JP7732386B2 - How to arrange a time-sharing schedule - Google Patents

Description

The present invention relates to technology for adjusting schedules in time-division networks using TSN (Time Sensitive Networking).

TSN, described in Patent Document 1 and Non-Patent Document 1, is a network technology that extends standard Ethernet (registered trademark) to enable interoperability between industrial networks and IT networks, and is composed of multiple standards such as "IEEE802.1AS" and "IEEE802.1Qbv."

(1) IEEE802.1AS
The "IEEE 802.1AS" standard is abbreviated as "gPTP." Time synchronization packets using this "gPTP" function differ from regular PTP time synchronization packets in that they are transmitted at the L2 layer and can only be used with the "PtoP (between adjacent devices)" transparency function.

(2) IEEE802.1Qbv
The IEEE 802.1Qbv time-aware scheduler divides communications on an Ethernet network into fixed-length, repeating time cycles, and within these cycles configures time slices that are assigned to one or more of eight Ethernet priorities (the "traffic scheduling" function).

This allows traffic classes that require guaranteed transmission and cannot be interrupted to be granted exclusive use of the Ethernet transmission medium for limited periods of time. The basic concept is time division multiple access (TDMA), which allows time-critical communications to be separated from unimportant background traffic by establishing virtual communications channels for specific periods of time.

Patent Publication No. 2020-107943

“Overview and Implementation of TSN, Integrating IT and OT Networks”, [online], Retrieved March 3, 2021, Internet <URL: http:ednjapan.com/edn/arrticies/1803/23/news014.html>

There are many cases where it is desirable to simultaneously use traffic shedding using the traffic scheduling function of the aforementioned TSN's "IEEE 802.1Qbv" on each communication device on the network.

In this case, time is synchronized between each communication device using TSN's "IEEE 802.1AS (gPTP)," and then gates are set on the target ports of each communication device at the same time using "IEEE 802.1Qbv" traffic sheeping.

In other words, the level of packets that can pass through is set using "IEEE802.1Qbv" for the target port of each communication device, a time-sharing schedule is created for each level, and gate opening and closing settings (hereinafter referred to as gate settings) are made to operate this time-sharing schedule at regular intervals.

After that, as time passes, differences in the gates and times set between each communication device will occur due to differences in crystal accuracy, etc., but the time on each communication device will be corrected at regular intervals through the "gPTP" time synchronization process.

In this case, it would be possible to reset the gate at a fixed time in line with the time synchronization process, but in order for each communication device to operate using the same time, it is necessary to detect the target time and reset the gate after that detection. As a result, there is a high degree of dependency on the CPU performance and processing status of the communication device, and there is a risk of non-negligible variations occurring between communication devices. This could have a negative impact on the accuracy of gate start timing between communication devices.

The present invention was made to solve these conventional problems, and its goal is to synchronize the start timing (gate opening and closing timing) of the time-sharing schedule set for each communication device on the network, thereby reducing variation in that start timing.

(1) One aspect of the present invention is a method for synchronizing time between communication devices on a network,
A method for adjusting a time division schedule set for each level of packets that can pass through a target port of each of the communication devices,
A gate that operates the time-sharing schedule for each level at a fixed cycle is set in the target port of each of the communication devices,
Calculating an adjustment time for the start timing of the time-sharing schedule by repeatedly opening and closing the gate in synchronization with time;
The start timing of the time-sharing schedule is offset in accordance with the calculated adjustment time.

(2) Another aspect of the present invention is a communication device that is arranged on a network in a time-synchronizable manner and in which a time-sharing schedule is set for each level of packets that can pass through a target port,
The target port is set to open and close a gate that operates the time-sharing schedule for each level at a fixed cycle,
Calculating an adjustment time for the start timing of the time-sharing schedule by repeatedly executing the gate opening/closing setting in synchronization with time synchronization;
The start timing of the time-sharing schedule is offset in accordance with the calculated adjustment time.

(3) Yet another aspect of the present invention is a method for synchronizing time between communication devices on a network,
A method for adjusting a time-sharing schedule, in which a gate opening/closing setting is made for a target port of each of the communication devices, which operates the time-sharing schedule for each level of packets that can pass through at a fixed cycle, the method comprising:
a step of calculating an adjustment time for the start timing of the time-sharing schedule by repeatedly executing the gate opening/closing setting in synchronization with time;
and a step of offsetting the start timing of the time-sharing schedule in accordance with the calculated adjustment time.

This invention synchronizes the start timing of the time-sharing schedules set for each communication device on the network, thereby reducing variations in the start timing.

FIG. 1 is a configuration diagram showing an example of a network according to a first embodiment. This is a flow diagram showing gate settings for each TSN hub. FIG. 3 is a sequence diagram showing details of S04 to S07 in FIG. 2 . 1A is a sequence diagram showing steps S04 to S07 of the operation process of the first embodiment, and FIG. 1B is an explanatory diagram showing steps S08 and S09 of the same. FIG. 10 is a configuration diagram showing an example of a network according to a second embodiment. FIG. 10 is a schematic diagram of an example of gate opening and closing according to a time-sharing schedule in the first embodiment. FIG. 1 is a sequence diagram showing time synchronization and gate synchronization in Patent Document 1. FIG. 10 is a schematic diagram showing an example of gate opening and closing according to a time-sharing schedule in the second embodiment. A sequence diagram showing time synchronization and synchronization with the gate. Other network diagrams. FIG. 10 is an explanatory diagram showing gate correction of an edge-side port using an edge-side correction packet. FIG. 10 is an explanatory diagram showing gate correction of a main-side port by a main-side correction packet. 10 is a flowchart showing the overall process of gate correction according to the second embodiment. 14 is a flowchart showing details of S27 in FIG. 13; 14 is a flowchart showing details of S24 in FIG. 13; 14 is a flowchart showing details of S26 in FIG. 13;

The following describes a time-sharing schedule adjustment method according to an embodiment of the present invention. Here, a TSN-compatible Layer 2 (L2) switching hub (hereafter referred to as a TSN hub) is used as an example of a communications device to which the present invention is applied.

TSN hubs are equipped with the IEEE 802.1AS gPTP function and the IEEE 802.1Qbv traffic scheduling function. This allows time synchronization between TSN hubs and the creation of time-sharing schedules to ensure priority packets are sent without delay.

In other words, it is assumed that the TNS hubs on the network are time-synchronized using the "gPTP" function, enabling highly accurate gate setting using time synchronization. Details are explained below based on Examples 1 and 2.

Example 1 will be explained based on Figures 1 to 4. The TSN hub can be used while connected to a bus, ring, or even more complex network. In this example, we will assume the network N shown in Figure 1 as an example.

Network N is made up of TSN hubs 1-5. Here, TSN hubs 1-5 are time-synchronized using the "gPTP" function, and for "gPTP" ports 1a-5b, which require a time-sharing schedule, gates are set for packet transmission directions A-D using the traffic scheduling function.

For example, TSN hubs 1 and 3 have gates set on "gPTP" ports 1a and 3a for packets sent in direction A, and gates set on "gPTP" ports 1b and 3b for packets sent in direction B.

TSN hubs 4 and 5 have gates set at "gPTP" ports 4a and 5a for packets sent in transmission direction C, and gates set at "gPTP" ports 4b and 5b for packets sent in transmission direction D.

TSN hub 2 has a gate set at "gPTP" port 2a for packets sent in transmission direction A, a gate set at "gPTP" port 2b for packets sent in transmission direction B, a gate set at "gPTP" port 2c for packets sent in transmission direction C, and a gate set at "gPTP" port 2d for packets sent in transmission direction D.

These TSN hubs 1-5 are equipped with a timer (capable of expressing time down to nanoseconds) used for gPTP time synchronization, and a counter with a count on the order of nanoseconds (cycles). The gate start adjustment time is calculated using this timer and counter, and the start timing of the next gate (the start timing of the time-sharing schedule's operating cycle) is set based on the calculated gate start adjustment time. Below, the timer is referred to as "ToD" and the counter as "TC," and gate setting details are explained.

<Gate setting processing details>
2 and 3, the gate setting processing steps (S01 to S09) in the TSN hubs 1 to 5 will be described. Here, it is assumed that the "gPTP" function and traffic scheduling function of the TSN hubs 1 to 5 are operating normally.

S01: When processing begins, TSN Hubs 1-5 check their own port status and obtain information about ports that require gate setting. For example, TSN Hub 2 obtains information about "gPTP" ports 2a-2d.

S02: Based on the port information obtained in S01, the target port for gate setting is selected. For example, for TSN hub 2, "gPTP" ports 2a → 2d are selected in that order.

S03: Check whether gate settings are complete for all target ports selected in S02 (for example, "gPTP" ports 2a-2d for TSN Hub 2). If the check shows that gate settings are complete, end the process; if not, proceed to S04.

S04: To measure the start timing of gate processing (gate opening and closing), the period is measured using "TC," which is not affected by time synchronization using the "gPTP" function.

As shown in Figure 3, gating begins when the "TC" period measurement is "0 (nsec)," i.e., when it is "TC0 (nsec)." For example, if the target port is the "gPTP" port 2a of TSN hub 2, gating begins for port 2a when it is "TC0 (nsec)."

S05: When "TC" has passed "any count number x T (nsec)", i.e., after "TC1 = T (nsec)", collect the "ToD" time "t1 = HH:MM:SS.zzzzzzzzzzz" that is synchronized with "gPTP" time (see Figure 3).

S06: Calculate backwards from the time "t1" collected in S05 to the gate start time of S04, i.e., the "ToD" time "t2" at "TC0 (nsec)".

Explaining the details based on Figure 3, first, to calculate the difference between TCs, the calculation "TC1 - TC0" is performed (S06-1). Next, the calculation result of S06-1 is subtracted from "t1" collected in S05 (S06-2), and the time "t2 = HH:MM:SS.yyyyyyyyy" is calculated (estimated).

S07: Calculate the gate start adjustment time based on the time "t2" calculated in S06-1. At this time, the time "t3" is defined as in equation (1).
Formula (1): Time "t3" = Gate start time (gate opening/closing start time) "t2" + 1 second = HH:MM:SS + 1.000000000
Then, the time "t2" is subtracted from the time "t3" to determine the difference between "t2" and "t3", i.e., "difference data (nsec)" (S07-1). The "difference data (nsec)" is then divided by the gate period (1 msec) to determine the remainder (S07-2). This remainder is set as the gate start adjustment time b (nsec).

S08: Compare the gate start adjustment time b (nsec) from S07 with the preset execution threshold a (nsec). If the comparison shows that the gate start adjustment time b (nsec) is greater, proceed to S09; if not, end the process.

S09: At the next gate start timing, adjust the settings to start gate processing after waiting for gate start adjustment time b (nsec). Here, the gate is assumed to have the function to set gate start adjustment time b (nsec), and processing ends after the setting.

<<Example of operation processing>>
An example of operation processing will be described with reference to Fig. 4. Here, a 125 MHz 32-bit counter [1 count = 8 nsec (1000000000/125000000)] is used as "TC", and the gate period is "1 msec" as before.

(1) S04-S07
An example of the operation process from S04 to S07 will be described with reference to Fig. 4A. Here, the gate start of S04 is performed when "TC0 = 5 count X 8 nsec = 40 (nsec)".

Also, when "TC1 = 1255 count x 8 nsec = 10040 (nsec)", "time t1 = 10:20:30.759004560" of S05 is collected.

When calculating gate start adjustment time b (nsec) (S07), time "t2" from S06 is calculated backwards in advance. That is, the TC difference "10040 - 40 = 10000 nsec" is calculated (S06-1). Furthermore, the TC difference "10000 nsec" is subtracted from "time t1 = 10:20:30.759004560" from S05 to calculate "time t2 = 10:20:30:758994560" (S06-2).

Then, the gate start adjustment time b (nsec) is calculated based on "time t2 = 10:20:30:758994560". At this time, the time "t3 = 10:20:31.00000000" is defined according to equation (1). Furthermore, as shown in equation (2) in Figure 4(a), "difference data (nsec) = 241005440 (nsec)" is calculated (S07-1).

The "difference data (nsec)" calculated here is divided by the gate period (1 msec = 1,000,000 nsec) as shown in equation (3) in Figure 4(a). The remainder is used to calculate the "gate start adjustment time b (nsec) = 5,440 (nsec)" (S07-2).

(2) S08, S09
An example of the operation process in S08 and S09 will be described with reference to Fig. 4B. Here, the calculated gate start adjustment time b "5440 (nsec)" is compared with the execution threshold value a (nsec) (S08).

At this time, as shown by arrow P, if "execution threshold a (nsec) = 10,000 (nsec)," gate start adjustment is not performed. On the other hand, as shown by arrow Q, if "execution threshold a (nsec) = 5,000 (nsec)," gate processing is started after waiting for the gate start adjustment time of "5,440 (nsec)" at the next gate start timing, as shown in S09. This achieves the following effects A to C.

A: As shown in S09, the gate start timing set for TSN hubs 1-5 is adjusted in advance according to the gate start adjustment time b (nsec). This synchronizes the gate start times between TSN hubs 1-5, making it possible to reduce variations.

In this case, as mentioned above, the process of obtaining the ToD time used in "gPTP" time synchronization places a load on the CPU, but this process is limited to S05, and the CPU load is reduced by using "TC" for other counts. In this respect, the impact on the accuracy of time synchronization is reduced, and the accuracy of time synchronization is improved.

B: As shown in S08, gate start adjustment time b (nsec) is used to adjust the start timing of gate processing when it is greater than execution threshold a (nsec). By setting execution threshold a (nsec) in gate start adjustment time b (nsec) in this way, it becomes possible to synchronize gate processing between TSN hubs 1-5 with greater than expected accuracy.

C: Patent Document 1 adjusts the time-sharing schedule by offsetting it according to the distance between TSN hubs. Here, a probe packet is sent from a TSN hub at the end of the network, and the offset value is adjusted according to the reception timing of the probe packet from other TSN hubs.

Therefore, by applying the offset value of Patent Document 1 to the gate start adjustment time b (nsec), it becomes possible to set gates that take into account packet delays between TSN hubs.

Example 2 will be explained based on Figures 5 to 16. In this example, the "IEEE802.1AS" of the TSN hub is used to synchronize the "IEEE802.1Qbv" between the TSN hubs with high precision, and the "IEEE802.1Qbv" is then corrected for delays due to the distance between the server and client devices via the TSN hub.

As mentioned above, the TSN hub has the functions of "IEEE 802.1AS (time synchronization)" and "IEEE 802.1Qbv (time sharing schedule)." For example, consider the network example shown in Figure 5.

In Figure 5, 1-3 indicate TSN hubs (hereafter referred to as Hubs 1-3), and the server and terminal devices are connected via Hubs 1-3. Also, although only one terminal device is shown in Figure 5, multiple terminal devices may be connected, and information (status, audio, images, etc.) is periodically transferred from the terminal device to the server in response to a request from the server. The information sent from this terminal device ranges from important information that must not be lost to image information, for which loss is not a problem depending on the situation.

Conventionally, when each of hubs 1-3 time-shares at the same timing through time synchronization, as shown in Figure 7, packets sent from TSN hub 1 arrive late due to transmission delays before reaching TSN hubs 2 and 3. This causes the gate opening and closing timing to not match, resulting in packets being sent later than expected, and in the worst case scenario, packets may be lost.

Patent Document 1 therefore proposes a method of offsetting the time-division schedule by the transmission delay. This offset makes it possible to operate TSN in long-distance systems by opening and closing gates at timings delayed by the transmission delay.

However, if the network geometry or route changes due to functions such as STP (Spanning Tree Protocol)/RSTP (Rapid Spanning Tree Protocol), the transmission delay also changes, so the offset value must also be changed.

However, because time synchronization packets with transmission delays travel back and forth, calculating the offset value takes time, which can result in processing delays. To address this issue, this embodiment proposes a method that makes it easy to obtain the offset value for a time-sharing schedule.

<<Operation/Processing Details>>
The operation and processing of this embodiment will be explained with reference to Figures 5 to 8. Here, the TSN hub to which the server and other devices are connected is called the main hub, and the TSN hub at the end of the network is called the edge hub. According to the network configuration in Figure 5, Hub 1 represents the main hub, and Hub 3 represents the edge hub.

(1) Basic Concept Fig. 6 shows gate settings (examples of gate opening and closing settings) for a time-sharing schedule that are set on the TSN-enabled ports of each of hubs 1 to 3 (synonymous with the "gPTP" ports in Example 1: hereinafter referred to as TSN ports), and the type and passing time of packets that are to be passed are set for each of intervals A to E. In Example 1, the gate settings are repeatedly executed in synchronization with the time.

As shown in Figure 6, Hubs 1 to 3 have time synchronization functions and functions to synchronize time synchronization with gates, and the gates of Hubs 1 to 3 are the backbone of the network set in the TSN port. Gates are opened and closed by packet transmission, and as shown in Figure 7, gates 1 and 3 are used for packet transmission from Hub 1 to Hub 3, and gates 2 and 4 are used for packet transmission from Hub 3 to Hub 1.

If each gate opens and closes at the same time due to time synchronization, packets will arrive delayed due to the transmission distance, which could result in the time-sharing schedule not being maintained. For example, in Figure 7, when a packet is sent from Hub 1 to Hub 3, a transmission delay occurs at times B and C.

As mentioned above, in Patent Document 1, the time-sharing schedule is offset to take such transmission delays into account. While this method enables operation in accordance with the transmission delays of long-distance systems, it may not be possible to calculate the offset in a short time when the network geometry or route changes.

Therefore, in this embodiment, as shown in Figure 8, a correction packet period is added to the beginning of the gate setting, and the time-sharing schedule is offset according to the passage time of the correction packet. This correction packet is transmitted at the main hub (hub 1) and edge hub (hub 3) with the gate open only during the correction packet period of the time-sharing schedule.

As a result, correction packets are sent from hubs 1 and 3 in accordance with the time synchronization and time-sharing schedule. Meanwhile, intermediate hub 2 can forward packets at any time, so it also performs simple forwarding processing on correction packets. Note that the gate interval S in Figure 8 is set to allow only one packet (correction packet) to pass through.

In this case, time synchronization is maintained even if the network geometry or route changes, but the transmission delay time changes depending on the transmission route. As a result, Hub 2 will receive the correction packet from a different port than before, and if it can obtain the reception timing, it will be able to calculate the transmission delay and apply a new offset.

In this respect, there is no need to wait for packets to travel back and forth due to time synchronization as in Patent Document 1, and offsets can be calculated in a short time even if the network geometry or route changes.

The correction packet sent from the main hub (Hub 1) reaches the edge hub (Hub 3) and transfer ends. On the other hand, the correction packet sent from the edge hub (Hub 3) is terminated at the main hub (Hub 1) by functions such as access lists when it reaches the main hub (Hub 1). Depending on the network topology, there may be multiple edge hubs, and in such cases, multiple correction packets will merge at the main hub.

(2) Operation Processing Example An operation processing example will be described with reference to Fig. 9. Here, as shown in Fig. 5, hubs 1 to 3 are connected by TSN ports, with hub 1, the main hub, serving as the grand master (GM) for time synchronization, and hubs 2 and 3 serving as slaves synchronized with hub 1. Therefore, the timer times (ToD time) of hubs 2 and 3 match the ToD time of hub 1 through time synchronization.

S11: First, Hub 1 opens the gate for the correction packet period at ToD time (XX:ZZ:00.000000000 seconds) from its own TSN port and sends a correction packet (1581 bytes) to Hub 3.

Here, the gate function of Hub 1 is assumed to be able to determine the transmission time as "00.000000000 seconds", the correction packet sent from Hub 1 is called Correction Packet 1, the ToD time at which Correction Packet 1 is sent is called ToD Time A, and Correction Packet 1 is sent including ToD Time A. Note that the size of the correction packet is not limited to "1581 bytes" and may be other sizes.

S12, S13: Correction packet 1 sent by hub 1 is received by hub 2 and forwarded to hub 3 (S12). In other words, since the gate is always open during the period of the correction packet from hub 2, it is forwarded using the store-and-forward method.

At this time, Hub 2 can obtain the timing at which Correction Packet 1 was received, and can obtain the transmission delay from the difference with Hub 2's timer time (ToD time) and use this as the offset value.

Here, the ToD time at which correction packet 1 collected by hub 2 was received is defined as ToD time B (XX: ZZ: 00.0000yyyy seconds), and the time it takes from when correction packet 1 arrives at the TSN port of hub 2 until ToD time B is collected is defined as "t1."

In this case, transit time B from Hub 1 to Hub 2 is calculated by subtracting ToD time A and "t1" from ToD time B, as shown in equation (4) in Figure 9. Since transit time B calculated using equation (4) is equal to the delay time of the packet from Hub 1 to Hub 2, transit time B is added as the gate correction time when synchronizing the ToD time of the next Hub 2 with Gate 3 (edge-side port) (S13). This enables accurate gate setting.

Correction Packet 1 is then received by Hub 3. Since Hub 3 is an edge hub, there is no need to calculate the transmission delay.

S14-S16: Next, Hub 3 opens the gate for the correction packet period at ToD time (XX:ZZ:00.00a000000 seconds) from its own TSN port and sends the correction packet (1581 bytes) to Hub 1 (S14).

Here, the gate function of Hub 3 is assumed to be able to determine the transmission time as "00.00000000 seconds", the correction packet sent from Hub 3 is called Correction Packet 2, the ToD time at which Correction Packet 2 is sent is called ToD Time C, and Correction Packet 2 is sent including ToD Time C.

Correction packet 2 sent by Hub C is received by Hub 2 and forwarded to Hub 1 (S15), where the same process as S13 is executed. Here, the ToD time at which correction packet 2 collected by Hub 2 is received is defined as ToD time B' (XX hours, ZZ minutes, 00.00a0zzzzz seconds), and the time it takes from when correction packet 2 arrives at Hub 2's TSN port until ToD time B' is collected is defined as "t2."

In this case, the transit time B' from hub 3 to hub 2 is calculated using equation (5) in Figure 9, and transit time B' is added as the gate correction time when synchronizing the ToD time of the next hub 2 with gate 2 (main side port) (S16).

(4) Other Operation Processing Examples Other operation processing examples will be described with reference to Figures 10 to 12. Figure 10 shows another example of a network configuration. Here, hub 1 corresponds to the main hub, hubs 5, 7, and 8 correspond to edge hubs, gates 2, 4, 6, 8, 9, 10, and 12 indicate gates of the main side ports, and gates 1, 3, 5, 7, and 11 indicate gates of the edge side ports.

As shown in Figure 11, Hub 1 sends an edge-side correction packet (corresponding to Correction Packet 1 described above) to Hubs 5, 7, and 8. This edge-side correction packet is then forwarded sequentially by Hubs 2-4 and 6. At this time, the transit times through Hubs 2-4 and 6 are added to Gates 1, 3, 5, 7, and 11, and gate correction is performed on the edge-side port.

On the other hand, as shown in Figure 12, Hubs 5, 7, and 8 send a main-side correction packet (corresponding to correction packet 2 described above) to Hub 1. This main-side correction packet is forwarded sequentially by Hubs 2 to 4 and 6.

In this case, the corrections for main side gates 4 and 10 of hubs 3 and 6 overlap for hub 2, and the corrections for main side gates 8 and 9 of hubs 5 and 7 overlap for hub 4. When the gate correction paths for the main side ports overlap in this way, the longer transit time is used as the offset value.

<<Detailed processing content>>
The gate correction process will be described in detail with reference to Fig. 13. Here, it is assumed that the following data is input in advance by command and written to the gate correction information file.
Edge delay number Main flag Edge flag Number of the TSN port that operates the TSN (hereinafter referred to as the TSN port number)
S21: At the start of processing, the gate correction information file is read. As a result, the edge delay number, main flag, edge flag, and TSN port number are set in the TSN hub as internal data. Note that the main flag is set in the main hub, while the edge flag is set in the edge hub.

S22-S27: A gate for gate correction packets is set for the TSN port (S22). Then, a check is made to see if an edge-side correction packet has been received (S23). If the check shows that an edge-side correction packet has been received, the process proceeds to S24, which is the process performed when an edge-side correction packet is received.

If not received, proceed to S25. In S25, it is checked whether a main side correction packet has been received. If the result of the check is that a main side correction packet has been received, proceed to S26, which is the process when a main side correction packet is received. On the other hand, if not received, proceed to S27, which is the process when a gate correction packet is sent.

(1) Details of the Gate Correction Packet Transmission Processing The gate correction packet transmission processing of S27 will be described in detail with reference to Fig. 14. Here, the processing content differs between the main hub and the edge hub.

S31: When processing begins, it is confirmed whether the device itself is the main hub. This confirmation is made by checking whether the main flag is set or not. If the main flag is set, the device is considered to be the main hub, and processing proceeds to S32 and subsequent steps. On the other hand, if the main flag is not set, the device is not considered to be the main hub, and processing proceeds to S36 and subsequent steps.

S32-S35: The time of time synchronization (ToD time) is obtained from the timer (S32). The remaining time from the obtained ToD time to the next second is calculated (S33), and it is confirmed whether the calculated remaining time is less than the gate period (S34).

If the result of the check is that the time is not equal to or less than the gate period, the process ends. On the other hand, if the time is equal to or less than the gate period, an edge-side correction packet is sent from the TSN port with the TSN port number read in S21 (S35), and the process ends.

S36: It is determined whether the device itself is an edge hub. This determination is made by checking whether the edge flag is set. If the edge flag is not set, processing ends, but if it is set, processing proceeds to S37 and subsequent steps.

S37-S42: It is determined whether the main side correction packet has been received (S37). This determination is made by checking whether the reception flag for the main side correction packet is set (S38). If the reception flag is not set, the main side correction packet has not yet been received, and processing ends. On the other hand, if the reception flag is set, the main side correction packet has already been received, and processing proceeds to S39.

In S39, the edge delay time (edge delay number x gate period) is set in the wait timer to avoid collisions of edge-side correction packets. It is then checked whether the delay time set in S39 has elapsed (S40).

If the check shows that the time has not elapsed, the process is terminated. If the time has elapsed, the main-side correction packet is sent from the TSN port whose TSN port number was read in S21 (S41). After this transmission, the reception completion flag for the edge-side correction packet is cleared (S42), and the process is terminated.

(2) Details of the Edge Side Correction Packet Reception Processing The edge side correction packet reception processing in S24 will be described in detail with reference to FIG.

S51: When processing begins, it is confirmed whether the device itself is the main hub. This confirmation is made by checking whether the main flag is set. If the main flag is set, processing ends, but if the main flag is not set, the process proceeds to S52.

S52: It is determined whether the device itself is an edge hub. This determination is made by checking whether the edge flag is set. If the edge flag is set, the edge-side correction packet reception flag is set (S59), and processing ends. On the other hand, if the edge flag is not set, processing proceeds to S53 and subsequent steps.

S53-S58: First, the receiving port of the edge-side correction packet is saved as the transmitting port of the main-side correction packet (S53). Next, the time of time synchronization (ToD time) is obtained from the timer (S54), and the time synchronization timestamp function is used to obtain the elapsed time "t1" since the edge-side correction packet was received at the port (S55).

After obtaining this elapsed time "t1", the gate correction time is calculated using equation (4) (S56). Additionally, information on edge-side ports other than the port that received the edge-side correction packet is obtained based on the edge-side port number and TSN port number (S57).

The TSN port of the information obtained here is called the gate correction port. For this correction port, the gate correction time is added to the next gate tuning (S58), after which processing ends.

(3) Details of the Main-Side Correction Packet Reception Processing The main-side correction packet reception processing in S26 will be described in detail with reference to FIG.

S61: When processing begins, it is confirmed whether the device itself is the main hub. This confirmation is made by checking whether the main flag is set. If the main flag is set, processing ends, but if the main flag is not set, the process proceeds to S62.

S62: It is checked whether the device itself is an edge hub. This check is performed by checking whether the edge flag is set. If the edge flag is set, processing ends, but if it is not set, processing proceeds to S63 and subsequent steps.

S63-S66: First, the time of time synchronization (ToD time) is obtained from the timer (S63), and the time synchronization timestamp function is used to obtain the elapsed time "t2" since the main side correction packet was received at the port (S64).

After obtaining this elapsed time "t2", the gate correction time is calculated using equation (5) (S65). The main port number is also obtained, and the gate correction time is added the next time the main port's gate is tuned (S66), after which the process ends.

In this embodiment, when calculating the offset value of the time-sharing schedule, "t1" and "t2" can be obtained from the timestamp function, making it possible to easily calculate using equations (4) and (5).

In particular, if time synchronization is already in place, even if the network geometry or route changes due to functions such as STP/RSTP, the offset value for the time-sharing schedule can be calculated in a short time by simply obtaining the reception timing of the correction packet.

In addition, it is possible to calculate efficient offset values when collecting data from servers, etc., even in complex network configurations, without being affected by factors such as the network shape or the number of TSN hubs.

For example, even with the network configuration shown in Figure 10, accurate gate correction is possible as shown in Figures 11 and 12. As a result, when a server is connected to the main hub (hub 1) and a terminal device is connected to the edge hubs (hubs 5, 7, and 8), important packets can be transmitted without loss while a certain amount of other packets transmitted between the server and terminal device are stably forwarded.

In this embodiment, gate settings for the time-sharing schedule are required, but the offset value is automatically set even if the network shape or route changes, so this setting work can be omitted.

1-8...TSN hub 1a-5b..."gPTP" port A-D...Packet transmission direction N...Network

Claims (5)

Time synchronization between communication devices on the network
A method for adjusting a time division schedule for each level of packets that can pass through a target port of each communication device, in which a gate opening/closing setting is made to operate at a fixed cycle, and the time division schedule is adjusted,
a step of calculating an adjustment time for the start timing of the time-sharing schedule by repeatedly executing the gate opening and closing in synchronization with time;
offsetting a start timing of the time-sharing schedule in accordance with the calculated adjustment time;
and
each of the communication devices includes a timer used in the time synchronization and a counter that is not affected by the time synchronization;
acquiring from the timer a time when a given number of counts has elapsed since the execution of the time-sharing schedule;
calculating the adjustment time based on a time obtained by calculating back a time corresponding to the count number from the acquired time and a start time of the time sharing schedule;
A method for adjusting a time-sharing schedule, comprising: When the counter is "0 (sec)", the time-sharing schedule is executed,
2. The method for adjusting a time sharing schedule according to claim 1, wherein the time is acquired after the count number has elapsed since the execution of the time sharing schedule. The adjustment time is
A time obtained by adding an arbitrary number of seconds to the time obtained by the reverse calculation;
A start time of the time-sharing schedule;
3. The method for adjusting a time sharing schedule according to claim 1, wherein the difference is a remainder obtained by dividing the difference by the period of the time sharing schedule. comparing the adjustment time to a predetermined execution threshold;
3. The method for adjusting a time sharing schedule according to claim 1, wherein if the comparison result shows that the adjustment time is large, the start of the next time sharing schedule is delayed by the adjustment time. Time synchronization between communication devices on the network
A method for adjusting a time division schedule for each level of packets that can pass through a target port of each communication device, in which a gate opening/closing setting is made to operate at a fixed cycle, and the time division schedule is adjusted,
a step of calculating an adjustment time for the start timing of the time-sharing schedule by repeatedly executing the gate opening and closing in synchronization with time;
offsetting a start timing of the time-sharing schedule in accordance with the calculated adjustment time;
and
In the setting of gate opening and closing,
The type of packet you want to pass and the time it takes to pass are set for each gate interval.
a period of a correction packet is added to the beginning of the gate interval;
calculating a correction time corresponding to a transmission delay by transferring the correction packet to each of the communication devices;
The calculated correction time is used as the adjustment time,
The correction packet is sent to a main communication device that is assumed to be connected to a server in the network,
the communication device at the end of the network;
While sent from
The data is sequentially transferred by the intermediate communication device disposed between the main and the other than the terminal,
The intermediate communication device
A timer time when the correction packet is collected;
The time it takes for the correction packet to arrive at the target port and collect the timer time;
Calculate the transmission delay time based on
The method for adjusting a time-sharing schedule according to claim 1, wherein the transmission delay time is calculated as the correction time.