KR101322841B1 - Method for executing time synchronization between nodes of network and appartus for executing the method - Google Patents

Abstract

The present invention relates to a method for performing time synchronization between nodes in a network and a device for performing the method. In the method for performing time synchronization between nodes of a network according to the present invention, a predetermined value of a measured value for propagation to a master node using a time stamp in a slave node is pre-determined. Calculating each time unit, calculating an estimate of an actual propagation time using the measured value and a sliding window at the slave node, and transmitting a message at the master node. Calculating an offset using a time, a message reception time at the slave node, and the estimated value.

Time synchronization, propagation time, digital controlled oscillator (DCO), phase locked loop (PLL), time stamp

Description

METHOD FOR EXECUTING TIME SYNCHRONIZATION BETWEEN NODES OF NETWORK AND APPARTUS FOR EXECUTING THE METHOD}

1 is an example for explaining an inter-node synchronization method in the prior art.

2 is an example of a method for calculating propagation time.

3 is an example of a method of measuring a propagation time when a relay device exists between a slave node and a master node.

4 is an example of an offset calculation method using a peer-to-peer transparent clock.

5 is a flowchart illustrating a method of performing time synchronization between nodes in a network according to the first embodiment of the present invention.

6 is a flowchart illustrating a method of calculating an estimated value of actual propagation time using a sliding window according to the first embodiment of the present invention.

7 is a block diagram for explaining an internal configuration of a device constituting a slave node according to the second embodiment of the present invention.

8 is a flowchart illustrating a method of performing time synchronization between nodes in a network according to a third embodiment of the present invention.

9 is a flowchart illustrating a method of calculating an estimated value of actual propagation time using a linear digital filter according to a third embodiment of the present invention.

10 is a block diagram for explaining an internal configuration of a device constituting a slave node according to the fourth embodiment of the present invention.

11 is an example for explaining a bidirectional method for time synchronization.

12 is a flowchart illustrating a method of performing time synchronization between nodes in a network according to a fifth embodiment of the present invention.

13 is a block diagram for explaining an internal configuration of a device constituting a slave node according to the sixth embodiment of the present invention.

14 is an example for explaining a one-way method for time synchronization.

<Explanation of symbols for the main parts of the drawings>

710: calculation unit

720: estimated value calculation unit

721: measurement value holding unit

722: second estimated value calculator

730: offset calculation unit

The present invention relates to a method for performing time synchronization between nodes of a network and a device for performing the method. More particularly, the present invention relates to jitter and wonder in performing time synchronization between nodes of a network based on a time stamp. And a method and apparatus for improving time synchronization performance.

There are currently a number of schemes for delivering timing using time stamps, such as the IEEE 1588 protocol and the Network Time Protocol (NTP).

In all such schemes, each of the nodes constituting the network attempts to synchronize with one of the nodes. At this time, one node belongs to a node called a master, that is, a master node. In this way, the timing is also traceable in one node, the grandmaster node, called the grandmaster.

The master node sends a message containing a time stamp to all nodes attached to the master node except the master node of the master node. In this case, the time stamp may mean a time at which the message is transmitted. Each slave node receives this time stamp and adds a propagation time between the slave node and the master node to the time stamp. That is, the slave node compares the time stamp plus propagation time from the time of receiving the message, that is, the transmission time from the master node plus propagation time, and calculates an offset that is a difference from the master node to perform synchronization. Can be.

1 is an example for explaining an inter-node synchronization method in the prior art.

The master node 101 sends a message to the slave node 102. At this time, the time stamp T1, which is the time at the moment 103 of transmitting the message, is recorded in the message. In addition, when a slave node receives a message, it checks the reception time T2 which is the time of the moment 104 of receiving a message. If the slave node 102 knows the propagation time D1 from the master node 101, the slave node 102 may calculate an offset that is a difference from the master node 101 as "offset = T2-T1-D1". In addition, the offset may be used to perform synchronization with the master node 101.

However, the slave node 102 must know the propagation time D1 from the master node 101 to calculate the offset as in the assumption above. That is, the slave node 102 must calculate the propagation time D1.

Such a method of calculating the propagation time and another method of calculating the offset will be described in detail in [Configuration of the Invention] as a background of the present invention.

The schemes described above and later do not specify how to calculate the offset used at the slave node. Although it can be adjusted directly at the slave node, this often leads to problems that result in excessive jitter, wonder and time synchronization inaccuracies. In many cases, the sequence of offsets in the slave node is filtered using a Digital Controlled Oscillator (DCO), and using an analog Phase Locked Loop (PLL) if lower jitter and wonder are desired. However, this scheme of using DCOs and PLLs in slave nodes is only used for stringent applications because of the higher cost.

One main source of time inaccuracy is due to the limited granularity of the time stamp measurement. For example, standard Ethernet bridges and terminators require a 25 MHz oscillator. In such an apparatus, the particle size of the time stamp measurement is 40 nanoseconds. This means that an error of 80 nanoseconds can be generated in the propagation time measurement.

This error accumulates at propagation time in schemes using peer-to-peer transparent clocks. In the case of N-hops, that is, when there are N-1 peer-to-peer transparent clocks between the master node and the slave node, the possible error is 80 N nanoseconds. In the case of a large number of hops, this error increases rapidly, exceeding several hundred nanoseconds or one microsecond. In this case, the PLL filtering described above can be used to reduce the error, but such filtering is expensive and requires a separate storage location for the timing signal. This means more cost. Therefore, there is an urgent need for a new time synchronization method that can improve the performance of jitter, wonder and time synchronization without using expensive PLL filtering.

The present invention proposes a method for performing time synchronization between nodes of a network and a new technology related to a device for performing the method, in order to solve the above problems of the prior art.

The present invention calculates measured values of propagation based on a time stamp for each predetermined time unit in a slave node, and uses a mean and sliding window of the measured values to master node. It is aimed at improving the performance of jitter, wander and time synchronization by calculating the offset from.

Another object of the present invention is to calculate the measurements of the propagation time based on the time stamp for each pre-determined time unit at the slave node, and master using the average of the measurements and a general linear digital filter. By calculating the offset from the node, you improve the performance of jitter, wonder, and time synchronization.

It is still another object of the present invention to calculate a first offset from a master node for each predetermined time unit in a system for measuring propagation time in a unidirectional direction, and using the first offset and offsets calculated in the previous time unit as input values. By calculating the second offset, which is the offset in the current time unit, through a linear digital filter, the performance of jitter, wonder and time synchronization is improved while taking into account the residence time in the relay.

In order to achieve the above object and to solve the above-mentioned problems of the prior art, a method of performing time synchronization between nodes of a network according to an embodiment of the present invention, using a time stamp (time stamp) in the slave node Calculating a measured value for a propagation time to a master node for each predetermined time unit, and actually propagating using the measured value and a sliding window at the slave node Calculating an estimate of an actual propagation time, and calculating an offset using the message transmission time at the master node, the message reception time at the slave node, and the estimate.

According to one aspect of the invention, the step of calculating the estimated value of the actual propagation time using the measured value and the sliding window in the slave node, the most recently calculated by the number corresponding to the length of the sliding window Maintaining a value and calculating a second estimate of the actual propagation time using the average of the retained measurements and a first estimate calculated over a previous time unit.

According to another aspect of the invention, the step of calculating an average of the retained measured values, and using the average and the first estimate calculated in the previous time unit to calculate a second estimate of the actual propagation time, As shown in Equation 1, the second estimated value may be calculated.

Where d _k is the measured value in _k -th time unit, D _k is the estimated value in k-th time unit, M is the length of the sliding window, and H (z) is D _k. Can mean the transfer function of the difference equation of

According to another embodiment of the present invention, a method of performing time synchronization between nodes in a network includes calculating a measurement value of a propagation time from a slave node to a master node by each predetermined time unit using a time stamp. Calculating an estimated value of actual propagation time at the slave node using the measured value and a linear digital filter, and at the message transmission time at the master node, at the message reception time at the slave node, and Calculating an offset using the estimated value. At this time, the step of calculating the estimated value of the actual propagation time using the measured value and the linear digital filter at the slave node, the step of maintaining the measured value by a predetermined first number, the second predetermined number Maintaining a first estimated value measured in a previous time unit by the second step and calculating a second estimated value which is an estimated value of the actual propagation time using the maintained measured value, the first estimated value and the linear digital filter. The linear digital filter may assign a plurality of filter coefficients to the held measurement and the first estimate and calculate the sum as the second estimate.

According to another embodiment of the present invention, a method of performing time synchronization between nodes of a network includes: calculating a first offset from a master node by a predetermined time unit using a time stamp at a slave node; and Calculating a second offset using the first offset and the linear digital filter.

Hereinafter, various embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

Background

2 is an example of a method for calculating propagation time.

The propagation time between the clocks attached to two nodes, namely the master node 201 and the slave node 202, respectively, is measured by the clocks that exchange time stamps. A requester, which is one of the clocks, records the transmission time T1 203 of the first message in the first message when the requester sends the first message to another clock. The other clock responder records the reception time T2 204 of the first message.

Thereafter, the responder transmits a second message corresponding to the first message to the requester. At this time, the requester records and transmits the transmission time T3 205 of the second message in the second message. Finally, the requester checks the reception time T4 206 of the second message.

The requester may calculate the propagation time represented by Equation 2 using the T1 203 to T4 206 identified as described above.

Propagation time = ((T2-T1) + (T4-T3)) / 2

Here, it is assumed that the method of measuring the propagation time satisfies the following conditions (1) to (4).

(1) The propagation time is symmetrical. That is, the same in both directions.

(2) The propagation time is stable. In other words, the timescale for obtaining the four time stamps T1 to T4 is constant.

(3) The frequency difference between the master node and the slave node is small.

(4) The value of the time stamp is correct.

However, the assumption of (1) and (2) means that there can be no source of variable delay between the requester and the responder. In other words, a relay device such as a network switch or a network bridge cannot exist between the master node and the slave node. The presence of any such device can cause jitter, wonder and inaccuracy of time synchronization.

The presence of the network switch or the network bridge between the master node and the slave node causes a change and asymmetry of propagation time, and as another approach to deal with this, such as the network switch or the network bridge. There is a method of recording a reception time and a transmission time of each message in the relay device in the message. This method requires that the relay device include a clock and time stamping hardware called a transparent clock.

The transparent clock calculates a difference between the reception time of the message and the transmission time of the message. In this case, this difference is called a residence time. The dwell time is accumulated in the field of the message, and the accumulated value means a changed portion of the propagation time. That is, the value may mean an amount of change of the propagation time changed by the relay device in the slave node.

3 is an example of a method of measuring a propagation time when a relay device exists between a slave node and a master node.

The slave node 301 transmits a first message to the master node 302, and the master node 302 transmits a second message corresponding to the first message to the slave node 301, thereby explaining FIG. 2. Measure T1 to T4 as in. In this case, the first message and the second message pass through the relay node 303, and the relay node 303 stays at the relay node 303 using a transparent clock included in the relay node 303. Measure your time.

That is, the relay node 303 calculates the difference R1 between the reception time 304 and the transmission time 305 of the first message, and the difference R2 between the reception time 306 and the transmission time 307 of the second message. By calculating and recording in the first message and the second message, respectively, the slave node 301 can know the change amount of the propagation time generated by the relay node 303.

In this case, the propagation time may be calculated by the slave node 301 as shown in Equation 3 below.

Propagation time = ((T2-T1-R1) + (T4-T3-R2)) / 2

As such, the approach using the transparent clock requires that the propagation time between the slave node and the master node be measured again after network reconfiguration. This is because such a mechanism does not provide a measure of the propagation time between all pairs of transparent clocks, but only a measure of the propagation time between each slave node and the master node of the slave node. . This mechanism may cause a loss of synchronization when the network is reconfigured.

Another approach is for the transparent clocks to exchange messages for themselves to measure the propagation time between the transparent clocks. In other words, the message exchange is bidirectional between all neighboring pairs of the transparent clocks. The transparent clocks have a property called peer-to-peer transparent clock. That is, when a master node and a slave node are connected between the peer-to-peer transparent clock, the master node transmits a message indicating a transmission time to the slave node, and each of the peer-to-peer Peer transparent clock accumulates the time of stay in the message. In addition, the propagation time measured between the peer-to-peer transparent clocks is also accumulated in the message. That is, the propagation time of the link on which the message arrived is accumulated in the message.

At this time, the slave node receiving the message directly from the master node using the transmission time, the accumulated dwell time, the accumulated propagation time and the reception time of the message in the master node recorded in the message. The offset of can be calculated.

4 is an example of an offset calculation method using a peer-to-peer transparent clock.

The master node 401 sends a message to the slave node 402. In this case, the message includes a transmission time T1 403 of the message. When the message passes through the relay node 404, the relay node 404 calculates the dwell time 405 at the relay node 404 and the propagation time 406 of the link that received the message and records it in the message. To the slave node 402.

The slave node 402 measures the reception time T2 407 of the message and the propagation time D1 408 for the link between the last relay node and the slave node 402. The slave node 402 is the transmission time T1 403 of the message, the dwell time 405 and the propagation time 406 of the link that received the message, the propagation time D1 408 and the reception time T2 of the message ( 407 may be used to calculate an offset from the master node 401.

In this case, the slave node 402 may calculate an offset from the master node 401 as shown in Equation 4 below.

Offset = T2-T1-R1-D1

Here, R1 may mean a cumulative value of a dwell time 405 and a propagation time 406 of a link receiving the message.

The propagation time measuring method described in FIGS. 2 and 3 is a bidirectional method of measuring an offset by measuring a propagation time while exchanging messages between a master node and a slave node, and the offset measuring method described in FIG. 4 is a slave node in a master node. This is a one-way way to calculate the offset in one message transmission.

2. Time synchronization method according to the present invention.

The time synchronization method according to the present invention can be used in the following cases (1) to (3).

(1) Each slave node measures the delay time with the master node without the transparent clock present between the master nodes.

(2) Each slave node measures the delay time with the master node, where there is one or more end-to-end transparent clocks between the slave node and the master node. .

(3) There is one or more peer-to-peer transparent clocks between the slave node and the master node, and each propagation time is measured separately in neighboring peer-to-peer transparent clock pairs. The propagation time between the master node and the first peer-to-peer transparent clock and the propagation time between the slave node and the last peer-to-peer transparent clock are also measured separately.

Here, when the stable propagation time is measured using a clock with a limited particle size of the phase measurement, the measured values of the propagation time tend to rise between the two values. That is, the measurement has a value between a maximum integer multiple of clock granularity smaller than the propagation time and a minimum integer multiple of the click granularity greater than the propagation time.

The propagation time between the two values can measure the actual propagation time through the average of successive measured values, and the following two methods can be used to calculate the average.

The first method uses a sliding window of length M, and the second method uses a general linear digital filter.

5 is a flowchart illustrating a method of performing time synchronization between nodes in a network according to the first embodiment of the present invention.

In operation S510, the slave node calculates a measurement value for the propagation time to the master node for each predetermined time unit using the time stamp. In this case, the slave node may store and maintain the measured value calculated for each time unit by a predetermined number. In order to calculate the measured value in the current time unit, the slave node may be performed by including steps S511 to S514 in step S510.

In step S511, the slave node records a first time stamp, which is a transmission time of a first message, in the first message and transmits it to the master node. That is, the slave node needs four times as described above to calculate the measurement for propagation time, and for this purpose sends the first message to the master node. In this case, the first time stamp corresponds to T1 of Equation 2.

In step S512, the master node checks a second time stamp that is a reception time of the first message. In this case, the second time stamp corresponds to T2 of Equation 2.

In step S513, the master node records the first time stamp, the second time stamp, and a third time stamp, which is a transmission time of a second message corresponding to the first message, in the second message. Send to node. Here, the third time stamp corresponds to T3 of Equation 2, and transmits the first time stamp to the third time stamp by transmitting the second message to the slave node.

In operation S514, the slave node checks a fourth time stamp that is a reception time of the second message, and calculates a measurement value for the propagation time using the first time stamp and the fourth time stamp. . In this case, the fourth time stamp corresponds to T4 of Equation 2. That is, the slave node may calculate the measured value as shown in Equation 5 below.

Here, d _k is the measured value, T _{1, k} is the first time stamp, T _{2, k} is the second time stamp, T _{3, k} is the third stamp, T ₄ , _k denotes the fourth time stamp, and k denotes that the first to fourth time stamps are measured in k-th time units, respectively.

In operation S520, the slave node calculates an estimated value of actual propagation time using the measured value and the sliding window. The sliding window of length M may comprise a buffer capable of storing M measurements. In this case, the sliding window may maintain M of the most recent measured values in the buffer, and calculate the estimated value of the actual propagation time through the M measured values. As described above, the step (S520) of calculating the estimated value of the actual propagation time through the sliding window will be described in more detail with reference to FIG. 6.

In operation S530, the slave node calculates an offset by using a message transmission time at the master node, a message reception time at the slave node, and the estimated value. In this case, the slave node may use the following equation (6) to calculate the offset.

Offset = T2-T1-D1

Here, T2 denotes the message reception time, T1 denotes the message transmission time, and D1 denotes the estimated value.

That is, subtracting the message transmission time from the master node from the message reception time of the slave node, the time until the message reaches the slave node remains. Here, except for the measurement of the propagation time calculated using the sliding window, a timing difference between the slave node and the master node may be known, and time synchronization between the slave node and the master node may be performed using the same. Can be done.

6 is a flowchart illustrating a method of calculating an estimated value of actual propagation time using a sliding window according to the first embodiment of the present invention. As illustrated in FIG. 6, steps S601 and S602 may be included in step S520 described with reference to FIG. 5 to calculate the estimated value using the sliding window.

In step S601, the slave node maintains the measured value most recently calculated by the number corresponding to the length of the sliding window. As described above, the sliding window of M length may include a buffer capable of holding the M measured values, and maintain the M measured values through the buffer.

In step S602, the slave node calculates a second estimate of the actual propagation time as the estimate using the average of the retained measurements and a first estimate calculated in a previous time unit. In this case, the slave node may calculate the second estimated value using Equation 1.

That is, the average of the continuous measured values is obtained by using the sliding window of M length, and the average of the current measured value and the most recent M-1 measured values among the average and the propagation time values are used. As shown in Equation 1, the second estimated value may be calculated as the estimated value.

Referring back to Equation 1, d _k denotes a k-th measurement value of propagation time, that is, a measurement value in time unit k, and D _k denotes a k-th estimate of actual propagation time. , An estimated value in the time unit k. In other words, the estimated value of the current time unit means the second estimated value, and the estimated value of the previous time unit means the first estimated value.

In addition, M is a predetermined length of the sliding window.

At this time, since the error occurs due to the use of a clock having a limited phase measurement granularity (d _k ), the difference between the d _k and d _kM by calculating the difference of the ratio to the M , The estimated value D _k may be calculated using the average of the most recent M measured values and the (k-1) -th estimated value D _k-1 .

H (z) represents a transfer function for a difference equation for the estimated value D _k .

Here, the length M of the slide window should be selected to be larger than the number of samples for changing the d _k . According to practical experiments, good results are obtained when M is chosen to be 32 or a larger number than 32.

7 is a block diagram for explaining an internal configuration of a device constituting a slave node according to the second embodiment of the present invention. As shown in FIG. 7, the device 700 includes a measured value calculator 710, an estimated value calculator 720, and an offset calculator 730.

The measured value calculator 710 calculates the measured value for the propagation time to the master node for each predetermined time unit by using the time stamp. In order to calculate a measurement value of a current time unit, the measurement value calculator 710 may record a first time stamp, which is a transmission time of a first message, in the first message and transmit it to the master node.

In this case, the master node transmits the first time stamp, the second time stamp that is the reception time of the first message, and a third time stamp that is the transmission time of the second message corresponding to the first message, to the second message. The device may be recorded and transmitted to the device 700 constituting the slave node, and the measurement value calculator 710 checks a fourth time stamp that is a reception time of the second message received from the device 700.

The measured value calculator 710 may calculate the measured value for the propagation time by using the first time stamp and the fourth time stamp.

The estimated value calculator 720 calculates an estimated value of the actual propagation time using the measured value and the sliding window. In this case, in order to calculate the estimated value, the estimated value calculator 720 may include a measured value retainer 721 and a second estimated value calculator 722.

The measured value holding unit 721 maintains the measured value most recently calculated by the number corresponding to the length of the sliding window. As described above, the sliding window of M length may include a buffer capable of holding the M measured values, and maintain the M measured values through the buffer.

The second estimated value calculator 722 calculates the second estimated value of the actual propagation time as the estimated value using the average of the retained measured values and the first estimated value calculated in a previous time unit. That is, the average of the continuous measured values is obtained by using the sliding window of M length, and the average of the current measured value and the most recent M-1 measured values among the average and the propagation time values are used. As shown in Equation 1, the second estimated value may be calculated as the estimated value.

Referring back to Equation 1, d _k denotes a k-th measurement value of propagation time, that is, a measurement value in time unit k, and D _k denotes a k-th estimate of actual propagation time. , An estimated value in the time unit k. In other words, the estimated value of the current time unit means the second estimated value, and the estimated value of the previous time unit means the first estimated value.

In addition, M is a predetermined length of the sliding window.

At this time, since the error occurs due to the use of a clock having a limited phase measurement granularity (d _k ), the difference between the d _k and d _kM by calculating the difference of the ratio to the M , The estimated value D _k may be calculated using the average of the most recent M measured values and the (k-1) -th estimated value D _k-1 .

H (z) represents a transfer function for a difference equation for the estimated value D _k .

Here, the length M of the slide window should be selected to be larger than the number of samples for changing the d _k . According to practical experiments, good results are obtained when M is chosen to be 32 or a larger number than 32.

The offset calculator 730 calculates an offset using the message transmission time at the master node, the message reception time at the slave node, and the estimated value.

As described above, when the time synchronization method and the device configuring the slave node are used, the measured values of the propagation time are calculated based on a time stamp for each predetermined time unit in the slave node, and the average of the measured values is calculated. And calculating the offset from the master node using the sliding window to improve the performance of jitter, wonder and time synchronization.

8 is a flowchart illustrating a method of performing time synchronization between nodes in a network according to a third embodiment of the present invention.

In operation S810, the slave node calculates a measurement value for the propagation time to the master node for each predetermined time unit by using the time stamp. In this case, in order to calculate the measured value in the current time unit, the slave node may be performed by including steps S811 to S814 in step S810.

In operation S811, the slave node records a first time stamp, which is a transmission time of a first message, in the first message and transmits it to the master node. That is, the slave node needs four times as described above to calculate the measurement for propagation time, and for this purpose sends the first message to the master node. In this case, the first time stamp corresponds to T _{1, k} of Equation 5.

In step S812, the master node checks a second time stamp that is a reception time of the first message. In this case, the second time stamp corresponds to T _{2, k} in Equation 5.

In operation S813, the master node records the first time stamp, the second time stamp, and a third time stamp, which is a transmission time of a second message corresponding to the first message, in the second message. Send to node. In this case, the third time stamp corresponds to T _{3, k} of Equation 5.

In operation S814, the slave node checks a fourth time stamp that is a reception time of the second message, and calculates a measurement value for the propagation time using the first time stamp and the fourth time stamp. . In this case, the fourth time stamp corresponds to T _{4, k} of Equation 5 _, and thus, the slave node may calculate the measured value for the propagation time in time unit k as shown in Equation 5.

In operation S820, the slave node calculates an estimated value of actual propagation time using the measured value and a linear digital filter. As described above, the step S820 of calculating the estimated value of the actual propagation time through the linear digital filter will be described in more detail with reference to FIG. 9.

In operation S830, the slave node calculates an offset using a message transmission time at the master node, a message reception time at the slave node, and the estimated value. In this case, the slave node may use Equation 9 to calculate the offset.

That is, subtracting the message transmission time from the master node from the message reception time of the slave node, the time until the message reaches the slave node remains. Here, except for the measured value of the propagation time calculated using the linear digital filter, the timing difference between the slave node and the master node may be known, and the time synchronization between the slave node and the master node may be used using the same. Can be performed.

9 is a flowchart illustrating a method of calculating an estimated value of actual propagation time using a linear digital filter according to a third embodiment of the present invention. As illustrated in FIG. 9, steps S901 to S903 may be performed by being included in step S820 described with reference to FIG. 8.

In step S901, the slave node maintains the measured value by a predetermined first number. Here, the first number may be modified according to the number of the measured values to be maintained.

In step S902, the slave node maintains a first estimated value measured in a previous time unit by a predetermined second number. Here, the second number may be modified according to the number of the first estimated values to be maintained.

In step S903, the slave node calculates a second estimated value that is an estimated value of the actual propagation time using the retained measured value, the first estimated value, and the linear digital filter. In this case, the linear digital filter may apply a plurality of filter coefficients to the held measured value and the first estimated value and calculate the sum as the second estimated value.

The sequence of measurement values is input to the general linear digital filter represented by Equation 7 to obtain an average of the continuously maintained measurement values of the propagation time using a general linear digital filter. The second estimate D _k may be calculated.

Where d _k is the measured value in _k -th time units, D _k is the second estimate in k-th time units, M is the length of the sliding window, and H (z) is a filter. In a filter transfer function, a _i (1 <i <n) and b _j (0 <j <m) mean the plurality of filter coefficients, respectively. In addition, n denotes the first number and m denotes the second number.

In addition, in order for the output of the linear digital filter to converge to the actual propagation time in Equation 7, the filter coefficient must satisfy the condition of Equation 8.

Here, the bandwidth of the linear digital filter should be smaller than the discrete frequency of the variation of dk.

10 is a block diagram for explaining an internal configuration of a device constituting a slave node according to the fourth embodiment of the present invention. As shown in FIG. 10, the device 1000 includes a measured value calculator 1010, an estimated value calculator 1020, and an offset calculator 1030.

The measured value calculator 1010 calculates the measured value for the propagation time to the master node for each predetermined time unit by using the time stamp. In order to calculate a measurement value of a current time unit, the measurement value calculator 1010 may record a first time stamp, which is a transmission time of a first message, in the first message and transmit it to the master node.

In this case, the master node transmits the first time stamp, the second time stamp that is the reception time of the first message, and a third time stamp that is the transmission time of the second message corresponding to the first message, to the second message. The device may be recorded and transmitted to the device 1000 constituting the slave node, and the measured value calculator 1010 checks a fourth time stamp that is a reception time of the second message received by the device 1000.

The measured value calculator 1010 may calculate the measured value for the propagation time by using the first time stamp and the fourth time stamp.

The estimated value calculator 1020 calculates an estimated value of the actual propagation time by using the measured value and the sliding window. In this case, in order to calculate the estimated value, the estimated value calculator 1020 may include a measured value retainer 1021, a first estimated value retainer 1022, and a second estimated value calculator 1023.

The measured value holding unit 1021 maintains the measured value by the first predetermined number. Here, the first number may be modified according to the number of the measured values to be maintained.

The first estimated value holding unit 1022 maintains the first estimated value measured in the previous time unit by a predetermined second number. Here, the second number may be modified according to the number of the first estimated values to be maintained.

The second estimated value calculator 1023 calculates a second estimated value that is an estimated value of the actual propagation time by using the maintained measured value, the first estimated value, and the linear digital filter. In this case, the linear digital filter may apply a plurality of filter coefficients to the retained measured value and the first estimated value and calculate the sum as the second estimated value. That is, the second estimated value calculator 1023 may calculate the second estimated value as shown in Equation (7).

The offset calculator 1030 calculates an offset using the message transmission time at the master node, the message reception time at the slave node, and the estimated value. In this case, the offset calculator 1030 may calculate the offset represented by Equation 6.

That is, the time taken for the message to reach the slave node from the master node can be calculated by subtracting the transmission time of the message from the reception time of the message, wherein the time calculated using the linear digital filter at the time By excluding the estimated value of the propagation time, the timing difference between the master node and the slave node can be confirmed, and thus time synchronization between the master node and the slave node can be performed.

As described above, using the method and the device for performing time synchronization according to the present invention, the measured values of the propagation time are calculated based on the time stamp for each predetermined time unit in the slave node, and the average of the measured values and a general linear digital filter are calculated. By calculating the offset from the master node using a liner digital filter, you can improve the performance of jitter, wonder, and time synchronization.

11 is an example for explaining a bidirectional method for time synchronization.

The master node 1101 transmits the first message including the transmission time T1 of the first message to all slave nodes except the master node of the master node 1101 for time synchronization. When all slave nodes receive the first message, the slave node checks the reception time T2 of the first message. In this case, if all the slave nodes know the propagation time with the master node 1101, the offset between all the slave nodes and the master node 1101 may be calculated, respectively, as described through Equation 6 above. Time synchronization can be performed.

One slave node 1102 of all the slave nodes transmits a second message to master node 1101 in time unit k to measure the propagation time. In this case, the second message may be transmitted including a time stamp of the moment when the second message is transmitted, that is, a transmission time T _1k 1103.

When the master node 1101 receives the second message, the master node 1101 checks a time stamp of the moment when the second message is received, that is, a reception time T _2k 1104, and is a response message to the second message. Send a message to the slave node 1102. In this case, the third message may include a transmission time T _1k 1103, a reception time T _2k 1104, and a time stamp of the moment when the third message is transmitted, that is, the transmission time T _3k 1105.

Upon receiving the third message, the slave node 1102 checks the time stamp at the moment of receiving the third message, that is, the reception time T _4k 1106, and includes the transmission time T _1k (included in the third message). 1103, the reception time T _2k 1104 and the transmission time T _3k 1105 can be confirmed.

Thus, the slave node 1102 obtains the transmission time T _1k (1103), the reception time T _2k (1104), the transmission time T _3k (1105), and the reception time T _4k (1106) for the time unit k. Through Equation 6, the measured value d _k of the propagation delay in the time unit k may be calculated.

Then, the estimated value D _k of the actual propagation delay can be obtained through a plurality of measured values using a sliding window of M length or a linear digital filter. That is, when using the sliding window of M length, the slave node 1102 calculates the estimated value D _k as shown in Equation 1 through the average of the M measured values and the estimated value D _k-1 of the previous time step. Can be. In addition, when using the linear digital filter, the slave node 1102 may calculate the estimated value D _k through m measured values, n most recent estimated values, and filter coefficients.

In addition, the offset in the time unit k can be calculated using the estimated value D _k , and time synchronization between the master node 1101 and the slave node 1102 can be performed as described above. This process is repeated for each time unit, and continuous time synchronization is performed on the network.

So far, the bidirectional time synchronization method using the sliding window or the linear digital filter has been described with reference to FIGS. 5 to 11. 12 to 14 will be described a one-way time synchronization method that can calculate the offset between the master node and the slave node in a single message transmission.

12 is a flowchart illustrating a method of performing time synchronization between nodes in a network according to a fifth embodiment of the present invention.

In operation S1210, the slave node calculates a first offset from the master node for each predetermined time unit by using a time stamp. In this case, as shown in FIG. 12, the slave node may be performed by including steps S1211 to S1214 in step S1210.

In step S1211, the master node writes a first time stamp, which is a transmission time of the message, to the slave node.

In step S1212, the relay node records the accumulated value of the first propagation time of the link receiving the message and the dwell time at the relay node in the message.

In step S1213, the slave node checks the second propagation time of the link existing between the second time stamp, which is the reception time of the message, and the last relay node.

In operation S1214, the slave node calculates the first offset using the first time stamp, the second time stamp, the accumulated value, and the second propagation time. In this case, the slave node may use Equation 9 to calculate the first offset by using the first time stamp, the second time stamp, the accumulated value, and the second propagation time.

First Offset = T2-T1-R1-D1

Here, T2 may mean the second time stamp, T1 may mean the first time stamp, R1 may mean the accumulated value, and D1 may mean the second propagation time.

In operation S1220, the slave node calculates a second offset using the first offset and the linear digital filter. In this case, the slave node may calculate the second offset using Equation 10 below by using a predetermined number of offsets measured in the first offset and the previous time unit to calculate the second offset.

Where u _k is the first offset, y _k is the second offset, a _i (1 <i <n) and b _j (0 <j <m) are filter coefficients of the linear digital filter, H (z) may mean a filter transfer function, respectively.

13 is a block diagram for explaining an internal configuration of a device constituting a slave node according to the sixth embodiment of the present invention. As shown in FIG. 13, the device 1300 includes a first offset calculator 1310 and a second offset calculator 1320.

The first offset calculator 1310 calculates the first offset from the master node for each predetermined time unit by using the time stamp in the slave node. In order to calculate the first offset, the first offset calculator 1310 receives a message from a master node and confirms a first time stamp as a transmission time of the message, as shown in FIG. 13. 1313, a cumulative value checking unit 1312 for checking accumulated values of a first propagation time and a dwell time in the message, and a reception time checking unit 1313 for confirming a second time stamp that is a reception time of the message. And a second propagation time verification unit 1314 and a first propagation time, the second time stamp, the accumulated value, and the second propagation time for confirming a second propagation time of a link existing between the last relay nodes. It may include a calculation unit 1315 for calculating the first offset.

In this case, the first propagation time may be a propagation time of a link receiving the message at the relay node, and the dwell time may be a dwell time of the message at the relay node. In addition, the calculator 1315 may calculate the first offset as shown in Equation (9).

The second offset calculator 1320 calculates a second offset using the first offset and the linear digital filter. In this case, the second offset calculator 1320 may calculate the second offset as shown in Equation 10 using the first offset and a predetermined number of offsets measured in a previous time unit.

As described above, using the time synchronization method and apparatus according to the present invention, in the system for measuring the propagation time in one direction, the first offset from the master node is calculated for each predetermined time unit, and the first offset and the previous time unit are calculated. By calculating a second offset, which is an offset of the current unit of time, through a linear digital filter that uses the calculated offsets as an input value, the performance of jitter, wonder, and time synchronization can be improved while considering the dwell time in the relay device.

14 is an example for explaining a one-way method for time synchronization.

The master node 1401 transmits a message T _1k to the third slave node 1404 through the first slave node 1402 and the second slave node 1403 to perform time synchronization in time unit k. And send the message 1405. In this case, the first slave node 1402 checks the propagation time 1406 of the link receiving the message and the dwell time 1407 of the first slave node 1402, and accumulates the message, and stores the message. Is sent to the next node. The second slave node 1403, which is the next node, also checks the dwell time 1408 of the second slave node 1403 and the propagation time 1409 of the link that received the message, and accumulates the message in the message. Send to node. In this case, the message includes a cumulative value R _1k in which a propagation time 1406, a residence time 1407, a residence time 1408, and a propagation time 1409 are accumulated.

The third slave node 1404, which is the next node and the destination, checks the propagation time D _1k 1410 of the link with the second slave node 1403, which is the last relay node, and receives the message at the received time T. Check _2k (1411).

At this time, the third slave node 1404 satisfies all the parameters of Equation 9 for calculating the first offset u _k . That is, the third slave node 1404 may calculate the first offset u _k . As such, the third slave node 1404 maintains the first offset u _k calculated during the time unit k in the most recently calculated order, and the second offset calculated in the previous time unit is also n. Keep your dog.

That is, the third slave node 1404 may calculate a second offset, which is an offset in the time unit k, through Equation 10, and the master node 1401 and the third slave node through the second offset. Time synchronization between 1404 is performed. The time synchronization can be performed through the same process as all other slave nodes, and the time synchronization of the network is performed.

Embodiments according to the present invention may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be those specially designed and constructed for the present invention or may be available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as a hard disk, a floppy disk, and a magnetic tape; optical media such as CD-ROM and DVD; magnetic recording media such as a floppy disk; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

As described above, the present invention has been described by specific embodiments such as specific components and the like. For those skilled in the art, various modifications and variations are possible from these descriptions.

Accordingly, the spirit of the present invention should not be construed as being limited to the embodiments described, and all of the equivalents or equivalents of the claims, as well as the following claims, belong to the scope of the present invention .

According to the present invention, measured values of propagation are calculated based on a time stamp for each predetermined time unit in a slave node, and the average and sliding window of the measured values are calculated. By calculating the offset from the master node, the performance of jitter, wander and time synchronization can be improved.

According to the present invention, the slave node calculates the measurement values of the propagation time based on the time stamp for each predetermined time unit, and uses the average of the measurements and a general linear digital filter to master node. By calculating the offset from, you can improve the performance of jitter, wonder, and time synchronization.

According to the present invention, in a system for measuring the propagation time in one direction, linear digital is calculated by calculating a first offset from a master node for each predetermined time unit, and using the first offset and offsets calculated in a previous time unit as input values. By calculating the second offset, which is the offset of the current time unit, through the filter, the performance of jitter, wonder, and time synchronization can be improved while considering the residence time in the relay device.

Claims (26)

In a method for performing time synchronization between nodes in a network, Calculating a measured value for a propagation time to a master node by a predetermined time unit using a time stamp at a slave node; Calculating an estimate of an actual propagation time using the measured value and a sliding window at the slave node; And Calculating an offset using a message transmission time at the master node, a message reception time at the slave node, and the estimated value Including, The step of calculating the measurement value for the propagation time to the master node by the predetermined time unit using the time stamp in the slave node, Transmitting, by the slave node, a first time stamp, which is a transmission time of a first message, to the master node by recording a first time stamp in the first message; And Calculating a measurement value for the propagation time by receiving a second message corresponding to the first message from the master node at the slave node; Time synchronization method comprising a. The method of claim 1, The calculating of the estimated propagation time using the measured value and the sliding window at the slave node, Maintaining the measured value most recently calculated by the number corresponding to the length of the sliding window; And Calculating a second estimated value of the actual propagation time as the estimated value using the average of the maintained measured values and a first estimated value calculated in a previous time unit. Time synchronization method comprising a. 3. The method of claim 2, Computing the average of the maintained measured value, and using the average and the first estimate calculated in the previous time unit, calculating the second estimate of the actual propagation time, A time synchronization method for calculating the second estimated value as shown in Equation (11).

Where d _k is the measured value in _k -th time unit, D _k is the estimated value in k-th time unit, M is the length of the sliding window, and H (z) is D _k. Means the transfer function of the difference equation. In a method for performing time synchronization between nodes in a network, Calculating a measurement value for the propagation time to the master node by the predetermined time unit using the time stamp at the slave node; Calculating an estimate of an actual propagation time at the slave node using the measured value and a linear digital filter; And Calculating an offset using a message transmission time at the master node, a message reception time at the slave node, and the estimated value Including, The step of calculating the measurement value for the propagation time to the master node by the predetermined time unit using the time stamp in the slave node, Transmitting, by the slave node, a first time stamp, which is a transmission time of a first message, to the master node by recording a first time stamp in the first message; And Calculating a measurement value for the propagation time by receiving a second message corresponding to the first message from the master node at the slave node; Time synchronization method comprising a. 5. The method of claim 4, The calculating of the estimated propagation time using the measured value and the linear digital filter at the slave node, Maintaining the measured value by a first predetermined number; Maintaining a first estimated value measured in a previous time unit by a second predetermined number; And Calculating a second estimated value that is an estimated value of the actual propagation time using the held measurement, the first estimated value, and the linear digital filter. Including, And said linear digital filter applies a plurality of filter coefficients to said held measurement and said first estimate and calculates the sum as said second estimate. The method of claim 5, The step of calculating a second estimated value that is an estimated value of the actual propagation time using the retained measured value, the first estimated value and the linear digital filter, A time synchronization method for calculating the second estimate value as shown in Equation 12 below.

Where d _k is the measured value in _k -th time units, D _k is the second estimate in k-th time units, M is the length of the sliding window, and H (z) is a filter. A filter transfer function, wherein a _i (1 <i <n) and b _j (0 <j <m) is the plurality of filter coefficients, n is the first number, and m is Each of the second numbers. The method according to claim 6, The a _i (1 <i <n) and the b _j (0 <j <m) satisfy the following equation (13).

The method according to claim 1 or 4, The step of receiving a second message corresponding to the first message from the master node at the slave node to calculate a measurement value for the propagation time, Confirming, by the master node, a second time stamp that is a reception time of the first message; Recording, by the master node, a third time stamp, which is a transmission time of the first time stamp, the second time stamp, and a second message corresponding to the first message, in the second message and transmitting to the slave node; ; And Confirming, by the slave node, a fourth time stamp that is a reception time of the second message, and calculating a measurement value for the propagation time using the first time stamp and the fourth time stamp Time synchronization method comprising a. 9. The method of claim 8, The step of calculating a measurement value for the propagation time using the first time stamp and the fourth time stamp, A time synchronization method for calculating the measured value as shown in equation (14).

Here, d _k is the measured value, T _{1, k} is the first time stamp, T _{2, k} is the second time stamp, T _{3, k} is the third stamp, T ₄ , _k denotes the fourth time stamp, and k denotes that the first to fourth time stamps are measured in k-th time units, respectively. The method according to claim 1 or 4, The step of calculating an offset using the message transmission time at the master node, the message reception time at the slave node and the estimated value may include: A time synchronization method for calculating the offset as shown in the following equation (15). Offset = T2-T1-D1 Here, T2 denotes the message reception time, T1 denotes the message transmission time, and D1 denotes the estimated value. In a method for performing time synchronization between nodes in a network, Calculating a first offset from the master node for each predetermined time unit using a time stamp at the slave node; And Calculating a second offset using the first offset and the linear digital filter Including, Computing the first offset from the master node for each predetermined time unit using the time stamp at the slave node, Transmitting, by a master node, a first time stamp, which is a transmission time of a message, to the slave node by recording in the message; And Calculating the first offset at the slave node using the first propagation time of the link that received the message at the relay node, the dwell time at the relay node, and the second propagation time of the link existing between the last relay node; step Time synchronization method comprising a. 12. The method of claim 11, Calculating the first offset at the slave node using the first propagation time of the link that received the message at the relay node, the dwell time at the relay node, and the second propagation time of the link existing between the last relay node; The step, Recording, in the message, a value in which the first propagation time of the link that received the message at the relay node and the dwell time at the relay node are accumulated; Confirming, at the slave node, a second time stamp that is a reception time of the message and the second propagation time of a link existing between the last relay node; And Calculating the first offset using the first time stamp, the second time stamp, the accumulated value, and the second propagation time at the slave node. Time synchronization method comprising a. The method of claim 12, The calculating of the first offset using the first time stamp, the second time stamp, the accumulated value, and the second propagation time at the slave node may include: A time synchronization method for calculating the first offset as shown in equation (16). First Offset = T2-T1-R1-D1 Here, T2 denotes the second time stamp, T1 denotes the first time stamp, R1 denotes the accumulated value, and D1 denotes the second propagation time. 12. The method of claim 11, The step of calculating a second offset using the first offset and the linear digital filter, And calculating the second offset using Equation 17 using the first offset and a predetermined number of offsets measured in a previous time unit.

Where u _k is the first offset, y _k is the second offset, a _i (1 <i <n) and b _j (0 <j <m) are filter coefficients of the linear digital filter, H (z) means a filter transfer function, respectively. A computer-readable recording medium storing a program for executing the method according to any one of claims 11 to 14. In the device constituting the slave node for performing time synchronization between the nodes of the network, A measured value calculator for calculating a measured value for a propagation time to a master node by using a time stamp for each predetermined time unit; An estimated value calculator for calculating an estimated value of actual propagation time using the measured value and the sliding window; And An offset calculator configured to calculate an offset using a message transmission time at the master node, a message reception time at the slave node, and the estimated value Including, The measured value calculation unit, A first message transmitter for recording a first time stamp, which is a transmission time of a first message, in the slave node in the first message and transmitting the same to the master node; And A first calculator configured to receive a second message corresponding to the first message from the master node at the slave node and calculate a measurement value for the propagation time Apparatus comprising a. 17. The method of claim 16, The estimated value calculator, A measurement value holding unit which holds the measurement value most recently calculated by the number corresponding to the length of the sliding window; And A second estimated value calculator for calculating a second estimated value of the actual propagation time as the estimated value using the average of the measured values and the first estimated value calculated in a previous time unit; Apparatus comprising a. 18. The method of claim 17, The second estimated value calculator, The apparatus for calculating the second estimated value as shown in Equation 18.

Where d _k is the measured value in _k -th time unit, D _k is the estimated value in k-th time unit, M is the length of the sliding window, and H (z) is D _k. Means the transfer function of the difference equation. In the device constituting the slave node for performing time synchronization between the nodes of the network, A measured value calculator for calculating a measured value for a propagation time to a master node by using a time stamp for each predetermined time unit; An estimated value calculator for calculating an estimated value of an actual propagation time using the measured value and the linear digital filter; And An offset calculator configured to calculate an offset using a message transmission time at the master node, a message reception time at the slave node, and the estimated value Including, The measured value calculation unit, A first message transmitter for recording a first time stamp, which is a transmission time of a first message, in the slave node in the first message and transmitting the same to the master node; And A first calculator configured to receive a second message corresponding to the first message from the master node at the slave node and calculate a measurement value for the propagation time Apparatus comprising a. 20. The method of claim 19, The estimated value calculator, A measurement value holding unit holding the measurement value by a first predetermined number; A first estimation value holding unit which maintains a first estimated value measured in a previous time unit by a predetermined second number; And A second estimated value calculator for calculating a second estimated value that is an estimated value of the actual propagation time by using the held measured value, the first estimated value, and the linear digital filter Including, And said linear digital filter applies a plurality of filter coefficients to said held measurement and said first estimate and calculates the sum as said second estimate. 21. The method of claim 20, The second estimated value calculator, The apparatus for calculating the second estimated value as shown in Equation 19.

Where d _k is the measured value in _k -th time units, D _k is the second estimate in k-th time units, M is the length of the sliding window, and H (z) is a filter. A filter transfer function, wherein a _i (1 <i <n) and b _j (0 <j <m) is the plurality of filter coefficients, n is the first number, and m is Each of the second numbers. 22. The method of claim 21, Wherein a _i (1 <i <n) and b _j (0 <j <m) satisfy Equation 20 below.

The method of claim 16 or 19, The first calculation unit, A second time stamp checking unit for checking a second time stamp that is a reception time of the first message in the master node; Recording, by the master node, a third time stamp, which is a transmission time of the first message, the second time stamp, and the second message corresponding to the first message, in the second message and transmitting the same to the slave node; A second message transmitter; And A second calculator configured to check a fourth time stamp that is a reception time of the second message in the slave node, and calculate a measurement value for the propagation time using the first time stamp and the fourth time stamp; Apparatus comprising a. In the device constituting the slave node for performing time synchronization between the nodes of the network, A first offset calculator configured to calculate a first offset from the master node for each predetermined time unit by using a time stamp in the slave node; And A second offset calculator configured to calculate a second offset using the first offset and the linear digital filter Including, The first offset calculator, A first time stamp confirming unit for recording a first time stamp, which is a transmission time of a message in a master node, in the message and transmitting the same to the slave node; And Calculating the first offset at the slave node using the first propagation time of the link that received the message at the relay node, the dwell time at the relay node, and the second propagation time of the link existing between the last relay node; First calculation unit Apparatus comprising a. 25. The method of claim 24, The first calculation unit, A cumulative value checking unit that checks the accumulated values of the first propagation time and the residence time in the message; A reception time confirmation unit confirming a second time stamp that is a reception time of the message; A second propagation time confirmation unit for confirming the second propagation time of a link existing between the last relay node; And A second calculator configured to calculate the first offset using the first time stamp, the second time stamp, the accumulated value, and the second propagation time Including, The first propagation time is a propagation time of a link receiving the message at a relay node, The dwell time is the dwell time of the message at the relay node. 25. The method of claim 24, The second offset calculator, And calculate a second offset as shown in Equation 21 using the first offset and a predetermined number of offsets measured in a previous time unit.

Where u _k is the first offset, y _k is the second offset, a _i (1 <i <n) and b _j (0 <j <m) are filter coefficients of the linear digital filter, H (z) means a filter transfer function, respectively.

Images