurality of time slots. FIG. 2 is a block diagram showing a configuration example of a network controller in FIG. 1. FIG. 2 is a block diagram showing a configuration example of a general node in FIG. 1. 3 is a flowchart showing a flow of time slot allocation control by the network controller in FIG. 1. 6 is a time chart showing a specific example of transmission / reception time slot allocation control in the embodiment. 6 is a time chart showing a specific example of transmission / reception time slot allocation control in the embodiment. FIG. 9 is a schematic configuration diagram of an optical communication system having a bucket brigade node according to another embodiment of the present invention. 6 is a time chart showing a specific example of transmission / reception time slot allocation control in the embodiment. 9 is a time chart for explaining a control signal transmission / reception operation on a control channel in another embodiment of the present invention. 9 is a time chart showing a specific example of time slot assignment control in the case of a pre-assignment method according to another embodiment of the present invention. 9 is a time chart showing a specific example of time slot assignment control when performing broadcast in another embodiment of the present invention. FIG. 7 is a schematic configuration diagram of an optical communication system having a monitoring device according to another embodiment of the present invention. FIG. 4 is a diagram showing an example of an alarm signal output from the monitoring device in the embodiment. FIG. 9 is a schematic configuration diagram of an optical communication system connecting a plurality of networks according to another embodiment of the present invention. FIG. 17 is a block diagram illustrating a configuration example of a gateway in FIG. 16. FIG. 17 is a block diagram illustrating a configuration example of a gateway in FIG. 16. FIG. 17 is a block diagram illustrating a configuration example of a gateway in FIG. 16. FIG. 17 is a block diagram illustrating a configuration example of a gateway in FIG. 16. FIG. 9 is a schematic configuration diagram of an optical communication system in which a network controller according to another embodiment of the present invention is duplicated. FIG. 11 is a schematic configuration diagram of an optical communication system having a connectionless data relay node according to another embodiment of the present invention. The figure which shows the example of time slot division | segmentation in which the number of time slots in one frame is 2n or more with respect to the number n of nodes regarding the optical communication network of another Example of this invention. The figure which shows the structure of the communication network of other Example of this invention. FIG. 11 is a diagram showing examples of various time slots in a communication network according to another embodiment of the present invention. FIG. 13 is a diagram illustrating a configuration of a communication network in which a network controller is a gateway to a public network according to another embodiment of the present invention. FIG. 6 is a diagram showing a configuration of a communication network in which a network controller obtains a clock from a public network according to another embodiment of the present invention. FIG. 14 is a diagram showing a configuration of a communication network in which a network controller obtains a clock from a node of a gateway to a public network according to another embodiment of the present invention. FIG. 13 is a diagram illustrating a configuration of a node that is a part of a network according to another embodiment of the present invention and that wavelength-multiplexes a control channel. FIG. 13 is a diagram illustrating an example of a time division method of a control channel in a communication network according to another embodiment of the present invention. FIG. 11 is a diagram illustrating an example of a method of superimposing a clock signal on a communication network according to another embodiment of the present invention. FIG. 11 is a diagram illustrating a virtual path of a communication network according to another embodiment of the present invention. FIG. 11 is a diagram for explaining a method of assigning identifiers in a communication network according to another embodiment of the present invention. FIG. 14 is a diagram showing a state in which a time slot is released regardless of an identifier in a communication network according to another embodiment of the present invention. FIG. 13 is a diagram showing a configuration of a node in which an input buffer is provided for each destination node, particularly, a configuration of a transmission unit, in a communication network according to another embodiment of the present invention. FIG. 11 is a diagram illustrating a logical configuration of a communication network according to another embodiment of the present invention and illustrating a path assignment method. FIG. 2 is a diagram illustrating a basic operation of a network according to the present invention, particularly a method of using a time slot. FIG. 4 is a diagram illustrating a call blocking rate of the network of the present invention, using the number of time slots as a parameter. FIG. 11 is a diagram illustrating an example of a procedure for organizing and moving time slots in a communication network according to another embodiment of the present invention. FIG. 40 is an exemplary view for explaining an example of the time slot arrangement and movement procedure related to FIG. 39; FIG. 40 is a time chart showing a specific example of transmission / reception time slot allocation control in the embodiment in FIG. 39.

Explanation of reference numerals

1 .... wavelength multiplexing optical network 2 .... general node 3 .... network controller 4 .... star coupler 5 ... optical fiber 6 .... ring fiber 7 ... monitoring device 8 .... gateway 9 ... Reserve network controller 10 ... Relay node 11 ... Control signal receiver 12 ... Control signal transmitter 13 ... Signal processing unit 14 ... CPU
15. Memory 21 ... Data transmitter 22 ... Control signal transmitter 23 ... Data receiver 24 ... Control signal receiver 25 ... WDM multiplexer / demultiplexer 26 ... WDM multiplexer Demultiplexer 27 ... Optical switch 31 ... Gateway receiver 32 ... Signal processing unit 33 ... Gateway transmitter 41 ... Wavelength selector 42 ... Wavelength converter 51 ... Wavelength component Waveguide 52 ... Optical switch 53 ... Wavelength multiplexer 53 ... Wavelength converter 54 ... Wavelength multiplexer 61 ... Coupler 62 ... Tunable filter 63 ... Wavelength converter 64 ... Coupler

Claims (2)

A wavelength division multiplexed optical network having a plurality of different wavelength optical transmission channels;
A plurality of nodes connected via this wavelength division multiplexing optical network and performing transmission and reception using time slots obtained by dividing an optical transmission channel of each wavelength into a plurality;
A network controller that centrally controls the assignment of the time slots to the plurality of nodes;
The network controller has a spare time slot when a free time slot is newly generated or a processing capacity of the network controller so that a common reception free time slot is secured for all of the plurality of nodes. An optical fiber communication system, wherein, when there is, an assigned time slot is moved. If a transmission request other than broadcast / multicast cannot be transmitted unless a common reception available time slot is used for all of the plurality of reserved nodes, the network controller temporarily stops the transmission of the plurality of nodes. By assigning a common reception free time slot to all of the nodes and moving other time slot allocations that can be moved, communication of the allocated slot is performed except for the reception free time slot common to all of the plurality of nodes. 3. The optical fiber communication system according to claim 2, wherein the optical fiber communication system is moved to the slot of (1).

Images