JP6684409B1 - Time synchronization system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform highly accurate time synchronization in a synchronization message exchange via an IP network. A slave device obtains a first time difference at a first time point by exchanging a synchronization message with a grand master device, and determines a first straight line connecting the first time difference and a reference value, Through the exchange of the synchronization message, the second time difference is acquired at the second time point, the second straight line connecting the second time difference and the reference value is determined, and the slope of the first straight line is the slope of the second straight line. If it is smaller than the above, the offset between the master time of the grand master device and the local time of the slave device is calculated based on the value corresponding to the second time point on the first straight line. [Selection diagram] Fig. 3

Description

  The present disclosure relates to a time synchronization system.

  In recent years, in many fields, time synchronization via a network is increasing using PTP (Precision Time Protocol). As an example, a mobile phone network or a broadcasting IP network can be used. On the other hand, as for networks, conventional SDH (Synchronous Digital Hierarchy) networks are being rapidly switched to Ethernet (registered trademark) -based IP networks. Therefore, it is required to perform time synchronization by PTP through the IP network.

IEEE Standard for a Precision Clock Synchronization Protocol for Networked Measurements and Control Systems, IEEE Std 1588-2008, July 24, 2008.

  Unlike the SDH network, in which clocks are synchronized in the entire network and the delay time in the network falls within the range of fluctuation due to stuffing, in the IP network, fluctuations in delay time due to buffering of packets in switches and routers in the network occur. A switch and a large PTP boundary clock, a peer-to-peer transparent clock, or an end-to-end transparent clock. Unless it is on a network consisting of routers, delay request / response mechanism (Delay request-resp nse Mechanism) and high accuracy synchronization thing is difficult using a simple offset calculation.

  On the other hand, there are usually various vendors and various generations of devices in a wide area IP network, and it is assumed that all devices in the network that require time synchronization support any of these operations of PTP. Building a network is not realistic.

  In the present embodiment, more accurate time synchronization is performed based on the time stamps acquired by the synchronization message exchange in the network in which the switches and routers in the network do not support PTP (Synchronization message exchange).

  A slave device according to an embodiment includes a local time that allows free running in a PTP slave of an ordinary clock, a means for creating a time stamp of a synchronization message exchange based on the local time, and a means for synchronizing with the means. A means to calculate the predicted value of the difference of the subsequent time stamps based on the difference of the time stamps acquired by exchanging messages, and the means to correct the local time by using the predicted value, and the predicted value based on the measured value of the time stamp. And means for correcting Specifically, the slave device is a slave device that operates according to PTP, and obtains the time difference between the transmission time and the reception time of the synchronization message through the exchange of the synchronization message with the grand master device, and obtains the time difference. And a reference value, and based on the value on the straight line, a controller configured to calculate the offset between the master time of the grand master device and the local time of the slave device is provided. .

  According to the present embodiment, it is possible to realize highly accurate time synchronization at the microsecond level necessary for broadcasting SFN (Single Frequency Network) and the like on an IP network that does not support PTP.

  As a result, it is not necessary to use an expensive dedicated network supporting GPS or PTP for time synchronization, which contributes to a reduction in the cost required for equipment or lines required for time synchronization, and a reduction in maintenance costs.

FIG. 1 is a diagram illustrating an overview of a system configuration according to an embodiment. FIG. 5 is a diagram showing a basic synchronous message exchange of PTP. FIG. 6 is a diagram showing a transition of a time difference in synchronous message exchange according to one embodiment. FIG. 6 is a diagram showing a transition of a time difference in synchronous message exchange according to one embodiment. FIG. 6 is a diagram showing a transition of a time difference in synchronous message exchange according to one embodiment. It is a figure which shows the structure of the PTP slave apparatus concerning one Embodiment. It is a figure which shows the logic module of the PTP slave apparatus concerning one Embodiment.

  Hereinafter, embodiments will be described in detail with reference to the drawings. The same elements are given the same numbers throughout the description of the embodiments.

  FIG. 1 is a diagram illustrating an outline of a configuration of a time synchronization system according to an embodiment. IEEE 1588 PTP (hereinafter "PTP") employs a hierarchical master-slave structure for clock distribution. The hierarchical master-slave structure includes grand master devices and slave devices. The grand master device (a node serving as a base point for transmitting its own clock to the outside) is also referred to as a grand master clock (GMC) and distributes time information to slave devices. The grand master device itself corrects its own master time using an external synchronization signal generated from a global source clock (GPS, atomic clock, etc.).

  The message exchange system according to the present embodiment adopts a hierarchical master-slave structure according to the PTP described above. The time synchronization system shown in FIG. 1 generally includes a PTP grandmaster device 101, and PTP slave devices 102, 103, and 104. The grand master device 101 and the slave devices 102, 103, and 104 are connected to the IP network 110 via Ethernets 1101, 1102, 1103, and 1104, respectively. IP network 110 includes intermediate nodes (switches and routers, not shown) that relay PTP messages (synchronization messages) exchanged between grandmaster device 101 and slave devices 102, 103, and 104. The slave devices 102, 103, and 104 each perform a synchronization message exchange (Synchronization message exchange) with the grand master device 101 via the IP network 110, and based on the offset from the master time of the grand master device 101. Correct your own local time.

  FIG. 2 shows the basic synchronization message exchange of PTP. In PTP, a synchronization message for correcting the local time of the slave device is exchanged between the grand master device 101 and the slave devices 102, 103 and 104 as described below. In the following description, only the slave device 102 will be taken as an example.

  First, the grand master device 101 transmits a Sync message to the slave device 102. The Sync message includes the transmission time (time stamp) (t1) of the Sync message. When the slave device 102 receives the Sync message, the transmission time (t1) and the arrival time (t2) (time stamp) of the Sync message are measured by the slave device 102.

  Next, the slave device 102 transmits a Delay_Req message to the grand master device 101. When the slave device 102 transmits the Delay_Req message, the transmission time (t3) (time stamp) of the Delay_Req message is measured by the slave device 102.

  Next, the grand master device 101 sends a Delay_Resp message to the slave device 102. The Delay_Resp message includes the actual arrival time (t4) (time stamp) of the Delay_Req message. When the slave device 102 receives the Delay_Resp message, the actual arrival time (t4) of the Delay_Req message is measured by the slave device 102.

  Although FIG. 2 shows the synchronous message exchange performed in the one-step mode, the present embodiment is naturally applied to the synchronous message exchange performed in the two-step mode. In the synchronous message exchange performed in the two-step mode, the grand master device 101 sends a Follow_Up message to the slave device 102. Here, in sending the Sync message, it is difficult to know the sending time of the message in advance, so the predicted value of t1 is measured in the Sync message, and the actual value of t1 is measured in the Follow_Up message. This compensates for the error. Hereinafter, the Sync message, Follow_Up message, Delay_Req message, and Delay_Resp message described above are referred to as synchronization messages.

  In PTP, it is assumed that the transmission delay time d between the grand master device 101 and the slave device 102 is the same on the forward path and the return path of the synchronous message exchange. Therefore, the offset Δt between the master time of the grand master device 101 and the local time of the slave device 102 can be calculated based on the four time stamps t1, t2, t3, and t4 measured by the slave device 102. . Since it is assumed that the delay time of the round-trip path of the synchronous message exchange is the same, the expressions (1) to (4) are established.

t2-t1 = Δt + d Formula (1)
t4-t3 = Δt + d Formula (2)
{(T2-t1) + (t4-t3)} / 2 = d Formula (3)
Δt = (t2-t1) -d Formula (4)

  The slave device 102 can calculate the offset Δt according to the equations (1) to (4) and correct its own local time based on the offset Δt.

  However, in reality, the transmission delay time of the IP network is constantly changing and fluctuating depending on the amount of traffic or the characteristics of the traffic, and the transmission delay time is different between the forward path and the return path.

  When an intermediate node in an IP network operates as a boundary clock, a peer-to-peer transparent clock, or an end-to-end transparent clock according to PTP, it measures a variable transmission delay time and notifies the slave device of the transmission delay time. To do. When the intermediate node implements the above-mentioned function, the slave device can grasp the transmission delay time. Therefore, the calculated value of the offset Δt is not affected by the variation in the transmission delay time.

  An intermediate node that does not support PTP cannot notify the slave device of the above-mentioned transmission delay time. Therefore, there is no means for quantifying this variation in transmission delay time on a network that does not support PTP. Therefore, the calculated offset value is affected by the variation in the transmission delay time, and as a result, the synchronization accuracy deteriorates.

  Also, in the local clock of the PTP standard slave device shown in IEEE Std 1588-2008, the corrected local time is used for the next synchronization message exchange. That is, the slave device measures t1 to t4 based on the variable transmission delay time, corrects its own local time based on the measured t1 to t4, and again t1 to t4 with the corrected local time. Will be measured. Therefore, the local time itself used for the synchronization message exchange is not stable and jitter or wander appears. On a network that does not support PTP, in addition to the above-mentioned variation in transmission delay time, instability of local time further deteriorates synchronization accuracy.

  In the present embodiment, it is possible to minimize the influence of variations in transmission delay time and instability of local time on the calculation of the offset Δt, and obtain a highly accurate offset value.

  In the present embodiment, each of the slave devices 102, 103 and 104 has two local times LT1 and LT2. The local time LT1 is not time-corrected according to the PTP described above and is only used for synchronous message exchanges (ie to calculate the four time stamps t1, t2, t3 and t4) (free running). The local time LT2 is corrected based on the offset from the master time of the grand master device 101 according to the PTP described above.

  By exchanging the synchronization message using the free-running local time LT1, it is possible to prevent the synchronization accuracy from further deteriorating due to the instability of the local time described above. The offset Δt between the master time of the grand master device and the local time of the slave device is due to the steady frequency difference between both clocks. Therefore, (t2-t1) and (t4-t3), which consist of the offset between the master time of the grand master device and the local time of the slave device and the transmission delay time, are fixed to minimize the influence of the ambient temperature of the equipment. Under the condition, it changes linearly in proportion to the transit time.

  The graph of FIG. 3 shows a case where the frequency of the local clock of the slave device is higher than the frequency of the clock of the grand master device as an example, and the time difference (t2-t1) and the time difference in the synchronous message exchange using the free-running local time LT1. The transition of (t4-t3) is shown. In the upper graph of FIG. 3 (hereinafter, “(t2-t1) graph”), the vertical axis represents (t2-t1) calculated when the slave devices exchange synchronization messages. The horizontal axis represents the time axis of the local time LT1 of the slave device based on t2. Similarly, in the lower graph of FIG. 3 (hereinafter, “(t4-t2) graph”), the vertical axis represents (t4-t3) calculated when the slave devices exchange synchronization messages. The horizontal axis represents the time axis of the local time LT1 of the slave device based on t3.

  In the following, both the time difference (t2-t1) and the time difference (t4-t3) refer to the time difference between the transmission time and the reception time of the synchronization message in the synchronization message exchange between the grand master device and the slave device.

  In the (t2-t1) graph, the polygonal line 301 represents (t2-t1) measured when the slave device exchanges the synchronization message. The polygonal line 304 represents the actual measurement value of the transmission delay time d calculated based on the actually measured (t2-t1). Similarly, in the (t4-t3) graph, the polygonal line 311 represents (t4-t3) measured when the slave device exchanges the synchronization message. A polygonal line 314 represents the measured value of the transmission delay time d calculated based on the measured (t4−t3).

  As the slave device local time LT1 (referenced to the time point of t2) on the horizontal axis of the graph elapses, the shortest value of (t2-t1) indicated by the straight line 302 is the clock of the grand master device and the local clock of the slave device. It changes linearly according to the frequency difference. Similarly, as the slave device local time LT1 (based on the time point t3) elapses, the shortest value of (t4-t3) indicated by the straight line 312 is the frequency difference between the clock of the grand master device and the local clock of the slave device. It changes linearly according to. Here, the shortest value is a theoretical value of (t2-t1) and (t4-t3) when it is assumed that the transmission delay time d is constantly the shortest fixed value.

  The actual measurement value of (t2-t1) or (t4-t3) in the actual synchronous message exchange changes like the broken line 301 or the broken line 311 because the transmission delay time d varies as shown by the broken line 304 or the broken line 314. To do. In this case, where the clock frequency of the local time LT1 of the slave device is higher than the clock frequency of the grand master device, t2> t1 always holds. Therefore, the absolute value of the measured value of (t2-t1) never falls below the shortest value of (t2-t1).

  In this embodiment, paying attention to this point, the predicted value of the value of (t2-t1) in the synchronous message exchange is calculated based on the time series of the measured value of (t2-t1). Similarly, the predicted value of the value of (t4-t3) in the synchronous message exchange is calculated based on the time series of the measured value of (t4-t3). Then, the local time LT2 is corrected using the predicted value instead of the measured values of (t2-t1) and (t4-t3). On the other hand, when the local time LT2 is corrected, the local time LT1 is not corrected, and the local time LT1 is exchanged during the synchronization message exchange (that is, when (t2-t1) and (t4-t3) are actually measured). To use.

  In FIG. 3, in the time series of the measured values of (t2-t1) in the synchronous message exchange, the reference value a1 at the start point (T00 in FIG. 3) and the measured value of the time series of (t2-t1) are connected. The value on the extension line (the straight line 305 at the time point T01 of FIG. 3) of the straight line having the smallest inclination (hereinafter, the predicted value straight line) in the straight line group is predicted at the time point of the next synchronization message exchange (t2-t1). Used as a value. The set value a1 is a reference value for the predicted value of (t2-t1).

  Similarly, the straight line with the smallest inclination (predicted value) among the straight line group connecting the reference value a2 at the start time point (T00 in FIG. 3) among the time series measured values of (t4−t3) in the synchronization message exchange. The value on the extension of the straight line (315 at time T02 in FIG. 3) is used as the predicted value of (t4-t3) at the time of the next synchronization message exchange. The set value a2 is a reference value for the predicted value of (t4-t3).

  Then, unless the slope of the straight line connecting the measured value in the synchronization message exchange and the time series start time point (a1 or a2 of T00) is less than the slope of the predicted value line at that time, the predicted value is used. The local time LT2 is corrected. That is, in the correction of the local time LT2, the predicted value on the predicted value straight line (straight line 305) is used as the time difference (t2-t1), and the predicted value on the predicted value straight line (straight line 315) is the time difference (t4-t3). ), The transmission delay time d is calculated. Here, the inclination is lower, that is, the inclination is smaller, means that the inclination angle with respect to the time axis, which is the horizontal axis, is small.

  In the graph (t2-t1) in FIG. 3, as described above, the straight line 305 connecting the (t2-t1) measured value 308 and the starting point a1 at T00 has the smallest inclination at time T01. Therefore, the straight line 305 is a predicted value straight line in the next synchronization message exchange. For example, at the time of T011 at which the synchronous message exchange next to T01 is performed, the measured value 307 of (t2-t1) is obtained, but the straight line connecting the reference value a1 and the measured value 307 is more than the straight line 305. The inclination angle is large. Therefore, when performing the time correction at the time point T011, the predicted value E1 corresponding to t2 at the time point T011 on the straight line 305 is used as the time difference (t2-t1).

  Similarly, in the (t4-t3) graph, the straight line 315 connecting the (t4-t3) measured value 318 and the starting point a2 at T00 has the smallest inclination at the time of T02 (the inclination angle is large in the negative direction). . Therefore, the straight line 315 is a predicted value straight line in the next synchronization message exchange. For example, the measured value 317 of (t4−t3) is obtained at the time of T021 at which the synchronous message exchange next to T02 is performed, but the straight line connecting the reference value a2 and the measured value 317 is more than the straight line 315. The inclination angle is large. Therefore, when performing the time correction at time T021, the predicted value E3 corresponding to t3 at time T021 on the straight line 315 is used as the time difference (t2-t1).

  In the (t2-t1) graph, the measured value 309 of (t2-t1) is obtained at the time T012 when the next synchronization message exchange is further performed. The slope of the straight line 306 connecting the measured value 309 and the starting point a1 at T00 is smaller than the slope of the predicted value line 305 at T012. Therefore, the predicted value straight line is changed to the straight line 306, and the actual measurement value 309 corresponding to t2 at the time T012 on the straight line 306 is used as the predicted value instead of the predicted value 310 on the straight line 305. The time difference (t2-t1) is calculated based on the predicted value 309, and the local time LT2 is corrected.

  In the subsequent correction of the local time LT2, the predicted value on the straight line 306 (for example, the straight line 306 at the time of T03 in FIG. 3 is obtained until the actual measurement value forming the predicted value straight line having the inclination smaller than that of the straight line 306 is obtained. The time difference (t2-t1) is calculated using the predicted value E2) corresponding to t2 at the time T03 above, and the local time LT2 is corrected.

  The same processing as above is performed for (t4-t3). Further, the measured value 319 of (t4-t3) is obtained at the time of T022 when the next synchronization message exchange is performed. The slope of the straight line 316 connecting the measured value 319 and the starting point a2 at T00 is smaller than the slope of the predicted value line 315 at T022. Therefore, the predicted value straight line is changed to the straight line 316, and the actual measurement value 319 corresponding to t3 at time T022 on the straight line 316 is used as the predicted value instead of the predicted value 320 on the straight line 315. The time difference (t4−t3) is calculated based on the predicted value 319, and the local time LT2 is corrected.

  In the subsequent correction of the local time LT2, the time difference (t4−t3) is calculated using the predicted value on the straight line 316 until an actual measurement value forming a predicted value straight line having a slope smaller than that of the straight line 316 is obtained.

  The measured value 321 of (t4−t3) is obtained in the synchronous message exchange at the time of T023. The slope of the straight line 322 connecting the measured value 321 and the starting point a2 at T00 is smaller than the slope of the straight line 316 of the predicted value at that time. Therefore, the predicted value straight line is changed to the straight line 322, the time difference (t4−t3) is calculated using the predicted value on the straight line 322, and the local time LT2 is corrected.

  In the subsequent correction of the local time LT2, the time difference (t4−t3) is calculated using the predicted value on the straight line 322 until the measured value of the inclination smaller than that of the straight line 322 is obtained. That is, in the present embodiment, when performing time correction of the local time LT2 in the slave device, on the straight line with the smallest inclination connecting the measured value of (t2-t1) obtained up to that point and the reference value a1, The predicted value corresponding to that time point is used as (t2-t1). Further, when performing the time correction of the local time LT2, the predicted value corresponding to that time point on the straight line having the smallest inclination connecting the measured value of (t4-t3) obtained up to that time point and the reference value a2 is obtained. Used as (t4-t3).

Regarding the above-mentioned predicted value, for example, the predicted value E1 shown in FIG. 3 can be calculated according to the equation (5).
E1 = a1 + (T011-T00) × (308-a1 / (T01-T00))
Formula (5)
T00 is the value of t2 (time stamp) at the time of T00, T011 is the value of t2 at the time of T011, and T01 is the value of t2 at the time of T01.

The predicted value E2 shown in FIG. 3 can be calculated according to the equation (6).
E2 = a1 + (T03-T00) * (309-a1 / (T012-T00))
Formula (6)
T00 is the value of t2 (time stamp) at the time of T00, T03 is the value of t2 at the time of T03, and T012 is the value of t2 at the time of T012.

Furthermore, the predicted value E3 shown in FIG. 3 can be calculated according to the equation (7).
E3 = a2 + (T021-T00) * (318-a2 / (T02-T00))
Formula (7) T00 is the value of t3 (time stamp) at the time of T00, T02 is the value of t3 at the time of T02, and T021 is the value of t3 at the time of T021.

  The slave device according to the embodiment operates according to one of two modes. The first mode is the "conventional PTP mode" which operates according to the conventional PTP. In the second mode, a process of calculating (t2-t1) and (t4-t3) using the above-mentioned predicted value on the predicted value straight line, correcting the local time LT2, and not correcting the local time LT1 is executed. "Extended PTP mode". The two modes may be switched manually, or may be automatically switched at the timing when the variation in transmission delay time is reduced in order to calculate reference values a1 and a2 described later.

  Next, how to set the values of a1 and a2 will be described. A1 in the (t2-t1) graph of FIG. 3 is a reference value for the predicted value of (t2-t1) at the time of T00, and is assumed to be smaller than the shortest value of (t2-t1). A2 in the (t4-t3) graph of FIG. 3 is a reference value for the predicted value of (t4-t3) at the time of T00, and is assumed to be smaller than the shortest value of (t4-T3).

  The shortest value of (t2-t1) and (t4-T3) cannot be grasped. Therefore, the reference value a1 is, for example, a predetermined arithmetic processing that is the shortest value (t2-t1) obtained as a result of the slave device operating in the "conventional PTP mode" and exchanging synchronous messages a predetermined number of times. A value obtained by (subtracting a value obtained by multiplying a predetermined value such as 0.5 times) may be set. In this case, when the slave device performs the synchronous message exchange in the "conventional PTP mode" and determines that the variation of the transmission delay time d (that is, the measured (t2-t1)) is small (t2-t1). Predetermined arithmetic processing may be performed on the shortest value of. Similarly, the reference value a2 is set to the shortest value of (t4−t3), which is obtained as a result of the slave device operating in the “conventional PTP mode” and exchanging the synchronous messages a predetermined number of times (the predetermined arithmetic processing ( For example, a value obtained by subtracting a value obtained by multiplying a predetermined value such as 0.5 times may be set. In this case, when the slave device performs the synchronous message exchange in the "conventional PTP mode" and determines that the variation of the transmission delay time d (that is, the measured (t4-t3)) becomes small (t4-t3). Predetermined arithmetic processing may be performed on the shortest value of.

  Further, the reference value a1 is, for example, a predetermined arithmetic processing to the average value of (t2-t1) obtained as a result of the slave device operating in the "conventional PTP mode" and exchanging synchronization messages a predetermined number of times. A value obtained by (subtracting a value obtained by multiplying a predetermined value such as 0.5 times) may be set. In this case, when the slave device performs the synchronous message exchange in the "conventional PTP mode" and determines that the variation of the transmission delay time d (that is, the measured (t2-t1)) is small (t2-t1). Predetermined arithmetic processing may be performed on the average value of. Similarly, the reference value a2 is a predetermined arithmetic processing ((t4−t3)) obtained as a result of the slave device operating in the “conventional PTP mode” and performing synchronous message exchanges a predetermined number of times. For example, a value obtained by subtracting a value obtained by multiplying a predetermined value such as 0.5 times may be set. In this case, when the slave device performs the synchronous message exchange in the "conventional PTP mode" and determines that the variation of the transmission delay time d (that is, the measured (t4-t3)) becomes small (t2-t1). Predetermined arithmetic processing may be performed on the average value of.

  As described above, when the slave device determines that the variation of the transmission delay time d in the synchronous message exchange has become small, the “conventional PTP mode” may be changed to the “extended PTP mode”. This is because it is possible to set appropriate a1 and a2 from the shortest value or the average value of (t2-t1) and (t4-t3) obtained as a result of the smaller variation in the transmission delay time d. The above-described setting of the reference values a1 and a2 and the transition between the modes are executed by the PTP protocol engine 611 described later.

  Alternatively, the reference value a1 may be set to zero. The same applies to the reference value a2. Regarding the reference value a2, since the accuracy of the local time LT1 of the slave device is low in the first place, it is possible that t3> t4 even when the frequency of the local clock of the slave device is higher than the frequency of the clock of the grand master device. . Therefore, in order to avoid that (t4-t3) falls below the shortest value of (t4-t3) when the value of the transmission delay time d is small, a2 may be set to a predetermined value less than 0. .

  Alternatively, when the network configuration is known, the total of the theoretical minimum values of the devices that configure the network may be set as the reference values a1 and a2.

  By performing the above process, the predicted value of (t2-t1) approaches the shortest value of (T2-t1) as the time from the starting point T00 elapses, and the predicted value of the synchronous message exchange becomes The influence of variations on offset calculation for local time correction will also be reduced. The same applies to (t4-t3).

  As a result, it is possible to minimize the influence of variations in transmission delay time on the offset calculation, and to obtain a highly accurate offset value, thereby achieving highly accurate time synchronization on an IP network that does not support PTP. it can.

  FIG. 4 shows the transition of the time difference (t2-t1) in the synchronous message exchange using the free-running local time LT1 in the case where the frequency of the local clock of the slave device is lower than the frequency of the clock of the grand master device. . The idea is the same as that explained in FIG. FIG. 5 shows the transition of the time difference (t4-t3) in the synchronous message exchange using the free-running local time LT1 in the case where the frequency of the local time of the slave device is lower than the frequency of the time of the grand master device. . The idea is the same as that explained in FIG.

  FIG. 6 is a diagram showing the configuration of the slave device 102 in FIG. Since slave devices 103 and 104 have the same configuration, only the configuration of slave device 102 is shown in FIG. The slave device 102 operates as a slave clock of the ordinary clock defined by PTP in terms of protocol. That is, it operates as a slave in synchronous message exchange. However, the configuration of the slave device 102 is different from the standard slave configuration of PTP shown in IEEE Std 1588-2008.

  The slave device 102 includes a control device 601, a communication device 602, a memory 603, and a storage device 604, each of which is coupled via an internal bus.

  The control device 601 is a microprocessor, and reads the program stored in the storage device 604 into the memory 603 and executes it. The control device 601 controls the entire slave device 102. The control device 601 may be implemented using hardware such as an ASIC (Application Specific Integrated circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array).

  The communication device 602 is a network interface that sends and receives a synchronization message to and from the grand master device 101 via the IP network 110. The communication device 602 implements an event interface 618 and a general-purpose interface 619 described later.

  FIG. 7 is a diagram showing logic modules of the slave device 102. The slave device 102 includes, as logic modules, a PTP protocol engine 611, a free-runnable local time unit 612, a time stamp generation unit 613, an offset calculation unit 614, a local time correction unit 615, a time difference calculation unit 616, and a predicted value straight line setting unit. 617 is included. These logic modules are implemented by the control device 601 executing and controlling a predetermined program for the communication device 602, or by executing a predetermined program in the control device 601. The slave device 102 also includes an event interface 618 and a general purpose interface 619, which are two logical interfaces on a single physical port connected to the Ethernet 1102.

  The event interface 618 is an interface for synchronous message exchange, and is connected to a time stamp generation unit 613 that generates a time stamp for synchronous message exchange.

  The general-purpose interface 619 is used for transmitting / receiving a message having no time stamp, and is connected to the PTP protocol engine 611. The PTP protocol engine 611 sends and receives PTP messages, updates the dataset, and executes the state machine associated with the port.

  The free-runnable local time unit 612 maintains the local time LT1 used for the synchronization message exchange. The free-runnable local time unit 612 is a counter capable of free running based on the clock from the local time LT1. The free-runnable local time unit 612 provides the time stamp generation unit 613 with the time stamp information (t2 and t3) for synchronizing message exchange shown in FIG. At the start point T00 of the predicted value line in FIG. 3, that is, when the "conventional PTP mode" is changed to the "extended PTP mode", the clock information (time stamp) of the local time LT2 is fetched from the local time correction unit 615, and Count up by free run based on the value.

  When the event interface 618 receives the synchronization message from the grand master device 101, the time stamp generation unit 613 exchanges the synchronization message to obtain the predicted value of FIG. 3 (that is, when the slave device 102 is in the “extended PTP mode”). In (), the time stamps t2 and t3 are generated using the local time LT1 from the free-runnable local time part 612. The time stamp generation unit 613 performs normal synchronization message exchange, for example, in the synchronization message exchange before the start point T00 of the predicted value line in FIG. 3 (that is, when the slave device 102 is in the “conventional PTP mode”). A time stamp is generated using the local time from the local time correction unit 615.

  At the time of exchanging the synchronous message for obtaining the predicted value of FIG. 3 (that is, when the slave device 102 is in the “extended PTP mode”), the offset calculation unit 614 transfers the time difference value ((( The offset is calculated based on the value of (t2-t1) or the value of (t4-t3), and is passed to the local time correction unit 615. In a normal synchronization message exchange, for example, a synchronization message exchange before the start point T00 of the predicted value straight line in FIG. 3 (that is, when the slave device 102 is in the “conventional PTP mode”), it is passed from the time difference calculation unit 616. The offset is calculated using the time difference value and passed to the local time correction unit 615.

  The time difference calculation unit 616 calculates a time difference value (value of (t2-t1) or value of (t4-t3)) based on the values of t1, t2, t3, and t4 passed from the PTP protocol engine 611, and makes a prediction. The calculated time difference value and the local time information (t2 or t3) for exchanging the synchronization message, to which the calculated time difference value is given, are passed to the value line setting unit 617 and the offset calculation unit 614.

  The predicted value straight line setting unit 617 has a function of performing processing of the time difference graphs shown in FIGS. 3, 4, and 5, sets a predicted value straight line (predicted value straight line), and based on the predicted value straight line. The time difference information (value of (t2-t1) or value of (t4-t3)) is passed to the offset calculation unit 614.

  The prediction value straight line setting unit 617 uses the time information (T00 in FIG. 3) of the start point of the prediction value line, the current time information, the reference value (a1 and a2 in FIG. 3) of the start point of the prediction value line, and the current prediction value line. 3 (measured value 308 at time T01 in the case of the predicted value straight line 305 in FIG. 3 or measured value 318 at time T02 in the case of the predicted value straight line 315) and the synchronous message exchange that obtains the measured value. Holds local time information for.

  The predicted value straight line setting unit 617 that has received the new time difference value and the new local time information from the time difference calculation unit 616 sets the point plotted by the time difference value and the slave time information on the time difference graph as the start point of the predicted value straight line. A straight line is drawn to compare the slope with the current predicted value straight line.

  If the inclination angle of the straight line drawn by the new time difference value and the new local time information with respect to the slave time axis is less than the current predicted value straight line, the straight line is set as the new predicted value straight line, and the new time difference value and new slave time Is passed to the offset calculation unit 614.

  If the slope of the straight line drawn by the new time difference value and the new local time information is larger than the current predicted value line, the predicted value line is not changed and the value on the predicted value line corresponding to the new local time is changed. The time difference value is passed to the offset calculation unit 614.

  The local time correction unit 615 corrects the local time using the offset passed from the offset calculation unit 614 and outputs the corrected local time information 620.

  Regarding the creation of the corrected local time information 620, a control loop using a normal DLL or PLL or holdover technology is used for wonder or jitter suppression in addition to logical slowing.

  The operation of the slave device 102 shown in FIG. 6 will be described below.

  The PTP packet received by the slave device 102 from the Ethernet 1102 is received by the event interface 618 in the case of a synchronous message exchange packet, and is received by the general-purpose interface 619 in the case of other messages.

  The synchronous message exchange packet received by the event interface 618 is passed to the PTP protocol engine 611 after the time stamp generating unit 613 confirms the time information in the packet.

  The packet information received by the general-purpose interface 619 is also passed to the PTP protocol engine 611. The PTP protocol engine 611 updates the data set with the information of the packet passed from the time stamp generation unit 613 or the general-purpose interface 619, and executes the state machine associated with the port.

  With respect to the information from the time stamp generation unit 613, the PTP protocol engine 611 executes the protocol sequence of the synchronous message exchange, and gives the resulting information of t1, t2, t3, and t4 to the time difference calculation unit 616. .

  The time difference calculation unit 616 calculates a time difference value (value of (t2-t1) or value of (t4-t3)) based on the information, and the predicted value straight line setting unit 617 and the offset calculation unit 614 The calculated time difference value and the local time information (t2 or t3) for exchanging the synchronization message, which gives the time difference value, are passed.

  The predicted value straight line setting unit 617 draws a straight line connecting the time difference value passed from the time difference calculation unit 616 and the point plotted in the local time information with the reference value of the predicted value straight line to determine the slope of the current predicted value straight line. If the inclination angle of the straight line drawn by the new time difference value and the new slave time with respect to the slave time axis is smaller than the current predicted value straight line, the straight line is set as the new predicted value straight line and the new time difference value and new The local time information is passed to the offset calculation unit 614.

  If not, the predicted value straight line is not changed, and the value on the predicted value straight line corresponding to the new local time information is passed to the offset calculation unit 614 as the time difference value.

  The slope comparison is performed as follows, for example, with reference to the time point T011 in the (t2-t1) graph shown in FIG. As shown in FIG. 3, the predicted value line at T011 is the straight line 305, and the measured value 307 (time difference (t2-t1)) is obtained. The actual measurement value for obtaining the straight line 305 is the actual measurement value 308 actually measured at time T01. Therefore, the storage device 604 stores the actual measurement value 308 (time difference (t2-t1)) actually measured at time T01 and t2 based on the local time LT1 at time T01.

Therefore, the predicted value straight line setting unit 617 calculates the predicted value (E1) corresponding to t2 at the time point T011 on the predicted value line (straight line 305) at the time point T011, and calculates the calculated predicted value E1 as the actual measurement value. Compare with 307. When the measured value is smaller, the predicted value straight line is changed to a straight line connecting the reference value a1 and the measured value, and t2 and the measured value at that time point (T011) are stored. In other words, the slopes are compared by measuring the first time difference (t2-t1) at the first time point TP1, t2 at the first time point TP1, and the second time difference (t2- at the second time point TP2). Based on t1) and t2 at the second time point TP2, the second time difference (t2-t1) is the second time point TP2 on the straight line connecting the first time difference (t2-t1) and the reference value a1. It is performed by determining whether it is larger than the predicted value (EV) corresponding to t2 in.
EV = a1 + (TP2-TP0) * (first time difference (t2-t1) -a1) / (TP1-T00) Formula (8)
TP2 is the value of t2 (time stamp) at the time of TP2, TP0 is the value of t2 at the time corresponding to the reference value a1, and TP1 is the value of t2 at the time of TP1.

  When the predicted value EV is smaller than the second time difference (t2-t1), the straight line connecting a1 and the second time difference (t2-t1) is the predicted value line. On the other hand, when the predicted value EV is larger than the second time difference (t2-t1), the straight line connecting a1 and the first time difference (t2-t1) is the predicted value straight line.

Similarly for t4 to t3, the slopes are compared by the first time difference (t4 to t3) measured at the first time point TP1, t2 at the first time point TP1, and the second time point measured at the second time point TP2. The second time difference (t4-t3) is on the straight line connecting the first time difference (t4-t3) and the reference value a2, based on the time difference (t4-t3) of 2 and the time t2 at the second time point TP2. It is performed by determining whether or not it is larger than the predicted value (EV) corresponding to t2 at the second time point TP2.
EV = a2 + (TP2-TP0) * (first time difference (t4-t3) -a1) / (TP1-T00) Formula (9)
TP2 is the value of t3 (time stamp) at the time of TP2, TP0 is the value of t3 at the time corresponding to the reference value a1, and TP1 is the value of t3 at the time of TP1.

  When the predicted value EV is smaller than the second time difference (t4-t3), the straight line connecting a2 and the second time difference (t4-t3) is the predicted value straight line. On the other hand, when the predicted value EV is larger than the second time difference (t4-t3), the straight line connecting a2 and the first time difference (t4-t3) is the predicted value straight line.

  The offset calculation unit 614 predicts, for example, in synchronization message exchange (that is, when the slave device 102 is in the “extended PTP mode”) after the start point T00 of the predicted value straight line in FIG. 3 when highly accurate time synchronization is performed. An offset is calculated based on the time difference value (value of (t2-t1) or value of (t4-t3)) passed from the value straight line setting unit 617, and is passed to the local time correction unit 615.

  The offset calculation unit 614 calculates the time difference in a normal synchronous message exchange, for example, in the synchronous message exchange before the start point T00 of the predicted value line in FIG. 3 (that is, when the slave device 102 is in the “conventional PTP mode”). The offset is calculated using the time difference value passed from the unit 616 and passed to the local time correction unit 615.

  As a result, when highly accurate time synchronization is performed (that is, when the slave device 102 is in the “extended PTP mode”), the local time is corrected based on the predicted value straight line, and normal synchronization is performed as a preliminary step. When exchanging messages (that is, when the slave device 102 is in the "conventional PTP mode"), it is possible to correct the local time as a normal ordinary clock.

  When the protocol sequence of the synchronous message exchange is executed by the PTP protocol engine 611, the time stamp generating unit 613 performs time synchronization with high accuracy (that is, when the slave device 102 is in the “extended PTP mode”). , A local time from the free-runnable local time unit 612 is used to generate a time stamp, and in a normal synchronous message exchange (that is, when the slave device 102 is in the “conventional PTP mode”), the local time correction unit. A time stamp is generated from 615 using the local time and is used for the synchronous message exchange.

  As a result, a stable free-run local time LT1 may be used for high precision time synchronization, and a corrected local time stamp may be used for normal synchronization message exchange. It will be possible.

  As described above, in the present embodiment, the predicted value of the next time stamp is calculated based on the time stamp information obtained by the synchronization message exchange and used for the correction of the local time. Further, in this embodiment, the PTP slave device is provided with two local times, that is, a free-running local time and a corrected local time corrected so as to be synchronized with the time of the grand master device. Therefore, by using the free-running local time for synchronization message exchange, it is possible to minimize the influence of variations in transmission delay time and instability of local time on offset calculation, and to obtain highly accurate time synchronization. There is.

  As a result, more accurate time synchronization can be achieved based on the time stamps acquired by the intermediate nodes (switches, routers, etc.) in the network in the PTP-unsupported network by the synchronization message exchange.

  The processes implemented in this embodiment may be executed by a computer program including computer-executable instructions. The computer-executable instructions, when executed by the controller, cause the controller to perform the processes implemented in this embodiment. The computer program may be stored in a computer-readable storage medium.

  The embodiments disclosed in the present specification and the drawings are merely examples, and should not be construed as being limited to the contents of the present disclosure when defining the technical scope of the present invention. Although each process is described separately for the sake of description, the processes may be integrated and coordinated so that the other part performs some or all of the processes of each process.

101 PTP grand master device 102, 103, 104 PTP slave device 1101, 1102, 1103, 1104 Ethernet 110 IP network 601 Control device 602 Communication device 603 Memory 604 Storage device 611 PTP protocol engine 612 Free run possible local time part 613 Time stamp generation 614 Offset calculation unit 615 Local time correction unit 616 Time difference calculation unit 617 Prediction value straight line setting unit 618 Event interface 619 General purpose interface

Claims (6)

A slave device operating according to PTP,
At a first time point, a first time difference between a transmission time and a reception time of the synchronization message is obtained through an exchange of a synchronization message with a grandmaster device, and the first time difference is the first time point. A first increment from a reference value corresponding to an earlier reference instant, said first increment corresponding to a time from said reference instant to said first instant,
Calculating a second time difference having a second increment from the first time difference at a second time point when exchanging the synchronization message after the first time point, the second increment being , Having the same rate of increase as the first increment, and corresponding to the time from the first time point to the second time point,
Calculating an offset between the master time of the grand master device and the local time of the slave device based on the second time difference ;
At a third time point after the first time point, a third time difference between the transmission time and the reception time of the synchronization message is acquired, and the third time difference is the third time from the reference value. Has an increment of
Calculating a fourth time difference having a fourth increment from the reference value, the fourth increment having the same rate of increase as the first increment;
Comparing the third increment with the fourth increment, the third increment and the fourth increment corresponding to a time from the reference time point to the third time point,
A fifth time difference having a fifth increment from the third time difference or a fourth time difference from the fourth time difference at a fourth time point when exchanging the synchronization message after the third time point. Calculating a sixth time difference having an increment of 6, the fifth increment having the same rate of increase as the third increment, and the time from the third time point to the fourth time point. Correspondingly, the sixth increment has the same rate of increase as the fourth increment, and corresponds to the time from the third time point to the fourth time point,
Calculating the offset based on the fifth time difference or the sixth time difference,
A slave device comprising a control device configured as described above. The control device is further configured to calculate the offset based on the fifth time difference when it is determined that the third increment is smaller than the fourth increment. The slave device according to Item 1 . The slave device includes a first local time and a second local time,
The control device is
Acquiring the first time difference based on the first local time,
Correcting the second local time based on the calculated offset,
The slave device according to claim 1, further configured as follows. The slave device according to any one of claims 1 to 3 , wherein the reference value is set to zero. The controller is further configured to calculate a transmission delay time through the exchange of the synchronization message,
The reference value, the slave device according to any one of claims 1 to 3 value obtained by subtracting a predetermined value to the minimum value of the transmission delay time the calculated is set, it is characterized. A method performed by a slave device operating according to PTP, comprising:
Obtaining a first time difference between a transmission time and a reception time of the synchronization message at the first time point through exchanging a synchronization message with a grandmaster device, wherein the first time difference is A first increment from a reference value corresponding to a reference time point prior to the first time point, said first increment corresponding to a time from said reference time point to said first time point; ,
Calculating a second time difference having a second increment from the first time difference at a second time when exchanging the synchronization message after the first time. An increment of 2 has the same rate of increase as the first increment and corresponds to the time from the first time point to the second time point;
Calculating an offset between the master time of the grand master device and the local time of the slave device based on the second time difference;
A step of obtaining a third time difference between the transmission time and the reception time of the synchronization message at a third time point after the first time point, wherein the third time difference is the reference value. With a third increment from
Calculating a fourth time difference having a fourth increment from the reference value, the fourth increment having the same rate of increase as the first increment;
Comparing the third increment and the fourth increment, the third increment and the fourth increment corresponding to a time from the reference time point to the third time point. When,
A fifth time difference having a fifth increment from the third time difference or a fourth time difference from the fourth time difference at a fourth time point when exchanging the synchronization message after the third time point. Calculating a sixth time difference having an increment of 6, said fifth increment having the same rate of increase as said third increment, and said third to fourth time points. The sixth increment has the same rate of increase as the fourth increment, and corresponds to the time from the third time point to the fourth time point.
Calculating the offset based on the fifth time difference or the sixth time difference;
A method comprising:

Images