JP2009004855A - Communication system and communication method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a reliable communication system and communication method that copes with a permanent fault. <P>SOLUTION: It is decided from the ID of a fragment header whether the fragment packet generated from the same source data is already received or a first received packet (S3). When the ID of the fragment header is already registered (NO), the fragment packet is discarded (S7) and then a state where a fragment packet is expected to be received (S14) is entered. Since a standard fragment function holds the packet ID which is already received, collation therewith is performed to decide whether the packet is a first packet (YES) or a following packet (NO), and the ID thereof is registered (S4) when the packet is the first packet (YES). When the ID of the fragment header is already registered (NO), and when the packet is the following packet (NO), the registered ID is deleted and the packet itself is a redundant part, the packet is discarded (S7), and then the fragment packet is expected to be received (S14). <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a communication system (hereinafter also referred to as “communication method”) and a communication method that are highly reliable and suitable for remote control.

  A plurality of nodes transmit data to each other via a network using the Internet Protocol (IP) as a conventional communication system (hereinafter also referred to as “IP network”, “field network”, or “Internet”). It is possible to receive. Such communication methods are being applied to applications that have strict requirements for transmission delays, data loss, and the like, as the performance of the devices improves. Also in industrial applications, attempts have been made to transmit control information using an IP network as an information transmission means in the field of plant management and control, so-called process control.

In the process control information transmission, all control information needs to be transmitted within a certain time period called a control period. For this reason, a delay exceeding the control period and a lack of information are not allowed. However, in the case of an IP network, a repeater such as a hub is connected to form a network. If this hub or cable fails, there is a drawback that a permanent communication failure occurs.
In addition, a temporary communication failure occurs due to poor contact of the cable connector or external noise.
Therefore, in the process control that requires a strict level, various ideas have been made to avoid these.

  On the other hand, in the field of consumer use, problems related to instability of real-time transmission due to uncertainties of IP networks are being discussed in applications such as video distribution using the Internet. As a countermeasure, it is known that a communication method combining an existing IPv6 multicast protocol and a forward error correction control method (FEC: Forward Error Correction) is effective in realizing reliable large-scale multicast communication. (For example, Non-Patent Document 1).

  In this communication method, the method of dividing IP packet data after making it redundant and distributing and transmitting in time axis is particularly effective as a countermeasure against intermittent packet loss found in unstable networks such as the Internet. It is said that it is. This countermeasure against packet loss is positioned as an attempt to increase the reliability of the IP network, as in the industrial field.

  A conventional example will be described with reference to FIGS. In each country, for convenience of explanation, the same reference numerals 1 to 5 are attached to nodes 1 to 5 and hubs (hereinafter referred to as “HUB”) 1 to 5 which are different functions.

  FIG. 11 is an explanatory diagram of a loop configuration by a HUB connected in a ring shape and switching control by STP in a conventional IP network. It is explanatory drawing which shows the conventional IP network, and is explanatory drawing of the loop structure by the HUB connected in the ring shape, and the switching control by STP. As shown in FIG. 11, when an IP network is connected in a ring shape, data transmitted from the HUB to the HUB like a relay is transmitted in a loop. In order to avoid this, a technique called IEBE802.1D Spanning Tree Protocol (STP) is used.

  HUBs connected to the network contact each other's connection status and break the loop. A network that is physically connected in a loop is logically separated according to an algorithm. In the network shown in FIG. 11, since HUB1 is set as a root bridge, a loop is cut by automatically setting a block point between HUB3 and HUB4.

In the present invention, the repeater of the IP network is represented by the most commonly used HUB. However, in the STP, this is also referred to as a bridge in terms of function.
When the STP algorithm is simplified, when there are a plurality of routes to the root bridge as seen from a certain bridge, the shortest route is selected and the ports connected to the other routes are blocked. In the example of FIG. 11, the route to HUB1, which is the root bridge for HUB3, is determined based on whether it passes through two HUBs toward the left or one HUB toward the right. As a result, the right direction is selected and the left direction is blocked.

  When the network fails, the HUB adjacent to the failure point detects this and notifies all the remaining HUBs that the status has changed. For example, when the HUB 2 fails, the downstream HUB 3 notifies the state change, and the block point moves between the HUB 3 and the HUB 1. As a result of this reconfiguration, the HUB 3 can continue communication on the opposite route.

  Since the redundant network using the ring topology and STP is simple as described above, it is frequently used in office LAN wiring or the like, but this method has a drawback. As indicated by HUB2 in FIG. 11, generally, a large number of nodes are connected to the HUB. For this reason, if one HUB fails, the nodes connected to the HUB become unable to communicate at the same time. It cannot be used for applications that want to minimize disruption of communication due to failures, such as industrial applications.

  FIG. 12 is an explanatory diagram of a communication system that copes with a path failure by using two complete networks using a node having two network interfaces according to a conventional example. Node 1 and node 2 communicate with each other via a five-stage HUB, and each of the five-stage HUBs has a pair of backup machines and wiring between the stages is also independent of each other.

  The two networks of the A side and the B side are in a standby state when one is used and does not contribute to communication. When communication is performed on the A side, if the HUB3-A fails as shown in FIG. 12, it is the node 1 and the node 2 that detect the failure, and each node uses the network interface used so far. And the B side interface is used.

  According to the communication system of FIG. 12, the nodes connected to the network equipment are not disabled at the same time due to the failure of the network equipment. However, it is necessary to detect a failure in the path by either judging whether each node is in the state of data communication or constantly monitoring the operating state of all related HUBs. As a route switching method, a method of switching all nodes at the same time is common, but notification to all nodes and synchronization of switching timing are not easy.

As described above, according to the conventional communication system shown in FIG. 12, although a highly reliable system can be configured, there is a drawback that a burden on the node is increased. Further, although this is a defect common to the conventional communication system shown in FIG. 11, there is also a defect that the time required for path switching becomes long.
FEC onIPv6 for Reliable Multicast Nara Institute of Science and Technology, Seijiro Yoneyama / Hideki Sunahara

  In industrial IP networks, ring topology networks in which repeaters are arranged in a ring or double repeaters for all repeaters are used to avoid disruption of transmission paths due to failure of repeaters such as hubs. There is a standby redundant type duplex network or the like in which one is in an operating state and the other is in a standby state, and the entire route is duplexed.

  In the ring topology network, when a repeater, that is, a hub fails, communication between other hubs is ensured by turning the communication path back and forth between the hubs before and after the failure. However, a plurality of nodes connected to the failed hub have a drawback that the influence of the failure is large because all communication is interrupted.

  Therefore, as in the case of the standby redundant type duplex network or the like, if the path is duplexed with all the repeaters, one failure of the repeater will not lead to communication interruption of a large number of nodes. A higher reliability system can be constructed.

  However, no matter which method is adopted, it takes a certain amount of time to detect a failure of the repeater and switch the data transmission path. In the process control information transmission, all control information needs to be transmitted within a certain time period called a control period, and it is difficult to switch the route in a short time to meet this requirement. Further, the data staying in the failed repeater is lost when the repeater fails. Furthermore, conventional methods cannot deal with accidents in which data is lost due to accidental noise such as lightning strikes and switching arcs.

  The present invention has been made in view of the circumstances as described above, and is a communication system and a communication method having both resistance to a transient failure and avoidance of a communication failure due to a permanent failure of network equipment. Highly reliable data transmission that is not affected by network failures is realized by distributing to multiplexed communication paths instead of the time axis, while maintaining the advantages of using redundancy, division, and fragment function, which is the standard of IP networks. An object of the present invention is to provide a communication system and a communication method that can be used.

  In order to solve the above problem, a communication system according to claim 1 is a communication system including a plurality of nodes connected to a multiplexed IP network, and each of the nodes is a multiplexed IP network. A multi-terminal for connection corresponding to the number of multiplexes, and an IP communication processing unit that is electrically and logically connected to the multi-terminal for connection inside each of the nodes. The communication processing unit redundantly encodes and divides the transmission data, and then forms each divided data as a fragment data part of a fragment packet, and a transmission function that distributes and sends the data to multiple paths of the multiplexed IP network; The fragment packet received from the multipath is reconstructed, and if it is redundantly coded data, it is decoded to the original data. A redundant function created from the transmission data is encoded with a redundancy according to the degree of multiplexing of the IP network and the number of packets allowed and transmitted by being divided into a plurality of packets. Thus, when a packet loss occurs on the communication path, the packet is reconstructed up to the number of defects corresponding to the redundancy.

  The multiplexed network path is obtained by multiplexing all the repeaters such as hubs and routers forming the network and the cables connecting the repeaters. Since the node is connected to the multiplexed repeater, it has connection terminals as many as the number of multiplexed terminals, and each terminal and the repeater are connected separately. Accordingly, a completely independent network is formed in parallel for the number of multiplexed communication paths between nodes.

  The node has an IP communication processing unit (also referred to as an IP stack) for processing IP, forms data inside the node into an IP packet, advances to the network, interprets the IP packet received from the network, and extracts the data Take a role. This IP communication processing unit is logically and electrically connected to a considerable number of external terminals for connection to a multiplexed network path, and a transmission packet is selected by the IP communication processing unit by selecting one of the external terminals. Sent out. On the other hand, all received packets from the multiplexed network path are delivered to the IP communication processing unit and sequentially processed.

  When the IP communication processing unit intends to transmit the data inside the node as an IP packet, the data is redundantly encoded and divided. The divided data is further formed as an IP packet in the form of a fragment packet, and is distributed and transmitted to the multiplexed network path. The data thus made redundant, divided, and sent out as fragment packets are received by the destination node.

  The received packet arrives from the multiplexed network route with different routes and times. However, since the IP communication processing unit accepts all packets received from the multiplexed network route, there is no problem even if the route is distributed. . For the distribution of arrival times, IP packets in the standard fragment format already correspond and are processed in the same way. Furthermore, if the minimum number of packets that can decode the redundancy code can be received, the IP communication processing unit does not wait for an unarrived fragment packet and reproduces the original data at that time.

  According to the communication system according to the first aspect, since the original data is made redundant and divided into packets to form communication packets, the original data can be restored even if the number of packets corresponding to the redundancy is lost. In addition, since it is spatially distributed over the multiplexed network route, it is not necessary to switch the route even when the route fails. Even when the route is controlled for other reasons, the switching time is not limited. Furthermore, there are no problems with intermittent failures or transient noise failures. As described above, according to this communication system, a highly reliable communication system that can reliably transmit data even when a communication path has a permanent failure or a transient failure has occurred. Can provide.

  A communication system according to claim 2 is the communication system according to claim 1, wherein the node has one terminal for connecting to a LAN device in addition to a multi-terminal for connecting to a multiplexed network. The IP communication processing unit is also electrically and logically connected to the LAN device connection terminal, has a function of processing IP communication with the LAN device, and can connect the general-purpose LAN device to the multiplexed network. .

  According to the communication system according to claim 2, a general-purpose LAN device that does not have an interface for connecting to the multiplexed network path is connected to the multiplexed network path, and communication with high reliability is performed as in the communication system according to claim 1. Can provide a system.

  A communication system according to claim 3 is the communication system according to claim 1, wherein the node includes one terminal for connecting to a serial communication device in addition to a multi-terminal for connecting to a multiplexed network, and an IP The communication processing unit is electrically and logically connected to a serial communication device connection terminal, has a function of processing protocol conversion with the serial communication device, and can serially connect the serial communication device to a multiplexed network.

  According to the communication system according to claim 3, a serial communication device that does not have an interface for connecting to the multiplexed network path is connected to the multiplexed network path, and communication with high reliability is performed as in the communication system of claim 1. A system can be provided.

  The communication method according to claim 4 is a communication method corresponding to a duplex IP network at a node, wherein redundancy is repetition of the same data as the original data, the number of fragments is 2, The data part of the fragment packet is the same as the original data itself, the offset of the fragment header is set to “0”, the M flag is set to “0”, and the transmission part of the node separates the two fragment packets into a duplex IP network. And the receiving unit of the node determines that the fragment packet received from the duplex IP network is a redundant fragment packet if the offset packet is “0” and the M flag is “0”. To determine whether to proceed to this step, or to proceed to the standard fragment processing step. The step of determining whether the fragment packet composed of the original data is a subsequent packet that has already been received or the first received fragment packet by the ID of the fragment header, and if it is the first fragment packet, register the ID Restoring the original data, discarding the packet if it is the subsequent packet, deleting the registered ID, and waiting for reception of the next new fragment packet.

  A communication method according to claim 4 is a communication system corresponding to an IP network duplexed by the nodes. The redundancy method of the node is repetition of the original data, the number of fragments is 2, the data part of the two fragment packets is the same original data itself, the offset of the fragment header is “0”, and the M flag is “ The redundant fragment packet format is determined to be “0”.

The transmission unit of the IP communication processing unit collects the original data in the redundant fragment packet format, forms two fragment packets, and distributes them separately to the duplex network for transmission.
When the reception unit of the IP communication processing unit receives a fragment packet from the duplex network, first, whether the offset and the M flag are both “0” or not, whether the packet is a redundant fragment packet, Determine whether it is a standard fragment packet. If it is a standard fragment packet, it is processed according to the procedure of the standard fragment packet. If it is a redundant fragment packet, the original data is restored by the following steps.

1) It is determined by comparing the ID of the fragment header with the received ID registered in the node whether a fragment packet made from the same original data has already been received or the first received packet.
2) If it is the first fragment packet, the ID is registered and the original data is restored. If it is a subsequent packet, the packet is discarded and the registered ID is deleted.
3) Wait for reception of the next new fragment packet.

  According to the communication system of the fourth aspect, the fragment format remains the IP standard format, and the network can be configured with existing IP network equipment. Further, since the redundancy method is repetition of the original data, and a complicated arithmetic operation is not required, the processing load can be reduced. And since the communication system is completely duplexed, even if a failure occurs in a repeater or a cable constituting the network, communication can be performed without any trouble on the remaining route. Further, since redundant data is always distributed and communicated in a redundant route, even if a packet is lost due to a communication failure due to external noise or the like, the remaining packet passing through another route is guaranteed.

  As described above, according to this communication system, it is possible to reliably transmit data even when a communication path has a permanent failure or a transient failure, and a standard IP A highly reliable communication system that can be easily configured in a network can be provided.

In the communication method according to claim 5, in the node in the communication method according to claim 4, when one of the duplex networks fails, the restriction to send the two fragment packets to the normal side, or normal The packet transmission operation is restricted by selecting any one of the restrictions on sending the fragment packet only on the side.
That is, this is a communication method that enables transmission control of redundant fragment packets when a path failure occurs with respect to the transmission unit of the IP communication processing unit in the node.

According to the communication method of the fifth aspect, the node distributes redundant fragment packets in two paths. When one of the two routes fails, select whether to continue sending to the failure side as it is, or to automatically switch to the normal side without using the failure side and send all packets to the normal side only The IP communication processing unit is configured so as to be able to. In other words, when a path fails, the mode can be switched from spatial dispersion of redundant packets to time dispersion.
According to this communication method, even if there is only one remaining communication path, redundancy is effective against accidental packet loss, and high reliability can be ensured even at the time of failure.

According to the communication system and the control communication method of the present invention, the following effects can be obtained.
1) Even if a part of a multiplexed network path fails permanently, it can be restored by data received via another normal redundant path. For this reason, it is not necessary to deal with the failure by switching the route. Depending on the required level of the system, it can be dealt with by adjusting to the data redundancy according to the required level. As the simplest example, there is an example of adjusting the redundancy that allows one packet to be lost by duplicating the network and simultaneously sending one copy of the data packet.

2) Restoration of data according to redundancy is guaranteed, so even if the system needs to switch routes for some reason, there is no need to switch quickly, and no special equipment or techniques are required for high-speed switching. It is. According to the present invention, since there are such effects, the present invention can be applied to an application having a severe tolerance even for a slight delay time.

3) Although accidental packet loss could not be dealt with by failure detection and path switching function, the failure of various factors can be remedied by distributing redundant data to multiple paths. It becomes possible. For example, it is possible to cope with intermittent failures caused by poor contact of cables. In addition, information loss due to external noise can be remedied, and packets that are lost due to failure because of stagnation inside the failed repeater can be remedied.
As a result of such an effect, if a method of switching the route distribution of the redundant packet to time distribution and sending it to the normal route when a route failure occurs, a communication system that is resistant to accidental packet loss can be configured even when the route failure occurs .

4) Since redundant data such as redundant codes and redundant packets can be generated by simply copying the original data before transmission, the processing is simple. Therefore, it can be applied to a simple node.
5) Since redundant data is generated using the standard format of the fragment as it is, it can coexist with the standard IP network. Further, it does not affect the upper protocols TCP, UDP, ICMP and the like. In addition, since the IP network device can be used for the hub and router constituting the network, the communication system can be easily changed and expanded, and the operation management and maintainability are excellent.

6) Redundancy of data using fragments can be realized only by making a small change to an IP stack used in a general IP communication node. Since the implementation is performed on the IP layer separated from the application layer of the node, the IP stack corresponding to redundancy can share the same resource with any device as in the general IP stack.
7) Since a multiplexed network path in which repeaters are multiplexed is used, there is no inconvenience that many nodes cannot communicate at the same time due to a failure of one repeater.
8) If the redundant node is changed slightly, a general-purpose LAN device or serial communication device can be used as a dedicated repeater that connects the multiplexed network path, so that the general-purpose device can be easily used even in a highly reliable network.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is an explanatory diagram showing a transmission process of a packet transmission unit in a communication system (hereinafter also referred to as “present system”) related to an embodiment of the present invention (hereinafter referred to as “present embodiment”). In the transmission processing shown in FIG. 1, transmission processing for packet loss countermeasures by adding a Reed Solomon code and packet division using a fragment mechanism is executed.

  FIG. 2 is an explanatory diagram showing reception processing of the packet reception unit in the communication system related to the present embodiment. In the reception process shown in FIG. 2, the reception process of the packet loss countermeasure by the addition of the Reed Solomon code and the packet division using the fragment mechanism is executed.

  In the transmission / reception processing shown in FIG. 1 and FIG. 2, “when the number of packets fragmented from the original data is n and the number of error correction codes assigned thereto is m, the total number of packets received by the receiving node is n or more. It is operating based on the principle that “restoration starts when n packets can be received and the original data can be restored.” This means that data can be restored even if m packets are lost.

  In other words, the method of dividing IP packet data after making it redundant and distributing and transmitting on the time axis has been proposed as an effective method for intermittent packet loss seen in an unstable network such as the Internet. ing. However, it does not guarantee the high availability required for industrial networks.

  In addition, there is a need for a failure avoidance method that can cope with a permanent failure of a relay device or media constituting a network. 1 and 2 are described in detail in the paper “FEC on IPv6 for Reliable Multicast” by Nara Institute of Science and Technology, which is a non-patent document 1, so further explanation is omitted. To do.

  This system combines the transmission / reception processing of FIG. 1 and FIG. 2 with the multiplexed network, and further distributes the fragmented packets to the multiplexed network path for transmission. That is, it is a communication system that has both a resistance against a transient failure and a function of avoiding a communication interruption due to a permanent failure of network equipment by distributing and transmitting redundant packets to a plurality of paths (space axes).

  FIG. 3 is an explanatory diagram showing transmission processing of a packet transmission unit that generates transmission data applied to the IPv6 fragment scheme in the communication system according to the present embodiment. In FIG. 3, fragment processing of IPv6 is generated by dividing fragmentable data shown in the packet structure into a necessary number and adding a fragment header. The following procedure will also be described with reference to FIG.

  FIG. 4 is a flowchart showing a procedure for restoring a redundant fragment packet when it is received via the network in the communication system according to the present embodiment. In the procedure shown in FIG. 4, the multiplexing of the network is duplex, and the redundancy of data is illustrated by repeating the same data simply, but the multiplexing is not limited to this. Redundant encoding of data or the like may be employed.

  In the case of triple or more multiplexing, redundant data encoding, etc., “If the number of packets fragmented from the original data is n and the number of error correction codes added is m, the number of packets not received by the receiving node is less than m. If the number of multiplexed networks is K, when the total number of packets n + m is evenly distributed over the K system networks, the condition under which data restoration is possible even if one system network fails is 1 Since the number of packets in the system only needs to be m or less, it is as follows.

(N + m) / K ≦ m
Looking at the ratio of the original data n and the correction code m,
n / m ≦ K-1
If K = 2, that is, a duplex network,
n / m ≦ 1
If K = 3, that is, a triple network,
n / m ≦ 2
It becomes.
In the communication system shown in FIGS. 3 and 4, both the original data n and the correction code m are 1 and K is 2.

  In this system, the fragmentable portion is not divided from the original data of the packet structure 31 but is copied as it is to generate two fragment-type packet structures 32 and 33. In general, a Reed Solomon code is generated and used as the data of the second fragment. However, in the case of duplication, there is no need to use such complicated processing. At this time, the fragment header has the contents shown in the packet structure 34. The main point is that the offset value is "0" and the M flag value is "0".

  The offset value is an index indicating where the divided fragment data is located in the original data. The offset “0” indicates that it was the top data. Further, when the M flag is “1”, it indicates that there is subsequent divided data, and when it is “0”, it indicates that it is the last data. Therefore, the offset “0” and the M flag “0” are both the first data and the last data.

This is a pattern that cannot be a normal fragment packet. In this system, this unused “impossible format” is used as a guideline to distinguish normal fragment packets from redundant fragment packets.
The “impossible format” is within the standard fragment as a packet format. Therefore, the redundant fragment packet can interoperate with the standard IP network.

  A transmitting unit (not shown) separately transmits the same two fragment packets having this packet structure 34 to two paths of the duplex network. Thereby, spatial danger dispersion | distribution can be aimed at.

  When a redundant fragment packet is received from the network (S1) according to the flowchart shown in FIG. 4, a procedure for restoring it (S5, S10) will be described. The left half of FIG. 4 shows a redundant fragment restorer (S2 to S7), and the right half of FIG. 4 shows a standard fragment restorer (S8 to S13). First, it is determined whether the received packet is a standard fragment or a redundant fragment (S2). As described above, if the offset is “0” and the M flag is “0”, the means determines that the packet is a redundant fragment packet (YES), and proceeds to the next process (S3).

  In the process (S3), it is determined by the ID of the fragment header whether a fragment packet made from the same original data has already been received. This function is the same as a standard fragment packet. Since the received packet ID is registered, if it is unregistered and the first packet (YES) in comparison with this, the ID is newly registered (S4). Since this example is duplexed, the original data can be restored with the first packet. The restoration process is immediately performed (S5), the original data is returned to the IP layer (S6), and the process waits for reception of the next fragment packet (S14).

  If it is determined in the process (S3) that the ID has been registered and the subsequent packet (NO), the packet is discarded (S7) at the same time as the ID is deleted, and the process waits for the next reception (S14).

  If it is determined in step (S2) that it is a standard fragment packet, it is determined in the next step (S8) whether or not a fragment packet made from the same original data has already been received by the ID of the fragment header. When it is not registered (YES), a new ID is registered at the same time, and a 60-second timer is set (S9), and when it is registered (NO), the packet restoration process (S10) is executed as it is.

  The standard fragment packet has an indefinite number of fragments, and the number is not transmitted to the receiving side. The packet restoration process (S10) is a process of filling between the first packet and the last packet with the offset and the M flag as clues. When the data is collected from the beginning to the end, the packet restoration completion determination (S11) is completed (YES), ID deletion and timer stop (S12) are performed, and the restored original data is returned to the IP layer (S13). . If the restoration is not completed (NO), the process waits for the next fragment packet to be received (S14).

  To supplement ID registration management, there is no problem if all fragment packets arrive without delay for ID deletion. However, if there is a loss in the communication path, registration deletion will not be completed, including restoration processing, and this will be organized at the end. Processing is necessary. The standard fragment uses a 60-second timer and forcibly deletes the registration ID if it cannot be restored within 60 seconds.

  However, in the redundant fragment, depending on the situation of the path failure, there is a possibility that lost packets occur frequently and memory resources are wasted. Aside from the timer mechanism, an upper limit of the number of registered IDs is set, and when the upper limit is exceeded, the old registration is deleted in order (not shown).

In this system, the following effects can be obtained by transmitting the redundant fragment packet on the transmission side via the network and restoring the original data on the reception side.
1) Since data is made redundant according to the multiplicity of the network and distributed and transmitted over the multiplexed network path, the data can be restored even if one path of the network is permanently broken.
2) In a duplex network, a method of generating duplex data by using the standard specification of a fragment as it is can be used, so that it can coexist with a standard IP network. The node configuration can also be realized by slightly changing a general IP stack.

3) Furthermore, the redundant code can be generated by a simple copy instead of generating the redundant code by the mathematical formula as in the conventional method, and the processing is simple.
4) Data redundancy using fragments can be realized only by making a small change to the IP stack used in a general IP communication node. Since it is separated from the application layer of the node and implemented in the IP layer, the IP stack corresponding to redundancy can share the same resources with any device as in the general IP stack.

  FIG. 5 is a block diagram showing a communication system in which a node with a data redundancy function (hereinafter abbreviated as “node”) according to the present embodiment is combined with a duplex network. As shown in FIG. 5, a duplex network is configured by a first route (hereinafter also referred to as a route) 10 on the A side and a second route (hereinafter also referred to as a route) 20 on the B side.

  In this duplex network, the HUB, which is a relay for communication data inserted in the route 10 and the route 20, respectively, and the cable connecting them are completely duplicated. The communication path is ensured. Nodes 1 and 2 are also duplexed nodes corresponding to the duplicated network connections.

  In general, there are a large number of nodes connected to the IP network, and the HUBs forming the network are interconnected in a tree form (see FIG. 10). Only one node and a HUB that relays communication data between them are drawn and simplified.

  The nodes 1 and 2 with the data redundancy function are equipped with an IP stack 6 for performing IP communication processing, and the IP stack 6 is a redundant fragment for processing the redundant data shown in FIGS. Block 7 has been added. In FIG. 5, since it is a duplex network, the redundancy system shown in FIGS. 3 and 4 will be described as an example.

  In the redundant fragment block 7 of the nodes 1 and 2, the transmission data is converted into two redundant fragment packets as transmission processing and distributedly transmitted to the paths 10 and 20. As a reception process, the redundant fragment packet received from the paths 10 and 20 can be restored to the original data evenly from either side.

  As shown in FIG. 4, when both fragment packets arrive, the first packet is restored and the subsequent packets are discarded. Therefore, even if a packet is lost on either of the routes 10 and 20, data communication is established on the other side.

  The arrows in FIG. 5 indicate the packet flow. It can be seen that the fragment packet is sent out from the node in two paths. Further, FIG. 5 shows a state in which a failure has occurred in the middle of the route 20 on the B side (cable connection between HUB1-B and 2-B). As the arrow indicating the packet on the route 20 is interrupted before the failure point, the communication data is interrupted here, but the other fragment packet has arrived on the route 10, and the communication data continues to be restored without delay. . The same applies even if the failure is on the path 10 side.

  As described above, according to this communication system, since the duplex network is used evenly, even if a route failure occurs, there is no need to further switch the route.

The main points of the communication system shown in FIGS. 1 to 5 are as follows.
1) In a node connected to a multiplexed network, the transmission unit fragments the transmission data by the number corresponding to the redundancy required for the system. It is generated by the transmitter as a packet encoded redundantly so that data can be reconstructed. In addition, this transmission unit has a function of distributing fragment packets over a multiplexed network path. The number of multiplexed network paths in the network and the number of fragmented packets need not match.

  The reception unit of the node has a connection end with all the multipaths of the redundant network and can receive all the distributed fragment packets. When data is reconstructed, even if a packet loss occurs due to a network path failure, the data format is redundantly encoded. Therefore, the original data can be restored up to the number of defects corresponding to the redundancy.

  When the receiving unit receives the minimum number of packets that can restore the original data from the fragment packets that are received separately, the receiving unit discards the remaining packets upon reception. In this "Waiting for Unnecessary Packets" operation that is simply discarded, the management of "Waiting for Unnecessary Packets" sets an upper limit for the number, and when "Waiting for Unnecessary Packets" occurs beyond the upper limit, the oldest ones are sequentially Terminate. This termination process can be overlapped with the timeout function of the standard fragment packet.

2) The above node corresponding to a duplex network, the number of fragments is 2, and the following configuration.
B) Fragment the original data into 2 packets.
M) The redundancy method is repetition of the same data. For this reason, the data of the two fragment packets are the same.
C) Set the offset of the fragment header to “0” and the M flag to “0”.
D) The two fragment packets are distributed and sent separately to the two routes of the duplex network.

  FIG. 6 is an explanatory diagram showing a data flow when a path failure occurs in both the A path and the B path in the node with the data redundancy function and the duplex network with the path control function according to the present embodiment. The duplex network shown in FIG. 6 represents a configuration in which five stages of HUBs are connected to relay data between the two nodes 1 and 2.

FIG. 7 is an explanatory diagram showing a situation when the HUB nearest to the node fails in the node with the data redundancy function and the redundant network with the path control function according to the present embodiment together with FIG.
The duplex network shown in FIGS. 6 and 7 has a configuration in which five stages of HUBs are connected to relay data of two nodes 1 and 2. The HUBs forming the IP network are generally interconnected in a tree shape (see FIG. 10), but in this figure, for the sake of explanation, only the HUBs that relay the data of the nodes 1 and 2 are simplified. .

  The operation of the duplex network with a path control function will be described with reference to FIG. Compared with the duplex network of FIG. 5, in FIG. 6, a pair of HUBs constituting the duplex are connected to each other by one boat. Five paths of HUBs are inserted in each of the A path 10 and the B path 20, and HUB1-A and HUB1-B, HUB2-A and HUB2-B, HUB3-A and HUB3-B, HUB4-A and HUB4 -B, each pair of HUB5-A and HUB5-B are connected to each other by one boat to form a detour.

  All the HUBs constantly monitor the connection with the adjacent HUBs to detect abnormalities in the paths 10 and 20 (x marks), and the HUBs that have detected the abnormalities bypass packets to the normal path side through the detour. In FIG. 6, the detour is used as a path for detouring the failure location when the HUB3-A or the B-side cable fails (x mark). If the paths 10 and 20 are normal, this detour is in a standby state and does not contribute to data transmission. Only when used as a detour, it is used as a data transmission path.

  The physical layer used in the IP network is mostly Ethernet (registered trademark), but the HUB, router, and node connected to it have a link detection function as a standard function, and can detect the quality of connected devices and cables in real time. . It is easy for the HUB adjacent to the failure location to detect the failure of the paths 10 and 20 based on this and use the detour.

  As shown in FIG. 6, HUB3-A fails in path 10, and cable connection between HUB1-B and HUB1-B fails in path 20. Here, in the transmission packet of the node 1, since the tip of the HUB1-B is broken on the route 20 side, the route is detoured to the HUB1-A and joined to the route 10 side. Since the HUB 8 -A has failed in the packet of the merged path 10, the packet detours from the HUB 2 -A to the HUB 2 -B, and then reaches the node 2 through the path 20 side.

  Since the HUB 8 -A on the path 10 side has a failure, the transmission packet of the node 2 detours from the HUB 4-A to the HUB 4-B and joins the path 20 side. The joined packet has a failure at the tip of HUB2-B, so it bypasses HUB2-A and reaches node 1 through HUB1-A.

  As described above, according to the duplex network with path control, communication can be continued while following available paths even in the case of simultaneous failures at a plurality of locations occurring in both duplex paths.

  The nodes 1 and 2 shown in FIG. 6 are both nodes with a data redundancy function described with reference to FIGS. According to the redundancy method using this, fragment packets are distributed and sent to the route 10 and the route 20, and even if the fragment packet of one route is lost, the communication is not interrupted.

  For this reason, it is possible to realize a highly reliable communication system that can withstand failures of various causes, ranging from device failures to transient packet loss, coupled with fault tolerance of route control using detour routes. it can. In addition, since communication is not interrupted even during path switching, it is possible to use even equipment with a slow switching speed, so that a system can be configured at low cost.

  FIG. 7 shows an example in which HUB1-B has failed in the communication system having the same configuration as FIG. At this time, the node 1 needs to detect the failure of the HUB1-B. However, in the IP network, it is the same as the HUB, and the node also has a link detection function as a standard feature. Can be detected. The node 1 can be configured to stop the transmission fragment packet on the path 20 side against this failure. In this case, the traffic on the route 10 side can be suppressed.

  It is also possible to send all fragment packets to the path 10 side. According to this method, since data redundancy is maintained, the performance of the highly reliable communication system can be maintained as it is. Whether to stop the transmission fragment packet on the path 20 side or to send all fragment packets to the path 10 side can be selected according to the request of the communication system.

  FIG. 8 is an explanatory diagram showing a configuration in which a commercially available general-purpose LAN device 9 is connected to the duplex network according to the present embodiment. As shown in FIG. 8, the redundant adapter 8 is used as a dedicated repeater. The redundant adapter 8 has one connection end for LAN device connection and two connection ends for a duplex network.

  The packets sent from the LAN device 9 are redundantly encoded, fragmented into a plurality of packets, and distributedly sent to the duplex network. Conversely, the redundant fragment packet received from the network is reconfigured by the redundant adapter 8 and delivered to the LAN device 9. Further, when a non-redundant packet is received, it is passed unconditionally and delivered to the LAN device 9.

  Note that the redundancy adapter can select the application according to the type of packet when the packet sent from the LAN device 9 is made redundant and fragmented. For example, UDP applies redundant fragmentation, but TCP passes as it is. The packet to be passed at this time can be sent to one side of the duplex network or can be sent alternately.

  FIG. 9 is an explanatory diagram showing a configuration to which a serial communication interface such as RS232C according to the present embodiment is applied. As shown in FIG. 9, a redundant adapter 13 is used as a dedicated repeater for connection to a duplex network. The redundant adapter 13 has one connection terminal for serial communication and two connection terminals for a duplex network, and a protocol conversion block 14 for interpreting the protocol of the serial communication device is arranged on the serial communication side, and the redundant adapter 13 is duplexed. The network side is constituted by an IP stack incorporating a redundant fragment block 15.

  The operation of the redundant adapter 13 shown in FIG. 9 is the same as that of the redundant adapter shown in FIG. 8, and realizes mutual conversion between the communication protocol of the serial communication device (node) 2 and IP and generation of redundant fragments. Send, receive and restore.

The main points of the invention disclosed along FIGS. 6 to 9 are as follows.
1) A communication system composed of an IP network having a duplex route and a plurality of nodes connected thereto. The node has two terminals connected to the duplex path. The transmission unit redundantly encodes transmission data and fragments it into a plurality of packets.

  In addition, it has a function of distributing and transmitting fragment packets to a duplex network. When receiving the redundant fragment packet distributed in the duplex network, the receiving unit restores the original data from the redundantly encoded data. In an IP network having a duplex route, one port of the duplex HUB is connected to each other to form a detour.

  When a HUB that forms a duplex path detects a failure in an adjacent HUB or cable, the HUB bypasses the data on the failure side to the normal side using this detour. Although the detour is in a standby state that does not contribute to communication when it is normal, a monitoring function such as failure detection for the detour itself is always continued.

2) In the communication system described in the above 1), a dedicated repeater is inserted to make a general LAN device a duplex network. This dedicated repeater has one connection end for LAN device connection and two connection ends for a duplex network. As a result, the packets transmitted from the LAN device are redundantly encoded, fragmented into a plurality of packets, and distributedly transmitted to the duplex network.

  Conversely, the fragment packet received from the network is reconstructed by this repeater and delivered to the LAN device. The dedicated repeater can select the application according to the harvest of the packet when the packet from the LAN device is made redundant and fragmented. For example, UDP applies redundant fragmentation, but TCP passes as it is. The packet to be passed at this time can be sent to one side of the duplex network or can be sent alternately. Also, packets that have just arrived from the duplex network and are not made redundant are passed unconditionally.

3) A communication system in which a dedicated repeater is inserted to connect a serial communication device such as RS232C or a wireless data communication device to a duplex network in the communication system described in 1) above. The dedicated repeater has the same function as the repeater described in 2) above in addition to the function of mutually performing protocol conversion of serial communication, wireless data communication and IP communication.

  FIG. 10 is an explanatory diagram of a duplex network shown as a reference example. The duplex network of the present invention is applied to a field network in which a control station (DCS) and a field instrument of a process control system are connected by an IP network.

  The right side of FIG. 10 shows the entire field network. In order to connect one control station and several hundred field instruments, switching HUBs (Switches) are connected in a tree shape. Each switching HUB is duplicated to form a duplicated network. Further, as shown in the left side of the portion surrounded by the two-point steel wire in the middle of FIG. 10, the connection between the HUBs is configured, and the state in which the path is controlled with respect to the failure is illustrated.

  The duplex network according to the present invention is intended for a field network in a factory and does not assume a wide area network such as the Internet beyond a router, but data redundancy is also being used for the Internet. If an environment where multiple routes can be used by multiplexing access points to the Internet, the possibility of continuous use is fully conceivable.

  In the above-described embodiment, the operation procedure shown or the shapes and combinations of the constituent members are merely examples, and various modifications can be made based on process conditions, design requirements, and the like without departing from the gist of the present invention. Is possible.

It is explanatory drawing which shows the transmission process of the packet transmission part in the communication system relevant to one Embodiment (this embodiment) of this invention. It is explanatory drawing which shows the reception process of the packet receiver in the communication system relevant to this embodiment. It is explanatory drawing which shows the transmission process of the packet transmission part which produces | generates the transmission data applied to the fragment scheme of IPv6 in the communication system which concerns on this embodiment. 4 is a flowchart showing a procedure for restoring a redundant fragment packet when it is received via a network in the communication system according to the present embodiment. It is a block diagram which shows the communication system which combined the node which concerns on this embodiment with the duplex network. It is explanatory drawing which shows the data flow when a path | route failure generate | occur | produces in both the A path | route and the B path | route in the redundant network with a data redundancy function which concerns on this embodiment, and a redundant network with a path control function. It is explanatory drawing which shows the condition when the HUB nearest to a node fails in the node with a data redundancy function and a duplex network with a path control function according to the present embodiment. It is explanatory drawing which shows the structure which connected the commercially available general purpose LAN apparatus to the duplex network which concerns on this embodiment. These are explanatory drawings which show the structure which applied serial communication interfaces, such as RS232C, to the duplex network which concerns on this embodiment. It is explanatory drawing of the duplication field network shown as a reference example. In the conventional IP network, it is explanatory drawing of the loop structure by HUB connected in ring shape, and the switching control by STP. It is explanatory drawing of the communication system which respond | corresponds to a path | route failure by using two systems of networks completely using the node with two network interfaces concerning a prior art example.

Explanation of symbols

  1, 2 ... node, 10 ... first route, 20 ... second route

Claims (5)

A communication system composed of a plurality of nodes connected to a multiplexed IP network,
Each node is connected to each of the multiplexed IP networks by a multi-terminal for connection corresponding to the number of multiplexing, and an IP that is electrically and logically connected to the multi-terminal for connection inside each node. A communication processing unit,
The IP communication processing unit encodes and divides transmission data redundantly, forms each divided data as a fragment data portion of a fragment packet, and distributes and transmits the divided data to multiple paths of the multiplexed IP network And a receiving function for reconstructing the fragment packet received from the multipath and decoding the original data if the data is redundantly encoded,
The redundant data generated from the transmission data is encoded with redundancy according to the degree of multiplexing of the IP network and the allowable number of missing packets, and is divided into a plurality of packets to be transmitted on the communication path. A communication system characterized by reconfiguring packets up to the number of deficiencies according to redundancy when a deficiency occurs. In the communication system,
The node includes one terminal for connecting to a LAN device in addition to a multi-terminal for connecting to a multiplexed network,
The IP communication processing unit is also electrically and logically connected to the LAN device connection terminal and has a function of processing IP communication with the LAN device.
The communication system according to claim 1, wherein the general-purpose LAN device can be relay-connected to the multiplexed network. The node has one terminal for connecting to a serial communication device in addition to a multi-terminal for connecting to a multiplexed network, and the IP communication processing unit is electrically and logically connected to a serial communication device connection terminal. It also has a function to process protocol conversion with serial communication devices.
The communication system according to claim 1, wherein the serial communication device can be relay-connected to the multiplexed network. A communication method for a duplex IP network at a node,
Redundancy is a repetition of the same data as the original data,
The number of fragments is 2,
The data parts of the two fragment packets are the same as the original data itself,
Set the fragment header offset to “0”, the M flag to “0”,
A transmitting unit of the node separately distributes the two fragment packets to a duplex IP network and sends them out;
The receiving unit of the node determines that the fragment packet received from the duplex IP network is a redundant fragment packet if the offset is “0” and the M flag is “0”, and proceeds to the following step. Determining whether to proceed to a standard fragment processing step;
Determining whether a fragment packet composed of the same original data is a subsequent packet that has already been received or a fragment packet that has been received first, based on the fragment header ID;
If the first fragment packet, the ID is registered and the original data is restored, and if it is the subsequent packet, the packet is discarded, the registered ID is deleted, and the reception of the next new fragment packet is waited. A communication method comprising:   In the node, when one of the duplex networks fails, packet transmission is performed by selecting either restriction for sending the two fragment packets to the normal side or restriction for sending the fragment packet only for the normal side. The communication method according to claim 4, wherein the operation is restricted.

Images