JP6287621B2 - Network communication system, its master node - Google Patents

Description

  The present invention relates to a network system that has common data on each node by time-dividing a communication band and each node broadcasting its own data at the transmission timing of the local station.

  In a plant control transmission system, it is necessary for each device constituting the system to exchange a large amount of data with each other. In this regard, if a transmission method in which mutual access is performed in an event according to the occurrence of an access request by an application installed in each device is adopted, the network load depends on the application and real-time performance can be guaranteed. Can not.

  Note that the real-time property is guaranteed, for example, when the common memory data transmission process of all nodes is completed within the TS band described later, and this is repeatedly executed in a communication cycle of a fixed period T described later.

  Therefore, each device (each station; each node) is provided with a shared memory (common memory), and each node transmits its own shared memory data to all stations on the network at the update timing of its own node. There is a communication method that realizes a high-speed data exchange method by each station that receives the received data by overwriting and storing the received data in a corresponding area of its own shared memory.

  The storage area of the shared memory is divided into storage areas allocated to each station, and each node arbitrarily updates the data in the storage area for its own station as the application is executed. The data in the storage area for other stations is only referenced and not updated. Each time the data transmitted from the other station is received, the data in the storage area for the other station is updated by overwriting the received data.

  For example, Patent Document 1 discloses that an arbitrary time slot is assigned to each station, and that each station performs data transmission or the like at the timing of the assigned time slot. In Patent Document 1, each station has a cycle timer that generates a communication cycle of period T and a send timer that generates the timing of the time slot assigned to itself. In addition, the master station transmits a synchronization frame, so that the cycle timers of all stations can be synchronized.

JP 2005-159754 A

  Conventionally, a network station number as unique information of each station on the network is used to generate the transmission timing of broadcast communication. However, when the network station number is used, there are the following problems.

  That is, in the transmission band allocation method based on the network station number, the network update cycle is determined depending on the maximum station number, and the station number cannot be determined unless the transmission data amount of each station is estimated in advance.

This will be described with reference to FIG.
First, FIG. 14A shows an example of storage area allocation for each station in the common memory corresponding to the station number. In the example shown in the figure, the storage area allocation unit is 100 words (address), and the head address of the allocation area to each station is “station number × 100”. Thus, the station number of each station is manually set according to the amount of common memory data handled by each station, for example. In the illustrated example, it is assumed that the first station needs a data amount of 200 words and the next station requires 300 words, and accordingly, as shown in the figure, the first station has a station number = '0' and the next station. The station number = '2' for the next station, and the station number = '5' for the next next station. In other words, in this example, station numbers = “1”, “3”, and “4” are not used, and if there is a node to which any of these station numbers is assigned, there is a problem. Will occur.

  On the other hand, as shown in FIG. 14B, one or more time slots are assigned to each station in the communication cycle (the TS band described later). Here, it is assumed that one time slot is required for the data transmission of 100 words. Therefore, as shown in the figure, for example, for the station with the station number = '0', two time slots are allocated as shown in the figure. For the station with the station number = '2', As shown, three time slots will be assigned.

  In this way, for example, assuming that station number = 'N' (N is an arbitrary integer) is the maximum station number, and if two time slots are assigned to this node, the total number of time slots is “N + 2”. In this way, the maximum station number (+ the amount of transmission data) determines the number of time slots in the entire system, thereby determining the length of the communication cycle.

  Here, since the amount of transmission data usually depends on the application running on the station, it is often determined at the end of the system construction. On the other hand, parameters such as the above station number are often used as parameters for data access in the application, and it is necessary to decide in advance when building the application. To change the station number at the end of the system construction, create the application. The influence of is great.

  For this reason, the common memory data amount (≈transmission data amount) handled by each station is often set to be large. For example, although the station with the station number = '0' may actually need only one time slot, two time slots will be allocated as described above, and the same will be done for the other stations. Then, the maximum station number N becomes very large.

  Thus, by setting a station number larger than the original one in advance, a wasteful free band is created in the transmission band, and communication efficiency deteriorates. In order to perform allocation efficiently, it is necessary to correctly estimate the data amount of all applications in advance, but in reality it is very difficult.

  It is an object of the present invention to realize appropriate allocation of transmission bands by performing dynamic transmission band allocation according to system subscription, or to realize appropriate reassignment of transmission bands even when the master is replaced It is to provide a network communication system that can be used.

  In the network communication system of this example, each of a plurality of nodes including a master node and a slave node has a shared memory, and each node has its own node in a transmission band assigned to the own node in a fixed communication cycle. A system for transmitting shared memory data to another node has the following components.

Each slave node has subscription request means for transmitting a subscription request including a requested bandwidth amount to the master node when joining / rejoining the system.
The master node, in response to the subscription request, assigns an arbitrary transmission band in the communication cycle to each slave node, and an allocation management in which a transmission band allocation result to each node is registered Allocation management table storage means for storing the table.

  The band allocation management means allocates the transmission band with reference to the allocation management table.

  According to the network communication system and the like of the present invention, it is possible to realize appropriate allocation of transmission band by performing dynamic transmission band allocation according to the subscription to the system, or even when the master is changed, Reallocation can be realized.

It is a figure which shows an example of schematic structure of a network communication system. It is an example of band division based on the example of the prior application. It is a process flowchart figure of a master node. (A), (b) is an example (the 1) and (the 2) of transmission time band allocation information. It is a figure which shows the joining / dropping state of each node in 5 timings, and a time slot allocation state. FIG. 10 is a process flowchart (part 1) of the master node according to the second embodiment. FIG. 10 is a process flowchart (part 2) of the master node according to the second embodiment. It is a figure which shows the joining / dropping state of each node in five timings in Example 2, and a time slot allocation state. It is a figure which shows the case where it is a node type determination process and becomes a master node. It is a figure which shows the case where it is a node type determination process and becomes a slave node. It is a figure which shows the process in connection with a master change. It is a figure which shows the hardware constitutions of a node. The functional block diagram of the basic function of a node is shown. It is a functional block diagram of the node concerning this method. (A), (b) is a figure for demonstrating the conventional time slot allocation.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a diagram illustrating an example of a schematic configuration of a network communication system according to the present embodiment.

  As an example, the network communication system 10 illustrated in FIG. 1 includes a plurality of node devices 11 and one or a plurality of HUBs (hubs) 12. In the example of FIG. 1, it is assumed that there are four HUBs 12, HUBs 12-1 to 12-4. Further, there are five node devices 11 including a node device 11-1, a node device 11-2, a node device 11-3, a node device 11-4, and a node device 11-5. Node A, node B, node C, node D, and node E will be described. However, the node device 11 is described unless otherwise distinguished. Note that the number and types of node devices and relay devices, and connection methods are not limited to the illustrated example. The HUB 12 is an example of a relay device, and is not limited to this example.

  Here, in an example description to be described later, the node A is basically a master node (master station), and the nodes B, C, D, and E are slave nodes (slave stations). Henceforth, it may be described as master node A, slave node B, C, D, E, etc. However, as will be described later, when the master node A is dropped, any one of the slave nodes becomes an alternative master node. This method can also solve problems described later that occur in such a case. Details will be described later.

  In the example of FIG. 1, the communication path of the network communication system 10 is a star type having a relay device (HUB 12) between the nodes. The node devices 11 are connected to each other via a communication path. Since a delay occurs when data is relayed in each relay device (HUB 12), the communication between any node devices 11 has a larger delay as the number of relay devices increases.

In addition, although a relay apparatus shows HUB12 as an example, it is not limited to this example, For example, a router, a repeater, an optical converter etc. may be sufficient.
Each node device 11 is, for example, a programmable controller (PLC), a PC (Personal Computer), a server, an input / output (I / O) module, a drive device (for example, an inverter, a servo, or the like). Not limited to.

  Here, it is assumed that the network communication system 10 uses the shared memory system described in the background art. Therefore, each node A, B, C, D, E has a shared memory (common memory) not shown in FIG. Each node repeats transmission of its own shared memory data to all stations on the network in a communication cycle of a predetermined period T. Each node that has received this data updates the data stored in the shared memory of its own node with the received data. In this way, a data exchange system that guarantees real-time performance is realized.

  In each node, the shared memory is accessed in accordance with execution of processing of an arbitrary application. However, the latest update data in real time can be accessed by the above processing. Although the said application is an application for control of various apparatuses, such as a plant, for example, it is not restricted to this example.

  Further, the above-described time slot allocation is performed so that the transmission timings of the shared memory data of the nodes are different from each other. That is, as described above, one or more bands (time slots) are allocated to each node, which is allocated within the TS band as shown in FIG. 2, for example. The TS band is divided into a plurality of time slots. In this method, the number of time slots is determined in advance. In other words, the length of the TS band is determined in advance, and thus the length of the communication cycle is also determined in advance. That is, in this method, the length of the communication cycle (period T) does not change. In this method, time slots are dynamically allocated to the nodes. For example, the master node assigns a time slot to each node in the order of joining the network. Details will be described later.

Hereinafter, the example of FIG. 2 will be described.
The example of FIG. 2 is a band division example based on the example of the prior application (WO 2013121568 A1), for example.
Here, the prior application (WO 2013121568 A1) will be briefly described.

  In the prior application, each node first has two timers, a cycle timer and a send timer, and a predetermined period T is set in advance in the cycle timer. Thus, by causing the cycle timer to repeat the operation of restarting immediately after the timer is up, a communication cycle having the period T shown in FIG. 2 is generated. In the prior application, the communication cycle is divided into two bands of the illustrated TC band and TS band.

  In this example, the MSG band shown in FIG. 2 is further added. However, this is an example, and the MSG band may be omitted. However, in this example, an MSG band is also provided. The MSG band is mainly used in one-to-one communication. In this example, a join request (request for network participation from the slave station to the master station), which will be described later, is also message-communicated in this MSG band. The network participation request message includes a transmission source station number, a requested transmission time band number (a requested bandwidth amount (time slot number), a transmission data amount, etc.), and the like.

  Further, like the previous application, the illustrated TC band is a band for transmission / reception of a synchronization frame. Similarly to the prior application, each slave node may synchronize its own cycle timer with the cycle timer of the master node based on this synchronization frame or the like, but this is not restrictive. Note that “synchronization” in this description refers to, for example, a state in which the counter operations (start timing; phase, etc.) of the timers of the respective nodes are matched, but is not limited to this. However, the TC band is not an essential component.

  As an example, the band (time slot) may be regarded as a division of the TS band. Each node is assigned one or more time slots. In the example shown in FIG. 2, it is assumed that one time slot is assigned to nodes A, C, and E, two time slots are assigned to node B, and three time slots are assigned to node D. However, in the past, time slots were fixedly allocated in advance, but in this method, time slots are dynamically allocated during operation. In particular, time slots are assigned in order of subscription.

  In this method, the length of the TS band is fixedly determined in advance. The master node allocates a time slot and issues a subscription permission every time there is a request to join the system from an arbitrary slave node. However, if there are no remaining unallocated time slots that satisfy the request, subscription permission is not issued.

This will be described below with reference to FIG.
FIG. 3 is a process flowchart (part 1) of the master node.
4A and 4B are examples (part 1) and (part 2) of transmission time band allocation information generated and held by the master node. The master node may generate and hold only one of the examples shown in FIGS. 4A and 4B, or may generate and hold both. Prior to the description of FIG. 3, FIG. 4 will be described.

  First, in the example shown in FIG. 4A, the transmission time band allocation information 30 includes a transmission band 31, a state 32, a station number 33, and the like. The transmission band 31 is, for example, the identification number of each of the time slots, and is from “1” to “N” in the figure. In other words, the number of time slots is N. As described above, since the length of the TS band is fixedly determined in advance, the value of N is also determined in advance accordingly.

  In the state 32, for example, information indicating whether or not the time slot of the identification number of the transmission band 31 has been allocated to an arbitrary node is stored. In the illustrated example, “allocated”, “unallocated”, and the like are stored. Is done. In the case of “allocated”, the station number of the allocated node is stored in the station number 33. On the other hand, in the case of “unassigned”, for example, “NULL” or “0” is stored in the station number 33.

In the example shown in FIG. 4B, the transmission time band allocation information 40 includes a station number 41, a state 42, a transmission start time 43, a data occupation amount 44, and the like.
The station number 41 stores the station number of each node. In other words, identification information of each node is stored. The station number 41 may store the station numbers of all nodes that can be joined regardless of whether or not the system is subscribed to the system at that time, but is not limited to this example. In the case of this method, there is no restriction such as “station numbers“ 1 ”,“ 3 ”and“ 4 ”cannot be used” as in the prior art shown in FIG. You can decide to. In addition, a situation such as “incorrectly setting an unusable station number” cannot occur.

  The state 42 stores whether the node of the station number 41 is currently “subscribed”, “not subscribed”, or “dropped”. Here, “subscription” and “non-subscription” need not be explained, but “drop off” will be explained.

  That is, after joining the system once, the node may be temporarily disconnected from the system (for example, turned off, for example) for reasons such as maintenance, failure, and update. In such a case, connection to the system (for example, power ON) should be performed again. Considering that this is the case, if a node that has joined the system is considered to have been disconnected from the system (receives transmission data from this node continuously for a certain period of time or more. If not, etc.), it is assumed to be “drop off”. After that, if there is a subscription request from the same node, it is assumed that “returned”.

  When one or more arbitrary time slots have already been assigned to the node of the station number 41 (including the state of “dropped”), the transmission start time 43 identifies the first time slot of the assigned time slot group. A number is stored, and the data occupation amount 44 stores the number of assigned time slots.

  Hereinafter, the processing of FIG. 3 will be described. Here, it is assumed that the transmission time band allocation information referred to / updated at this time is the transmission time band allocation information 30 shown in FIG. 4A. However, the present invention is not limited to this example.

  Each time the master node receives a subscription request from an arbitrary slave node (step S11, YES), the master node executes the processes of steps S12 to S16. Here, the subscription request includes information such as the station number of the requesting node and the amount of transmission data (or the number of required time slots) of the requesting node. Since this process is already performed during operation, the request source clearly knows the amount of transmission data (or the number of required time slots). Therefore, more time slots are not allocated as in the conventional case, and no unnecessary bandwidth is generated.

  The master node first checks whether the station number of the requesting node overlaps with the station number of the already subscribed node by referring to the various pieces of information of the received subscription request (step S12). If the station number is duplicated (step S12, YES), the joining is not permitted and, for example, the process ends abnormally (step S15).

  Note that an abnormal response may be returned to the requesting node at the time of the above step S15, but a message addressed to the duplicate station number may not arrive correctly, so that the abnormal response may not be returned. In this case, for example, if a normal response is not returned from the master station even if a request is made a plurality of times, the subscription requesting station determines a subscription failure by itself.

  Here, each node is provided with a configuration (such as a station number SW to be described later) in which a worker or the like can manually set a station number. For this reason, there is a possibility that the same station number as that of an existing node is set for a node newly added to the system due to an operator's mistake. For example, in such a situation, the determination in step S12 is YES.

  However, even if the station number of the requesting node is registered in the transmission time band allocation information 30, the process of step S15 is not necessarily performed. That is, when the requesting node is a node that has returned from the “dropped” state, a return process (not shown) may be executed instead of the process of step S15. In this example, in the transmission time band allocation information 30, “dropped” is also stored in the state 32 as in the case of the transmission time band allocation information 40. The determination in step S12 is YES because the station number of the requesting node is registered in the transmission time band allocation information 30, and the state 32 corresponding to this station number 33 is “allocated”. Only if there is.

  On the other hand, if the station number of the requesting node is registered in the transmission time band allocation information 30, but the state 32 corresponding to this station number 33 is “dropped”, the return processing is performed. In this return processing, the transmission band 31 in one or more records whose station number 33 is the same as the station number of the request source node is allocated to the request source node. That is, the same bandwidth as that previously allocated to the requesting node (before dropping) is allocated again. Then, the state 32 in this record is changed from “dropped” to “allocated”.

  However, as will be described later, there are cases where the return processing cannot be performed substantially. For example, when the master node is an alternative master to be described later and does not have the transmission time band allocation information 30 (or when the transmission time band allocation information 30 is newly created after becoming the master), the master node has recovered from the dropout. In some cases, it is not known that the node is a node, and the request is a new subscription request. In this case, for example, a problem such as the example shown in FIG. 5 described later may occur. This will be described later.

  If the station number of the requesting node is not registered in the transmission time band allocation information 30 (step S12, NO), it is determined that the request is a new subscription request from an unsubscribed node, and the requested time slot is set. If allocation is possible, the subscription is permitted.

  That is, first, whether or not the requested bandwidth amount included in the subscription request (the amount of transmission data and the number of required time slots) is equal to or less than the current “remaining bandwidth” (such as the number of unused time slots). Is determined (step S13). Here, the initial value of the variable “remaining bandwidth” is determined in advance and registered in all nodes. In the case of the example of FIG. 4, for example, the initial value of “remaining bandwidth” = “N”.

  If the requested bandwidth exceeds the current “remaining bandwidth” (step S13, NO), a necessary time slot cannot be allocated to the requesting node, so it is determined that subscription is not possible. (Step S15).

  On the other hand, if the requested bandwidth is equal to or less than the current “remaining bandwidth” (step S13, YES), the “remaining bandwidth” is updated (“remaining bandwidth” = “remaining bandwidth” −required bandwidth) ( In step S14, the requesting node is permitted to join the network, and a predetermined process corresponding to this is executed (step S16). This predetermined process is a new registration process to the transmission time band allocation information 30, for example. This is a process of setting the status 32 to “allocated” for the records corresponding to the number of the requested bands in the records in which the status 32 is “unallocated” and storing the station number of the subscription requesting station in the station number 33. Etc. Further, the transmission request band of the record to which the allocation is made is notified to the subscription request source node together with the subscription permission. As a result, the subscription requesting station thereafter transmits its own common memory data in the time slot of the transmission band 31 within the TS band.

  The initial value of the “remaining bandwidth” is a numerical value that is uniquely determined when determining the network operation cycle, for example, and is assumed to be downloaded in advance to all devices (nodes) from a setting tool such as a personal computer. Since the network operation cycle is a parameter that can be determined independently of the application of the node, it is easy to change the cycle later.

  Here, the master node may not hold the transmission time band allocation information (30, 40). This is the case when one of the slave nodes becomes a new master node (substitute master) because the original master node has failed. It should be noted that even if the original master node returns and then becomes the master again, it is said that it has become an alternative master.

  Even if the alternative master does not hold the transmission time band allocation information (30, 40) as described above, the alternative master performs the process of FIG. 3 based on the common memory data actually received in the TS band. Can do.

  That is, first, in addition to the data used by the application, the common memory data includes the transmission source station number, the transmission start time and the transmission data amount (or the bandwidth allocated to the transmission source node (the time used) for network management. Information such as an identification number of a slot) is added (not shown). The transmission start time and the transmission data amount are data corresponding to the transmission start time 43 and the data occupation amount 44, for example.

  Therefore, even in the case of a slave node, it is possible to know which time slot is assigned to which station based on the reception timing of common memory data transmitted from each node and these additional information. Even if it is a slave node, if another node is dropped, it can be estimated that this node has been dropped because the common memory data from this node is not received.

  In this way, even a slave node can recognize a node that has joined the network or a dropped node. Further, not only the node being joined but also the dropped node can recognize the time slot assigned to that node. Therefore, when the slave node becomes an alternative master, the processing of FIG. 3 can be executed without holding the transmission time band allocation information (30, 40). In particular, even when the dropped node returns and transmits a join request, it can be recognized that the request source is the dropped node, and the same time slot as before can be allocated.

  However, for example, when another node is dropped while an arbitrary node is dropped, any node that has returned after that cannot recognize the presence of the dropped other node. The same applies to a master node in which an arbitrary node holds transmission time band allocation information (30, 40). This is because the transmission time band allocation information (30, 40) disappears due to dropping. Alternatively, even if the transmission time band allocation information (30, 40) does not disappear even if it is dropped, the time slot allocation status may change during the dropout, so the transmission time band allocation information (30, 40) 40) cannot be used. For this reason, there is no choice but to guess the time slot allocation status based on the common memory data received as described above. Since the other dropped nodes do not send the common memory data, their existence itself cannot be recognized. The above description is based on the premise that the other nodes that have dropped out have not yet returned.

For example, when the situation described above occurs, the problem described below may occur.
That is, for example, in a network applied in a plant control system, for example, in order to partially maintain the system, a part of equipment (node) is temporarily disconnected from the network, or a temporary node is used to update a control application. May be turned off. In this case, after that, by reconnecting the node or turning on the power, the network is rejoined.

  When dynamic transmission bandwidth allocation is performed according to the order of joining the network as in this method, the transmission bandwidth becomes fragmented while the transmission bandwidth allocation at the time of re-subscribing to the network is performed, and subscription is possible with the total amount of transmission data There is a case where a new station cannot participate. This will be described below using an example shown in FIG.

  In FIG. 5, the vertical axis is the time axis t, and the upper side in the figure is old and the lower side is new. FIG. 5 shows the joining / dropping status of each node A, B, C, D, E and the time slot allocation state at five timings on the time axis.

  Here, in accordance with the example of FIG. 2, it is assumed that one time is required for each of the nodes A, C, and E, two time slots are required for the node B, and three time slots are required for the node D. Node A is assigned time slot 1, Node B is assigned time slots 2 and 3, Node C is assigned time slot 4, and Node D is assigned time slots 5, 6, and 7. It is assumed that the time slot 8 is assigned to the node D.

  In the oldest state, it is assumed that nodes A, B, C, D, and E are all joined and functioning normally. Note that the node with a diagonal line in the rectangle means the master node at that time, and therefore node A is the master node in the oldest state. A node whose rectangle is a dotted line is a node that is managed as being dropped at that time. In addition, it is assumed that time slots without rectangles are managed as free bands. For example, at the latest timing shown in the figure, the time slot 3 and the time slot 8 are considered to be free bands. It is the master node at that time that performs these managements.

  In the oldest state, the master node A can grasp the time slot allocation status of each node shown by referring to the transmission time band allocation information (30, 40). However, in this situation, since all nodes normally transmit common memory data in the TS band, from the above, each of the slave nodes B, C, D, E is substantially the same as the master node A. The time slot allocation status of each node shown in the figure can be grasped. In addition, each slave node B, C, D, E records this time slot allocation status. However, since it is stored in the volatile memory, it is erased when the power is turned off. The same applies to the transmission time band allocation information 30.

  Then, at any time, it is assumed that the oldest state becomes the second oldest state. That is, it is assumed that the nodes A, B, and E are dropped and the master node is dropped, so that the node C becomes a new master node (substitute master).

  The alternative master node C does not have the transmission time band allocation information 30, but can grasp the time slot allocation status of all the nodes A, B, C, D, and E as described above. In addition, in this case, in the TS band, the common memory data from the node D is received, but the common memory data from the nodes A, B, and E is not received. Therefore, the nodes A, B, and E are dropped. I can guess. Note that the node C can also newly create the transmission time band allocation information 30 based on the estimation result of such allocation status. However, the estimation result is not always correct.

  Assume that node A has recovered in this situation. At this time, the node A transmits the above subscription request to the master node C, and the node C can recognize that there has been a subscription request from the node A that has been dropped. The assigned time slot 1 can be assigned to node A again. As a result, the third oldest state shown in FIG.

  During operation in this state, the nodes A, C, and D each transmit common memory data at the transmission timing (allocation time slot) of the local station in the TS band, so that the node A performs the time of the nodes C and D. The slot assignment status is known. However, regarding the nodes B and E, the node A does not know its existence in the first place. That is, the slave station can recognize only the information of the station that is currently transmitting valid data.

It should be noted that the node A after such dropout → returning is indicated as a node A ′ as shown in the figure to distinguish it from before the dropout.
In such a situation, when the node C is dropped and the node A ′ becomes the alternative master, the state is the fourth oldest (second from the bottom) shown in the figure.

  In this case, as described above, the node A ′ does not know the existence of the nodes B and E in the first place. For this reason, as shown in the fourth oldest state (second from the bottom) in the figure, the time slots allocated to the nodes B and E are regarded as free areas. In this state, for example, when the node E returns and issues the above subscription request, it is treated as a completely new node, so that it is not the time slot 8 originally assigned to the node E, but the head of the free area. Will be assigned. For this reason, it will be in the state shown in the latest state (lowermost) in the figure. That is, a part of the band (time slot 2) originally assigned to the node B is assigned to the node E.

  In this state, when Node B returns, if it is possible to allocate up to eight time slots, two time slots remain but are fragmented as shown in FIG. It can not be assigned. In this way, even if the node B tries to recover, there is no continuous allocatable area and it cannot be recovered.

Hereinafter, a second embodiment that solves such a problem will be described.
As described above, the slave station that does not have the transmission time band allocation information (30, 40) recognizes the station (valid station) that has received the common memory data, but once the station has joined, A station that has not received common memory data cannot recognize its presence. For this reason, when the slave station becomes a new master station, the above-described problem occurs.

  Therefore, in the second embodiment, as shown in FIG. 6A, the master node sends the transmission time band allocation information (30, 40) held by itself to, for example, all slave nodes periodically (step S21, YES). (Step S22). As a result, each slave node also holds the most recent transmission time band allocation information (30, 40). The notification process of the transmission time band allocation information (30, 40) may be performed, for example, in the form added to the synchronization frame in the TC band, but is not limited to this example. As described above, both the allocation information 30 and the allocation information 40 may be held, or only one of them may be held.

Further, the master node executes, for example, the joining process in FIG. 6B.
Here, description will be made assuming that the transmission time band allocation information 40 shown in FIG.

The process of FIG. 6B executes the process of steps S32 to S40 every time a join request from an arbitrary slave node is received (step S31, YES).
First, it is determined whether the station number of the transmission source node included in the subscription request has been registered in the transmission time band allocation information 40 (step S32). If it is not registered (step S32, NO), the processes of steps S36, S37, S39 or S40, which are the processes at the time of new subscription, are executed. Note that the processing of steps S36, S37, S39 or S40 is substantially the same as the processing of steps S13, S14, S15 or S16 of FIG. 3, and therefore description thereof is omitted here.

On the other hand, if the subscription request source node has already been registered in the transmission time band allocation information 40 (step S32, YES), first, another station has the same station number because the dropped station has returned or because of a setting error or the like. It is determined whether a subscription request has been made (step S33). That is, when the state 42 in the record (corresponding record) corresponding to the subscription request source node in the transmission time band allocation information 40 is “dropped”, it is assumed that the dropped station has been restored ( Step S33, NO), the process proceeds to step S34. On the other hand, if the state 42 in the record is “subscribed”, it is assumed that another device has accidentally requested to join with the same station number (step S33, YES), and the subscription is not permitted. It determines with a thing (step S39).

  In step S34, it is determined whether or not the number of requested time slots included in the subscription request matches the data occupancy 44 of the record. If they match, that is, if there is no change in the requested bandwidth (step S34, YES), the dropped station return process is performed (step S38). That is, the same time slot as before is assigned to the subscription requesting station. That is, a time slot indicated by the transmission start time 43 and the data occupation amount 44 in the corresponding record is assigned. Further, the state 42 of the corresponding record is changed from “dropped” to “subscribed”.

  In this way, only when it is determined in steps S33 and S34 that the dropped station has recovered and is requesting the same bandwidth as before, the band (time slot) that was originally allocated is reassigned. It will be.

  On the other hand, when there is a change in the requested transmission bandwidth (step S34, NO), the subscription is not permitted (step S39). This is a process that assumes, for example, the case where the same station number as that of a station that is dropped out is mistakenly set for a completely different new subscriber station. In other words, unless the new subscriber station accidentally has the same transmission bandwidth as that of the dropped station, the determination in step S34 is NO, thereby preventing the time slot of the dropped station from being assigned by mistake.

  Alternatively, when a device is replaced due to a failure or the like, normally, a new device (referred to as a replacement device) to be replaced with the failed device is set with the same station number as the failed device, and the same application as the failed device is installed. Since the application is the same as that of the failed device, the “required transmission bandwidth” is also the same as that of the failed device. Accordingly, in this case, the determination in step S34 is YES, and the same time slot as that assigned to the failed device is assigned to the replacement device. In other words, the replacement device can normally take over after the failed device.

  In the above example, the station number is used, but the present invention is not limited to this example. For example, in the case of Ethernet, a number (referred to as a device identification number) that can uniquely identify a device such as a MAC address may be used. However, in this case, since the device identification number is naturally different between the replacement device and the failed device, there is a possibility that the replacement device will not normally take over after the failed device.

According to the processing of the second embodiment described above, time slot allocation as shown in FIG. 7 is performed in the case shown in FIG.
FIG. 7 also shows time slot allocation statuses at five timings from the upper side in the figure in the chronological order, similarly to FIG. Since the oldest, second oldest, and third oldest timings are the same as those in FIG.

  At the fourth oldest timing (second from the bottom), the master node C is dropped as in FIG. 5 and the node A ′ becomes a new master node. However, unlike FIG. 5, the node A ′ The existence of the nodes B and E being dropped can be recognized, and the time slot previously assigned to the nodes B and E can also be recognized. This is because the node A ′ that was a slave node also holds the transmission time band allocation information 40 as described above. At this time, information on the nodes B and E should also be registered in the transmission time band allocation information 40. Of course, the state 42 should be “dropped”.

The above is based on the premise that the transmission time band allocation information 40 has been notified from the master node C by the process of step S22 at least once after the node A has returned.
Thus, in this state, when the node E returns and issues a join request in the same manner as in FIG. 5, when the node A ′ that has received this join request executes the process of FIG. 6B, the determination in step S32 is YES. Further, the determination in step S33 is NO. If the node E requests the same number of time slots (= '1') as before, the determination in step S34 is YES, and thus the process in step S38 is executed.

  Thus, as shown in the latest timing (bottom) in FIG. 7, the node A ′ assigns the previously assigned band (time slot 8) to the node E. That is, as shown in FIG. 5, a part of the bandwidth (time slot 2) previously allocated to the node B is not allocated to the node E. Thus, although not shown in FIG. 7, when there is a joining request from the restored node B, the node A ′ performs the processing of FIG. The bandwidths (time slots 2 and 3) assigned to can be assigned.

  Here, as described above, in this method, when the master node is dropped, any of the slave nodes becomes a new master node (substitute master). This will be described in detail below.

  Each node performs a process of determining whether its own node is a master node or a slave node when it is started up. Although this process itself is one application process, the flow of processing when it becomes a master is shown in FIG. 8, and the flow of processing when it becomes a slave is shown in FIG.

  First, in FIG. 8 and FIG. 9, the process immediately after the start is the same, and the process branches depending on the situation. That is, in both FIG. 8 and FIG. 9, first, resource initialization processing is performed (steps S41 and S51). The resource initialization process itself is irrelevant here, and a description thereof will be omitted. Next, in both FIG. 8 and FIG. 9, line monitoring processing is performed (steps S42 and S52). This process is, for example, a process of monitoring whether or not common memory data, a TC frame (synchronization frame), or an initial TC is received via a network after starting a monitoring timer in which a predetermined time is set in advance. is there.

  If the common memory data or TC frame from another node is received before the monitoring timer expires, execute the process of making itself a slave station, that is, the processes of steps S53, S54, and S55 of FIG. become.

  Further, even when the initial TC from another node is received before the monitoring timer expires and the transmission source node has a higher priority than the own station, the process of making itself a slave station, that is, FIG. Steps S53, S54, and S55 are executed. The initial TC and priority will be described later.

  On the other hand, if neither the common memory data nor the TC frame (synchronization frame) is received and the initial timer from the station having a higher priority than that of the own node is not received, Execute the process to make the master station. That is, the processing of steps S43 and S44 in FIG. 8 is executed. That is, the local station is temporarily set as a temporary master, and then, depending on the situation, whether to determine the local station as a master or a slave station is determined. Therefore, during this process, there may be a transition to a process in which the self is a slave station.

  In this way, the network is monitored at the time of start-up of the device, and basically operates as a temporary master when another station is not already a master. Alternatively, even when another station is operating as a temporary master but has a lower priority than the own station, it operates as a temporary master. If another station is already a master, this master station should have repeatedly transmitted a TC frame with a period T. If another station is already a master, this master station or another slave station should have repeatedly executed common memory data transmission.

Thus, as described above, when the common memory data or the TC frame is received, it is assumed that the other station is already the master, and the process of making the own station as a slave is executed.
On the other hand, when the local station is set as a temporary master as described above, predetermined processing as a temporary master is executed (step S43). That is, for example, the initial TC (ITC) is repeatedly transmitted a predetermined number of times (for example, three times) at a predetermined cycle. The initial TC includes, for example, information indicating that it is an ITC, a station number of a transmission source node, and the like. Here, the priority is higher as the station number is smaller. Thus, the other station that has received this ITC determines whether the priority is higher or lower than that of the own station by comparing the station number of the ITC with the station number of the own station.

  If an TC frame or an ITC from a station having a higher priority than the own station is received during the process of step S43, the process proceeds to a process in which the own station is a slave. On the other hand, when the ITC is transmitted for the predetermined number of times without receiving them, it is determined that the own station is the master. Then, the operation as a master is performed. That is, for example, the process of repeatedly transmitting the TC frame at a fixed period T and the process of FIG. 3 and FIG. 6B described above are performed in response to a subscription request from another station.

  On the other hand, with regard to the processing of making the self a slave station (the processing of steps S53, S54, and S55 in FIG. 9), first, it is assumed that another station has already become a master node by receiving a TC frame or the like as described above. In this case, for example, the transmission source node of the TC frame is regarded as a master node (step S53), and the join request is transmitted to the master node to enter a slave operation confirmation waiting state (step S54). ).

  Alternatively, even when an ITC having a higher priority than the own station is continuously received a predetermined number of times (for example, three times), the transmission source should start the master operation by the process of FIG. The above joining request is transmitted to the node, and the slave operation confirmation waiting state is entered (step S54).

  Then, in the state of waiting for the slave operation confirmation, when a subscription permission is returned from the master node, the operation is started as the slave station. At this time, for example, the time corresponding to the time slot assigned by the master is set in the send timer (step S55). In this case, as in the prior application (WO 2013121568 A1), the own cycle timer may be synchronized with the master cycle timer based on the transmission delay time notified from the master. The send timer and cycle timer may be the same as those of the prior application, and are not specifically described here.

Here, FIG. 10 shows processing related to the master change.
If each slave node detects the drop of the master at any time during operation (step S61) , first, the slave bit turns on the inquiry bit when transmitting the common memory data (step S62). Then, it operates as a temporary master (step S63).

  Here, regarding the process of step S61, for example, when the TC frame is not continuously received a predetermined number of times or more, it is determined that the master is dropped. Although not specifically shown, the inquiry bit is included in the common memory data frame and is normally transmitted with the bit turned OFF. Each node checks ON / OFF of the inquiry bit every time it receives common memory data from another station.

  In addition, when the TC frame is transmitted from another node before the common memory data transmission timing is detected after detecting the drop of the master, the ITC frame is transmitted from another node having a higher priority than the own node. If it has been transmitted, the process waits for master switching without performing the process of step S62 (step S64).

  Further, the process of step S63 may be the same as the process of step S43. That is, if the ITC frame can be transmitted a predetermined number of times (for example, three times), the own node is determined as an alternative master, and the operation as the master node is started (step S65).

  However, if a TC frame is transmitted from another node before transmitting the ITC frame a predetermined number of times, or if an ITC frame is transmitted from another node having a higher priority than the own node, A master switching waiting state is entered (step S64).

  When the TC frame is received in the master switching waiting state, a node other than the own node has become an alternative master, so that the operation as the slave note is resumed.

For example, the master / slave is dynamically determined by the above-described processing, and the alternative master is dynamically determined when the master is dropped.
FIG. 11 shows the hardware configuration of each node.

  Note that the master node in this example has a time slot allocation function and a function for synchronizing the period T, etc., compared to the slave node, but these functions can be configured by software. The master node and the slave node can be realized with the same hardware configuration. Therefore, FIG. 11 may be regarded as a hardware configuration diagram common to the master node and the slave node.

  In the example shown in FIG. 11, each node includes a CPU 51, a network I / F (interface) 52, a flash memory (Flash MEM) 53, an SRAM 54, an SDRAM 55, a station number SW (switch) 56, a USB I / F (interface) 57, and the like. Is provided. Note that SRAM is an abbreviation for Static Random Access Memory, SDRAM is an abbreviation for Synchronous Dynamic Random Access Memory, and USB is an abbreviation for Universal Serial Bus.

  The network I / F 52 is an interface for connecting to a network and communicating with other nodes. The USB I / F 57 is an interface for downloading network setting data from a personal computer (not shown). The station number SW 56 is a configuration for an operator or the like to manually set an arbitrary station number. The set station number is stored in the SRAM 54 or the like, and the frame number transmitted to the other node as described above includes the station number of the transmission source node.

  Software (application program) for realizing the above-described various processing functions of the node is stored in a flash memory (Flash MEM) 53 in advance. When the CPU 51 executes this software, the various processing functions described above (for example, the processes shown in FIGS. 3, 6A, 6B, 8, 9, 10, etc.) and FIGS. 12 and 13 described later are shown. Processing functions and the like are realized. In this case, for example, the configuration may be such that it is booted and executed in the SDRAM 55 or the like in order to increase the processing speed.

  For example, parameters such as the amount of transmission data (required number of time slots) are input via the USB I / F 57 using a setting tool on a personal computer (not shown), and the flash memory (Flash MEM). ) 53.

  Further still, as described above, each node, with can become either a master and a slave, because not is predetermined either, the software for the master for the slave is present. Of course, when the self is the master, the master software is executed.

FIG. 12 shows a functional block diagram of basic functions of the node.
In the illustrated example, the node includes various functional units such as a network management unit 61, a system management unit 62, a loader command server unit 63, a storage management unit 64, a communication I / F unit 65, and a SW monitoring unit 66.

  The network management unit 61 executes and controls data transmission / reception with other nodes via the network, network state management, and the like. The loader command server unit 63 manages communication with a host device such as a personal computer (not shown). Note that communication with the host device is performed, for example, via the communication I / F unit 65. The storage management unit 64 performs software booting, storage and reading of various setting values, and the like. The system management unit 62 manages these various functions in an integrated manner and manages and controls the entire node.

  Further, when the station number SW 56 is operated, the SW monitoring unit 66 notifies the station number set by the operation to the system management unit 62, for example. As a result, the system management unit 62 recognizes and stores the station number of the own node.

Note that FIG. 12 mainly shows functions related to the network operation, and the node functions are not limited to the illustrated functions, and may include other functions.
FIG. 13 shows a functional block diagram of a node related to this method.

  The network communication system according to this method is composed of a plurality of nodes each having a shared memory, and each node transmits and receives data in the shared memory of its own node to each other via the network in a predetermined communication cycle. In this system, the data stored in the memory is the same. One of the nodes is the master node 70 and the other is the slave node 80.

  Each of the slave nodes 80 includes a subscription request unit 81 that transmits a subscription request including a requested bandwidth amount to the master node 70 when joining / rejoining the network communication system.

  In response to the subscription request, the master node 70 registers a bandwidth allocation management unit 71 that allocates an arbitrary transmission bandwidth in the communication cycle to each slave node 80, and a transmission bandwidth allocation result to each slave node. An allocation management table storage unit 72 for storing the allocation management table to be stored. Then, the bandwidth allocation management unit 71 refers to the allocation management table and allocates the transmission bandwidth. This determines, for example, the bandwidth to be allocated from the free bandwidth at that time. That is, a part or all of the free bandwidth is allocated. Also, for example, basically, allocation is performed sequentially from the beginning of the free area. In other words, the assignments are made so as not to leave a gap.

For example, the bandwidth allocation management unit 71 updates the free bandwidth every time transmission bandwidth is allocated, and does not permit subscription when the requested bandwidth exceeds the free bandwidth.
For example, the assignment management table storage unit 72 also stores the current status of each registered node. Then, for example, when the master node 70 detects that an arbitrary node has dropped out of the nodes registered in the allocation management table, the state monitoring unit 73 sets the current state relating to the node as being dropped. It has further.

  In addition, for example, the bandwidth allocation management unit 71 returns the node after the drop-off when the transmission source node of the subscription request is registered in the allocation management table and the current state is being dropped. Is determined.

  Then, in response to a join request from a node that has been determined to have recovered after dropping, the band allocation management unit 71 reallocates the transmission band that has been allocated to the node and that has been registered in the allocation management table.

  Alternatively, the bandwidth allocation management unit 71 has the same requested bandwidth amount of the join request from the node that has been determined to have recovered after being dropped as the bandwidth amount of the transmission bandwidth that has been assigned to the node and that is registered in the allocation management table. In this case, the allocated transmission band is reassigned to the node.

  Further, for example, the master node further includes an allocation management information notification unit 74 that periodically transmits the allocation management table to each slave node and stores it. Then, for example, the slave node that has become a new master node in response to the drop of the master node uses the allocation management table stored in the own node by the allocation management information notification unit 74 to set the transmission bandwidth. Make an assignment.

Further, for example, each of the slave nodes may further include a master monitoring unit 82, an alternative mastering unit 83, and the like.
The master monitoring unit 82 determines whether or not the master node has dropped out. For example, as described above, when the TC frame (synchronization frame) is not received, it is determined that the master node has dropped out.

  When the master monitoring unit 82 determines that the master node has dropped, the alternative mastering unit 83 may set the own node as a new master node (substitute master) depending on the situation. For example, when the alternative mastering confirmation frame can be transmitted a predetermined number of times, the local node is set as the alternative master. On the other hand, if the alternative mastering confirmation frame is received from a slave node having a higher priority than the self node before the predetermined number of times is reached, transmission of the alternative mastering confirmation frame is stopped. That is, in this case, the own node does not become an alternative master.

  Then, when an arbitrary slave node becomes the new master node, the bandwidth allocation management unit 71 stores the bandwidth management information stored in the own node by the allocation management information notification unit 74 of the old master node. The transmission band is allocated using the allocation management table.

Further, for example, when the band allocation management unit 71 determines that the already registered node has recovered after dropping based on the subscription request and the allocation management table, the same bandwidth as before dropping is set. Assign .

  In addition, for example, the communication cycle includes a common memory band that is a band in which the nodes transmit and receive the shared memory data to and from each other, and the length of the common memory band is fixedly determined in advance. ing. The transmission band is allocated in such a manner that the common memory band is divided.

  As described above, according to the present technique, it is possible to efficiently use the transmission band in the common memory network using the time division band. For example, it is possible to realize appropriate allocation of transmission bands by performing dynamic transmission band allocation according to the order of joining the network. Furthermore, it becomes possible to cope with the drop-out and restoration of the network component equipment (node). For example, by periodically notifying each node of the bandwidth allocation information of each node held by the master node and sharing the information, even if the master node is dropped, the alternative master can obtain information on stations that have been subscribed in the past. Thus, it is possible to distinguish between the dropout recovery station and the new subscriber station and to recognize the band allocated to the dropout recovery station. As a result, it is possible to appropriately allocate the band, particularly with respect to the dropout recovery station, so that the dropout recovery can be performed correctly.

10 Network Communication System 11 Node Device 12 HUB (Hub)
30 Transmission time band allocation information 31 Transmission band 32 State 33 Station number 40 Transmission time band allocation information 41 Station number 42 State 43 Transmission start time 44 Data occupation amount 51 CPU
52 Network I / F (Interface)
53 Flash memory (Flash MEM)
54 SRAM
55 SDRAM
56 Station number SW (switch)
57 USB I / F (interface)
61 Network management unit 62 System management unit 63 Loader command server unit 64 Storage management unit 65 Communication I / F unit 66 SW monitoring unit 70 Master node 71 Bandwidth allocation management unit 72 Allocation management table storage unit 73 Status monitoring unit 74 Allocation management information notification Unit 80 slave node 81 subscription request unit 82 master monitoring unit 83 alternative master unit

Claims (11)

Each node, including the master node and slave node, has a shared memory, and each node transmits its own shared memory data to other nodes in the transmission band assigned to that node within its regular communication cycle. A system that
Each slave node has a subscription request means for transmitting a subscription request including a requested bandwidth amount to the master node when joining / rejoining the system,
The master node is
Band allocation management means for allocating an arbitrary transmission band in the communication cycle to each slave node in response to the subscription request;
An allocation management table storage means for storing an allocation management table in which a transmission band allocation result to each node is registered ;
Allocation management information notification means for periodically transmitting and storing the allocation management table to each slave node ;
The band allocation management means refers to the allocation management table, a network communication system and performs allocation of the transmission band.   2. The bandwidth allocation management unit updates a free bandwidth amount every time the transmission bandwidth is allocated, and does not permit subscription when the requested bandwidth amount exceeds a free bandwidth amount. Network communication system. The allocation management table storage means also stores the current state of each registered node,
When the master node detects that any node among the nodes registered in the allocation management table has dropped out , the master node further includes state monitoring means for dropping the current state relating to the node;
The bandwidth allocation management means determines that, when the transmission source node of the subscription request is registered in the allocation management table and the current state is being dropped, the node has recovered after dropping. The network communication system according to claim 1 or 2, characterized in that:   The bandwidth allocation management means reallocates the transmission bandwidth allocated to the node registered in the allocation management table in response to the join request from the node determined to have recovered after the drop. The network communication system according to claim 3, wherein:   The bandwidth allocation management means is configured such that the requested bandwidth amount of the subscription request from the node determined to have recovered after the drop is registered in the allocation management table, and the bandwidth amount of the transmission bandwidth allocated to the node 4. The network communication system according to claim 3, wherein if the same transmission frequency is the same, the allocated transmission band is reassigned to the node. 5. The slave node that has become the alternative master node that is the new master node in response to the drop of the master node uses the allocation management table stored in the own node by the allocation management information notifying means, and uses the transmission bandwidth. the network communication system according to any one of claims 1 to 5, characterized in that the assignment. Each slave node is
Master monitoring means for determining whether or not the master node has dropped;
When it is determined by the master monitoring means that the master node has dropped, it further comprises alternative mastering determination means for determining whether or not the own node is the alternative master node,
The bandwidth allocation management means allocates the transmission bandwidth using the allocation management table stored in the own node when the alternative node is determined as the alternative master node by the determination of the alternative mastering determination means. The network communication system according to claim 6 . When the bandwidth allocation management means determines that the already registered node has recovered after dropping based on the subscription request and the allocation management table stored in the own node, the same transmission as before dropping 8. The network communication system according to claim 7 , wherein a bandwidth is allocated. The alternative mastering determination unit determines that the own node is the alternative master node when the alternative mastering confirmation frame can be transmitted a predetermined number of times, but has a higher priority than the own node before the predetermined number of times is reached. network communication system according to claim 7 or 8, wherein when receiving the substitute master of confirmation frame from the slave node and judging not to said alternate master node its own node. The communication cycle includes a common memory band that is a band in which the nodes transmit and receive the shared memory data to and from each other.
The length of the common memory band is fixedly determined in advance,
The transmission band, a network communication system according to any one of claims 1 to 9, characterized in that assigned in the form of dividing the common memory bandwidth. A system that consists of a plurality of nodes each having a shared memory, and each node transmits its own shared memory data to other nodes in a transmission band assigned to that node within a regular communication cycle, In the master node in the network communication system in which one of the nodes is a master node and the other is a slave node,
Band allocation management means for allocating an arbitrary transmission band in the communication cycle to each slave node in response to a subscription request transmitted when each slave node joins the system;
An allocation management table storage means for storing an allocation management table in which a transmission band allocation result to each node is registered;
An allocation management information notification means for periodically transmitting and storing the allocation management table to each slave node;
A master node characterized by comprising: