JP2012015895A - Communication device and packet relay method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a communication device which relays a packet requiring real-time processing with a low delay.SOLUTION: A communication device 1 of the invention relaying a plurality of kinds of packets input from the outside comprises: an RT packet arrival time prediction part 17 predicting an arrival time of a first packet which is input periodically and requires relay processing with a low delay; a queue 12 accumulating a second packet which does not require relay processing with a low delay; and a transmission scheduler 15 determining a packet to be transmitted to a communication channel based on the prediction result by the RT packet arrival time prediction part 17 and the length of the packets retained by the queue 12.

Description

  The present invention relates to a communication apparatus that multiplexes a plurality of types of packets having different allowable transmission delay times and outputs the multiplexed packets to a communication path.

  In a communication device that multiplexes a plurality of input packets and outputs them to a communication path, packet multiplexing control (in what order packets are transmitted) is called a scheduling method. A conventional scheduling method is described in Non-Patent Document 1, for example. In this method, packets are temporarily stored in a queue and then read out from the queue in accordance with priority.

  Further, for example, in the case of priority queuing, which is one of scheduling schemes, packets are divided according to priority, a queue is provided for each priority, and the packets are read from the packets in the queue with high priority and transmitted. For a communication device that multiplexes and transmits packets, consider the case where a queue with high priority, medium (medium), normal (normal), and low (low) is provided. When a packet is input, the priority of the input packet is identified by the data type or the like, and is written in a priority queue corresponding to the identification result (packet priority). Further, when transmitting a packet, the communication device reads and transmits the packet in the high queue with the highest priority. Packets in the medium queue are read and sent if there are no packets in the high queue. Similarly, packets in the normal queue and the low queue are read and transmitted when there is no higher priority packet (when there is no packet in the higher priority queue). By this mechanism, a packet with high real-time property (hereinafter referred to as a real-time property packet) is transmitted with a lower delay than other packets by assigning it to high.

Cisco Systems, Inc. Next Generation Router Working Group, “Cisco ISR Router Textbook”, Impress Publishing, March 1, 2006, p.86-94

  In the above-described conventional communication apparatus, as a scheduling method for multiplexing packets, a control method is adopted in which packets are temporarily stored in a queue and then read from the queue according to priority. Real-time packets are transmitted with lower delay than other packets by increasing the priority. However, when a real-time packet arrives at the communication device and is accumulated in the high-priority queue, if the packet is being transmitted from the low-priority queue, the real-time packet is finally read from the high-priority queue after transmission of this packet is completed. Sent out and sent. That is, the real-time packet is accumulated in the queue from the start of reading of the packet from the low priority queue until the transmission is completed, and this accumulation time becomes a relay delay. As an example, when a 1500-byte packet is transmitted at a speed of 10 Mb / s, the maximum delay is 1.2 ms. For example, this delay becomes a control processing delay in the relay of control data in an instrumentation control system that performs plant control, leading to a problem that fine control cannot be performed.

  The present invention has been made in view of the above, and in a communication device that multiplexes a plurality of input packets and outputs them to a communication path, a communication device that relays packets that need to be handled preferentially with low delay, and The purpose is to obtain a packet relay method.

  In order to solve the above-described problems and achieve the object, the present invention is a communication device that relays a plurality of types of packets input from the outside, and requires periodically input and low-delay relay processing. Arrival time predicting means for predicting the arrival time of the first packet to be transmitted, packet storing means for storing the second packet not requiring relay processing with low delay, prediction results by the arrival time predicting means, and Transmission packet determination means for determining a packet to be transmitted to the communication path based on the length of the packet held by the packet storage means.

  According to the present invention, it is possible to relay a packet that requires real-time processing with low delay.

FIG. 1 is a diagram illustrating a configuration example of a packet transmission system to which a communication device according to the present invention is applied. FIG. 2 is a diagram illustrating an internal configuration example of the communication apparatus according to the first embodiment. FIG. 3 is a diagram illustrating an internal configuration example of the RT packet arrival time prediction unit according to the first embodiment. FIG. 4 is a diagram illustrating a pseudo code example illustrating the operation of the transmission scheduler. FIG. 5 is a diagram for explaining the operation of the communication apparatus according to the first embodiment. FIG. 6 is a diagram illustrating an internal configuration example of the communication apparatus according to the second embodiment. FIG. 7 is a diagram illustrating an internal configuration example of the RT packet arrival time prediction unit according to the second embodiment.

  Embodiments of a communication apparatus according to the present invention will be described below in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

Embodiment 1 FIG.
FIG. 1 is a diagram illustrating a configuration example of a packet transmission system to which a communication device according to the present invention is applied. This packet transmission system includes a communication device 1 according to the present invention, a data generator 20 that generates real-time data, a data generator 21 that generates non-real-time data, and a data processing device that is a communication partner of the communication device 1. 30, the communication device 1 and the data processing device 30 communicate via the communication network 200.

  The data generator 20 generates data to be processed in real time as a communication packet. For example, a device that accommodates a sensor (not shown) for control in the instrumentation control system corresponds. The data generator 21 generates data that does not require real-time processing as a communication packet. For example, a device that accommodates a monitoring device (not shown) in the instrumentation control system corresponds. Although FIG. 1 shows an example in which there are two data generators 21, one or three or more data generators 21 may be used. In the following description, a communication packet (data to be processed in real time) generated by the data generator 20 is expressed as an RT packet, and a communication packet generated by the data generator 21 (data that does not need to be processed in real time). Is expressed as an NRT packet.

  The communication device 1 multiplexes data (RT packet, NRT packet) input from the data generator 20 and the data generator 21 and transmits the multiplexed data to the communication network 200. The data generator 20 and the communication device 1 are connected via a communication path 101, and each data generator 21 and the communication device 1 are connected via communication paths 102a and 102b. The communication device 1 is connected to the communication network 200 via the communication path 103. The communication paths 101, 102a, 102b, and 103 may be wired communication paths or wireless communication paths.

  The data processing device 30 is a device that processes data output from the data generators 20 and 21 and multiplexed by the communication device 1. The data generators 20 and 21 and the data processing device 30 are installed remotely via the communication network 200. When the data processing device 30 processes the data generated by the data generators 20 and 21, the data processing device 30 issues a control instruction corresponding to the processing result to a control target device (not shown). For fine control in the data processing device 30, the data generator 20 periodically transmits a value of a sensor (not shown) accommodated to the data processing device 30.

  FIG. 2 is a diagram illustrating an internal configuration example of the communication device 1 illustrated in FIG. 1. In FIG. 2, the same reference numerals as those in FIG.

  As illustrated, the communication device 1 includes an input I / F unit 11, a queue 12, a queue management unit 13, a multiplexing unit 14, a transmission scheduler 15, an output I / F unit 16, and an RT packet arrival time prediction unit 17.

  In the communication apparatus 1, the input I / F unit 11 receives a packet input from the communication paths 101, 102 a and 102 b, specifically, processes in the physical layer, identification of the beginning of the packet, and tail of the packet Identify.

  The queue 12 is a buffer for temporarily storing NRT packets received from the corresponding data generator 21 (the generator 21 connected via the communication path 102a or 102b). FIFO operation for outputting the first packet is performed.

  The queue management unit 13 manages packets (NRT packets) accumulated in the queue 12, and indicates the accumulation state of NRT packets (presence / absence of NRT packets) and the length of the stored first NRT packet in the transmission scheduler 15 The NRT packet is read from the queue 12 according to the instruction of the transmission scheduler 15 and output to the multiplexing unit 14. The notification to the transmission scheduler 15 is made, for example, when the state of the queue 12 changes (when a packet is newly written or when a packet is read out). Notification may be made in response to a request from the transmission scheduler 15.

  The RT packet arrival time prediction unit 17 measures the period (RT packet arrival period) at which an RT packet is input from the communication path 101, and predicts the next time when the packet is input. Further, the prediction result is output to the transmission scheduler 15.

  The multiplexing unit 14 selects either the NRT packet (NRT # 1, NRT # 2) input from each queue or the RT packet (RT) input from the input I / F unit 11 connected to the communication path 101. And output to the output I / F unit 16. The selection is instructed from the transmission scheduler 15 and executed in units of packets.

  Based on the prediction result in the RT packet arrival time prediction unit 17 and the information on the length of the NRT packet notified from each queue management unit 13, the transmission scheduler 15 instructs the queue management unit 13 to read out the NRT packet and the multiplexing unit 14. Packet output instruction for. Specifically, it is determined whether or not to transmit an NRT packet, and when it is determined that transmission is to be performed, the queue management unit 13 is instructed to output the NRT packet to be transmitted from the queue 12 to the multiplexing unit 14. Do. Further, the multiplexing unit 14 is instructed to select and output an RT packet or an NRT packet read from any queue 12. The transmission scheduler 15 performs scheduling so that the arrival timing of the RT packet input periodically does not collide with the execution timing of the transmission processing of the NRT packet. In other words, it is predicted whether or not an RT packet will arrive during execution of the NRT packet transmission process, and if it is determined that it will arrive, it is decided not to start the NRT packet transmission process and the next RT packet arrives Wait to do. As a result, the delay time required for relay processing of NRT packets that do not require real-time processing increases, but RT packets that require real-time processing can be relayed with low delay. The timing for performing the scheduling is, for example, a point in time when it is detected that an RT packet or an NRT packet is newly input. Further, the state of the queue 12 is confirmed when transmission of the RT packet or NRT packet is completed, and scheduling is performed if an NRT packet exists.

  The output I / F unit 16 performs processing in the physical layer and transmits a packet.

  FIG. 3 is a diagram illustrating an internal configuration example of the RT packet arrival time prediction unit 17. The RT packet arrival time prediction unit 17 includes a packet detection unit 171, an up counter 172, and a down counter 173.

  In the RT packet arrival time prediction unit 17, the packet detection unit 171 detects the head of the RT packet input from the data generator 20. When detected, a pulse is output to the up counter 172. The up counter 172 is a counter that counts (increments) a value every clock (not shown), and resets the count value when a pulse indicating the start detection of the RT packet is input from the packet detection unit 171. The count value is output to the down counter 173. The down counter 173 is a counter that decrements the value every clock. When a pulse is input from the packet detection unit 171, the count value of the up counter 172 at that time (the count value immediately before resetting) is changed to the count value. Preset. The count value of the down counter 173 is output to the transmission scheduler 15.

  FIG. 4 is a diagram illustrating a pseudo code example illustrating the operation of the transmission scheduler 15. The transmission scheduler 15 determines whether the time until the arrival of the RT packet is the maximum on the line C1 shown in FIG. 4 based on the information input from the RT packet arrival time prediction unit 17. If it is the maximum, the operation of the line C2 is selected (as a scheduling result), and if it is not the maximum, the line C3 is further determined.

  Here, the time until the arrival of the RT packet is obtained from the count value of the down counter 173 notified from the RT packet arrival time prediction unit 17 by the following equation.

(Time to arrival of RT packet)
= Count value of down counter 173 × clock cycle of down counter 173

  Whether the time until the arrival of the RT packet is the maximum can be determined by whether the count value of the down counter 173 is preset. That is, it shows a case where an RT packet has arrived.

  In line C2, the multiplexing unit 14 is instructed to select and output RT (RT packet). In the line C3, the time for reading the head packet in the queue 12 from the queue (head packet reading time) is compared with the time until the arrival of the next RT packet. The leading packet read time is obtained from the length of the leading packet in the queue 12 notified from the queue management unit 13 by the following equation.

(First packet read time)
= Length of first packet (bytes) x Time to read 1 byte from queue

  Returning to the description of the line C3, the operation of the line C4 is selected when the read time of the leading packet is larger than the time until the arrival of the RT packet, and the operation of the line C6 is selected otherwise. This determination (determination of line C3) is performed for each queue 12. Here, the determination of the line C3 is established when the transmission of the NRT packet is started and the next RT packet arrives before the transmission is completed. Therefore, the operation of the line C4 that is selected when the determination of the line C3 is established does not transmit the NRT packet in the queue (does not start the transmission operation). On the other hand, in the operation of the line C6, the multiplexing unit 14 is instructed to select and output NRT (NRT packet) #n. Here, n indicates the number of the queue 12 that satisfies the condition for performing the operation of C6 in the determination of C3. Therefore, the NRT packet #n indicates an NRT packet written in the queue 12 having the number n. When the conditions are satisfied in a plurality of queues 12, an instruction is issued to select and output the first NRT packet in any queue. This selection method may be performed by any method such as round robin or selecting a packet having a larger number of packets stored in the queue 12.

  Next, the operation will be described. FIG. 5 is a diagram for explaining the operation of the communication apparatus 1, and an example of a correspondence relationship on the time axis between the packet input from the communication paths 101, 102 a, and 102 b and the packet output to the communication path 103. Is shown. In FIG. 5, lines 51, 52, and 53 indicate the time flow of the input packet from the communication path 102b, the time flow of the input packet from the communication path 102a, and the time flow of the input packet from the communication path 101, respectively. Show. A line 54 indicates the time flow of the output packet from the output I / F unit 16 (see FIG. 2) to the communication path 103. In the figure, (A) to (C), (a), (b), and (1) to (5) indicate packets. In each packet, the left end is the beginning of the packet and the right end is the end of the packet. The RT packets input from the data generator 20 (see FIG. 1) are the packets (1) to (5) indicated on the line 53, and are periodically input at the time interval α. The NRT packets input from the data generator 21 are the packets (A) to (C), (A), (B) indicated on the line 51 or 52, and input asynchronously (non-periodically). Is done. t1 to t12 indicate times.

  Hereinafter, the detailed operation of the communication apparatus 1 will be described with reference to FIG. It is assumed that no NRT packet is stored in any queue 12 at the time t1 shown in FIG.

  At time t1, the input packet (1) from the data generator 20 is input to the multiplexing unit 14 and the RT packet arrival time prediction unit 17 via the input I / F unit 11. Accordingly, in the RT packet arrival time prediction unit 17, the packet detection unit 171 detects the head of the RT packet, and outputs a pulse indicating packet detection (RT packet arrival) to the up counter 172 and the down counter 173. As a result, the up counter 172 resets the counter value, and the down counter 173 presets the counter value. In this preset, a counter value corresponding to a cycle α (= maximum value of time until arrival of RT packet) indicating an interval of arrival times of RT packets is preset in the down counter 173, and the next RT packet is sent from the data generator 20 to the next RT packet. A count value indicating the time until is input from the down counter 173 to the transmission scheduler 15. In this example, the down counter 173 outputs a count value indicating the time until the time t3 when the packet (2) is input to the transmission scheduler 15. Further, when the transmission scheduler 15 detects that the RT packet has arrived, it performs scheduling according to the operation according to FIG. The arrival of the RT packet is detected by monitoring whether the count value received from the down counter 173 is preset, for example. At time t1, the count value input from the RT packet arrival time prediction unit 17 is the maximum (the time until RT packet arrival is the maximum), so the transmission scheduler 15 executes the process of C2 according to the determination of the line C1 in scheduling. Decide on. That is, it instructs the multiplexing unit 14 to select and output an RT packet. Based on this instruction, the multiplexing unit 14 transmits the input packet (1), which is an RT packet, to the data processing device 30 via the output I / F unit 16.

  After the packet (1) is input, the packet (A) that is an NRT packet arriving from the data generator 21 is written into the queue 12 via the input I / F unit 11. When the last writing of the packet (A) is completed at time t2, the queue management unit 13 outputs information indicating the packet length and the packet existence of the packet (A) to the transmission scheduler 15. Receiving this information, the transmission scheduler 15 starts scheduling, first determines the line C1 in FIG. 4, and the count value input from the RT packet arrival time prediction unit 17 is not the maximum (until the RT packet arrives). Next, the line C3 is determined. In this example, since the determination of the line C3 is not established, that is, the reading of the NRT packet (packet (A)) from the queue 12 can be completed before the next RT packet (packet (2)) arrives, the transmission scheduler 15 further executes processing of lines C5 and C6. As a result, the transmission scheduler 15 instructs the multiplexing unit 14 to select and output the NRT packet # 2. In addition, the transmission scheduler 15 instructs the queue management unit 13 to read a packet from the queue 12. The queue management unit 13 causes the NRT packet # 2 to be output from the queue 12 to the multiplexing unit 14 in accordance with the instruction. The multiplexing unit 14 transmits the packet (A) that is the NRT packet # 2 input from the queue 12 via the output I / F unit 16 in accordance with the instruction from the transmission scheduler 15.

  Note that since the writing of the packet (A) to the queue 12 is not completed when the transmission of the packet (1) is completed, the scheduler 15 determines that no packet exists in any queue 12. Judgment is not performed and scheduling is not performed.

  Thereafter, the packet (2) from the data generator 20 is input to the multiplexing unit 14 and the RT packet arrival time prediction unit 17 via the input I / F unit 11 at time t3. Accordingly, the RT packet arrival time prediction unit 17 resets the counter value of the up counter 172 and presets the counter value of the down counter 173. At this time, the count value input from the RT packet arrival time prediction unit 17 is the maximum (the time until RT packet arrival is the maximum), so the transmission scheduler 15 is the same as when the packet (1) described above is input. This process is performed, and the multiplexing unit 14 is instructed to output the RT packet. The multiplexing unit 14 transmits the packet (2) via the output I / F unit 16.

  At the time when the transmission of the packet (2) is completed, there is no packet in any queue 12, so the scheduler 15 does not perform scheduling.

  Thereafter, the packet (B) that is an NRT packet arriving from the data generator 21 is written into the queue 12 via the input I / F unit 11. When the last writing of the packet (B) is completed at time t4, the queue management unit 13 outputs information indicating the packet length of the packet (B) and the existence of the packet to the transmission scheduler 15. Receiving this information, the transmission scheduler 15 starts scheduling, first determines the line C1 in FIG. 4, and since the count value input from the RT packet arrival time prediction unit 17 is not the maximum, the line C3 Judgment is made. In this example, the determination of the line C3 is established, that is, the packet (B) from the queue 12 cannot be read before the packet (2) which is the next RT packet arrives. It is decided not to read (B). In this case, the transmission scheduler 15 does not control the queue management unit 13 and the multiplexing unit 14 (does not issue a packet read instruction, selection / output instruction).

  Thereafter, the packet (C) following the packet (B) arrives from the data generator 21, and the packet (3) from the data generator 20 arrives at time t5 while writing to the queue 12 is in progress. Accordingly, the RT packet arrival time prediction unit 17 resets the counter value of the up counter 172 and presets the counter value of the down counter 173. In this case, since the count value input from the RT packet arrival time prediction unit 17 is the maximum, the transmission scheduler 15 performs the same process as when the packet (1) and packet (2) are input, and the RT packet Is output to the multiplexing unit 14. The multiplexing unit 14 transmits the packet (3) via the output I / F unit 16.

  When the transmission of the packet (3) is completed at time t6, the transmission scheduler 15 performs scheduling and determines the line C1 in FIG. 4 and then the line C3. In this case, the determination of the line C3 is not satisfied, that is, the packet (B) can be read from the queue 12 before the packet (4) which is the next RT packet arrives. The processes C5 and C6 are executed. That is, the transmission scheduler 15 determines to transmit the NRT packet # 2, and instructs the multiplexing unit 14 to select and output the NRT packet # 2. In addition, the transmission scheduler 15 instructs the queue management unit 13 to read a packet from the queue 12. The queue management unit 13 causes the NRT packet # 2 to be output from the queue 12 to the multiplexing unit 14 in accordance with the instruction. The multiplexing unit 14 transmits the packet (B), which is the NRT packet # 2 input from the queue 12, via the output I / F unit 16 in accordance with the above instruction from the transmission scheduler 15.

  In addition, while the transmission operation of the packet (B) is started and the packet (B) is being read from the queue 12, the packet (a) arrives from the communication path 102a, and the packet (A) enters the queue 12 Writing and writing of the packet (C) that arrived from the communication path 102b before that to the queue 12 are completed.

  After that, when the transmission of the packet (B) (reading from the queue 12) is completed at time t7, the transmission scheduler 15 performs scheduling and determines the line C1 in FIG. 4 and then the line C3. In line C3, the determination is made for the leading packet (here, packet (C) and packet (a)) of each queue 12, but the determination is satisfied for packet (C) and the determination is satisfied for packet (a). do not do. Therefore, the operation of line C4 is selected for packet (C), and the operation of line C6 is selected for packet (a). That is, it decides to transmit the packet (a). As a result, the transmission scheduler 15 instructs the multiplexing unit 14 to select and output the NRT packet # 1. The transmission scheduler 15 instructs the queue management unit 13 to read out the packet from the queue 12 in which the packet (a) is written. In accordance with the instruction, the queue management unit 13 causes the queue 12 to output the packet (A) that is the NRT packet # 1 from the multiplexing unit 14. The multiplexing unit 14 transmits the NRT packet # 1 (packet (A)) input from the queue 12 via the output I / F unit 16 in accordance with the instruction from the transmission scheduler 15.

  In the above scheduling, the determination of the line C3 may not be established for both the packet (C) and the packet (a). In such a case, the transmission scheduler 15 selects either one of the packets. Select and decide to send. Alternatively, first, the total value of the read time of packet (C) and the read time of packet (a) is compared with the time until RT packet arrival. If the time until RT packet arrival is larger, packet (C) and packet ( A) is determined to be transmitted, and if the time until the arrival of the RT packet is shorter, it is determined to transmit one of the packets. Which packet is transmitted when the time until the arrival of the RT packet is shorter is the higher priority if the priorities are different. If the priorities are the same, either may be selected, but it is desirable to select the one with the larger data amount in view of transmission efficiency.

  When the transmission of packet (a) (reading from queue 12) is completed, since the packet is present in queue 12, transmission scheduler 15 performs scheduling and determines line C1 in FIG. 4 and then line C3. . In this case, the determination of the line C3 is established, that is, since there is no NRT packet that can complete transmission before the next arrival of the RT packet, it is determined not to perform transmission.

  Thereafter, the packet (4) from the data generator 20 is input to the multiplexing unit 14 and the RT packet arrival time prediction unit 17 via the input I / F unit 11 at time t8. Accordingly, the RT packet arrival time prediction unit 17 resets the counter value of the up counter 172 and presets the counter value of the down counter 173. The transmission scheduler 15 performs the same process as when the packets (1) to (3) described above are input, and instructs the multiplexing unit 14 to output the RT packet. The multiplexing unit 14 transmits the packet (4) via the output I / F unit 16.

  When the transmission of the packet (4) is completed at time t9, since the packet exists in the queue 12, the transmission scheduler 15 performs scheduling and determines the line C1 in FIG. 4 and then the line C3. In this case, the determination of the line C3 is not satisfied, that is, the transmission scheduler 15 can complete the reading of the NRT packet (packet (C)) from the queue 12 before the next RT packet (packet (5)) arrives. Further, processing of lines C5 and C6 is executed. As a result, the transmission scheduler 15 instructs the multiplexing unit 14 to select and output the NRT packet # 2. In addition, the transmission scheduler 15 instructs the queue management unit 13 to read a packet from the queue 12. The queue management unit 13 causes the NRT packet # 2 to be output from the queue 12 to the multiplexing unit 14 in accordance with the instruction. The multiplexing unit 14 transmits the packet (C) that is the NRT packet # 2 input from the queue 12 via the output I / F unit 16 in accordance with the instruction from the transmission scheduler 15.

  In addition, while the packet (C) transmission operation is started and the packet (C) is being read from the queue 12, the packet (b) arrives from the communication path 102a, and the packet (b) is sent to the queue 12. Writing is complete.

  After that, when the transmission of the packet (C) (reading from the queue 12) is completed at time t10, the transmission scheduler 15 performs scheduling and determines the line C1 in FIG. 4 and then the line C3. In this case, the determination of the line C3 is established, that is, since the reading of the NRT packet (packet (b)) from the queue 12 cannot be completed before the next RT packet (packet (5)) arrives, transmission is performed. Decide not to.

  Thereafter, when the packet (5) from the data generator 20 is input to the multiplexing unit 14 and the RT packet arrival time prediction unit 17 via the input I / F unit 11 at time t11, the RT packet arrival time prediction unit 17 The counter value of the up counter 172 is reset and the counter value of the down counter 173 is preset. The transmission scheduler 15 performs the same processing as when the packets (1) to (4) described above are input, and instructs the multiplexing unit 14 to output the RT packet. The multiplexing unit 14 transmits the packet (5) via the output I / F unit 16.

  Thereafter, when the transmission of the packet (5) is completed at time t12, the transmission scheduler 15 performs scheduling and determines the line C1 in FIG. 4 and then the line C3. In this case, the determination of the line C3 is not established, that is, the reading of the NRT packet (packet (A)) from the queue 12 can be completed before the next RT packet (packet (6) not shown) arrives. Further, it is determined that the processing of the lines C5 and C6 is further performed to transmit the packet (A). Then, the transmission scheduler 15 instructs the multiplexing unit 14 to select and output the NRT packet # 1. Further, the transmission scheduler 15 instructs the queue management unit 13 to read out the packet from the queue 12 in which the packet (A) is written. In accordance with the instruction, the queue management unit 13 causes the packet 12 to output the NRT packet # 1 from the queue 12 to the multiplexing unit 14. The multiplexing unit 14 transmits the NRT packet # 1 (packet (A)) input from the queue 12 via the output I / F unit 16 in accordance with the instruction from the transmission scheduler 15.

  As described above, in the communication apparatus according to the present embodiment, the RT packet arrival time predicting unit 17 outputs the next arrival time of the RT packet input from the data generator 20 to the transmission scheduler 15, and the transmission scheduler 15 The reading of the NRT packet from the queue 12 is controlled so as not to collide with the packet arrival time, and the output packet from the multiplexing unit 14 is selected, so that the RT packet input from the data generator 20 is The data is output to the communication network 200 via the output I / F unit 16 without delay in the communication device. For this reason, the data generated by the data generator 20 is transmitted to the data processing device 30 without delay, and the data processing device 30 can issue an instruction to the control target device (not shown) according to the value, thereby enabling fine control. Become.

Embodiment 2. FIG.
In the first embodiment, the RT packet arrival time prediction unit 17 realizes the prediction of the arrival time of the packet by resetting the up counter 172 and presetting the down counter 173 by detecting the head of the packet. However, an embodiment in which arrival of an RT packet is input from the outside of the communication apparatus 1 will be described next. The configuration of the packet transmission system is the same as that of the first embodiment (see FIG. 1).

  FIG. 6 is a diagram illustrating an internal configuration example of the communication apparatus according to the second embodiment. In FIG. 6, the same components as those of the communication device 1 (see FIG. 2) described in the first embodiment are denoted by the same reference numerals. That is, the communication device 1a of the present embodiment is obtained by replacing the RT packet arrival time prediction unit 17 of the communication device 1 described in Embodiment 1 with an RT packet arrival time prediction unit 17a. In the present embodiment, the operation of the RT packet arrival time prediction unit 17a, which is a different component from the first embodiment, will be described.

  In FIG. 6, RT packet arrival period input indicates the input of a pulse indicating the arrival time of the RT packet. The RT packet arrival time prediction unit 17a predicts the next packet input time based on the input of this pulse.

  FIG. 7 is a diagram illustrating an internal configuration example of the RT packet arrival time prediction unit 17a. The same components as those in the RT packet arrival time prediction unit 17 according to the first embodiment are denoted by the same reference numerals. The RT packet arrival time prediction unit 17a is configured to delete the packet detection unit 171 from the RT packet arrival time prediction unit 17 and to input pulses indicating the RT packet arrival time to the up counter 172 and the down counter 173. It is a thing.

  For example, an RT packet arrival pulse that is a pulse indicating the arrival time of an RT packet is input from the data generator 20 to the communication device 1a in synchronization with the packet output (not shown). The only difference in operation from the communication apparatus 1 of the first embodiment is that the reset of the up counter 172 and the preset of the down counter 173 of the RT packet arrival time prediction unit 17 are executed in response to the input of the RT packet arrival pulse. It is. Therefore, as in the first embodiment, the RT packet arrival time prediction unit 17a outputs the next arrival time of the packet input from the data generator 20 to the transmission scheduler 15, and the transmission scheduler 15 determines the packet arrival time. The packet reading from the queue 12 is controlled so as not to collide, and the output packet from the multiplexing unit 14 is selected. Thereby, like the communication device 1 of the first embodiment, the packet input from the data generator 20 is output to the communication network 200 without delay in the communication device 1a. That is, the data generated by the data generator 20 is transmitted to the data processing device 30 without delay, and the data processing device 30 can issue an instruction to the control target device based on the value, so that fine control is possible.

  As described above, the communication device according to the present invention is useful when multiplexing packets input from a plurality of communication paths and relaying them to the communication path, and in particular, packets that are required to be relayed with low delay. It is suitable for a communication apparatus that handles packets that do not require relaying with low delay.

1, 1a Communication device 11 Input I / F unit 12 Queue 13 Queue management unit 14 Multiplexing unit 15 Transmission scheduler 16 Output I / F unit 17, 17a RT packet arrival time prediction unit 20 Data generator (real time)
21 Data generator (non-real-time)
30 Data processing device 171 Packet detection unit 172 Up counter 173 Down counter 101, 102a, 102b, 103 Communication path 200 Communication network

Claims (10)

A communication device that relays multiple types of packets input from the outside,
Arrival time predicting means for predicting the arrival time of the first packet that is periodically input and required to be relayed with low delay;
Packet storing means for storing a second packet that is not required to be relayed with low delay;
A transmission packet determination unit that determines a packet to be transmitted to a communication path based on a prediction result by the arrival time prediction unit and a length of the packet held by the packet storage unit;
A communication apparatus comprising: The arrival time prediction means includes:
A packet detector for detecting the arrival of the first packet;
An up-counter that resets when the arrival of the first packet is detected;
A down counter that presets the count value of the up counter immediately before the up counter is reset when the arrival of the first packet is detected;
With
The communication apparatus according to claim 1, wherein the count value of the down counter is output to the transmission packet determination unit as a predicted result of the arrival time of the first packet. The arrival time prediction means includes:
An up-counter that resets when a signal indicating that the first packet has been transmitted is input from the transmission source device of the first packet;
A down counter that presets the count value of the up counter immediately before the up counter is reset when the signal is input;
With
The communication apparatus according to claim 1, wherein the count value of the down counter is output to the transmission packet determination unit as a predicted result of the arrival time of the first packet. The transmission packet determining means includes
Based on the prediction result, a first time that is the time until the next arrival of the first packet is calculated, and further, based on the length of the packet held by the packet storage means, Calculate the second time, which is the time required from the start of transmission to the completion of transmission,
4. The communication device according to claim 1, wherein the second packet is determined to be transmitted when the second time is shorter than the first time. 5. The packet storage means comprises a plurality of queues;
The transmission packet determining means includes
When packets are accumulated in two or more queues of the plurality of queues,
For each head packet in each queue in which packets are accumulated, a required time from the start of transmission to the completion of transmission is calculated, and the calculated required time is shorter than the first time. The communication apparatus according to claim 4, wherein if it exists, it is determined to transmit a packet corresponding to a required time shorter than the first time. The transmission packet determining means includes
If there are a plurality of calculated required times shorter than the first time, any one of the packets corresponding to the required time shorter than the first time is selected and the communication is performed. The communication apparatus according to claim 5, wherein the communication apparatus determines a packet to be transmitted to the path. A packet relay method executed in a communication device that relays a plurality of types of packets input from the outside,
A prediction step for predicting the next input time of the first packet that is periodically input and is required to be relayed with low delay;
Whether or not to transmit the second packet held when the packet transmission operation to the communication path is completed or when a second packet not requiring relay processing with low delay is newly input A packet transmission step of transmitting the second packet if it is determined to transmit based on the prediction result in the arrival time prediction step,
A packet relay method comprising: In the packet transmission step,
Based on the prediction result, a first time which is a time until the next arrival of the first packet is calculated, and transmission of the packet is started based on the length of the second packet held. And calculate the second time, which is the time required to complete the transmission,
The packet relay method according to claim 7, wherein the second packet is transmitted when the second time is shorter than the first time. If there are a plurality of queues for holding the second packet,
In the packet transmission step,
If packets are accumulated in two or more queues of the plurality of queues, the time required from the start of transmission to the completion of transmission for each of the head packets of each queue in which packets are accumulated And a packet corresponding to the required time shorter than the first time is transmitted if the calculated required time is shorter than the first time. Item 9. The packet relay method according to Item 8. In the second packet transmission step,
If there are a plurality of calculated required times shorter than the first time, any one of the packets corresponding to the required time shorter than the first time is selected and the communication is performed. The packet relay method according to claim 9, wherein the packet relay method transmits to a path.

Images