JP2009077207A - Client device and synchronization system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To achieve stable frequency-time synchronization even under a network environment having fluctuation in transmission delay and path switching. <P>SOLUTION: An RTT (round trip time) is filtered to remove packets largely deviated from a mean value of a delay time, client time is controlled by using only the data approximate to the mean value of the delay time. Alternatively, an RTT is filtered, and the client time is controlled by using only the data within an error range estimated from the previous RTT. Alternatively, when an RTT or a one-side delay time suddenly changes, it is considered that path switching is caused, and the process shifts to a hold-over operation. Further, an offset value is donated so as to cancel a change amount at that time. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a client device that synchronizes frequency or time with a server that generates a reference time. In the following description, a server that generates a reference time is referred to as a synchronization server, and a client device that synchronizes frequency or time with the synchronization server is referred to as a synchronization client.

NTP (Network that is used as a standard synchronization method in packet networks)
The basic principles of Time Protocol (see Non-Patent Document 1) and SNTP (Simple Network Time Protocol: see Non-Patent Document 2) are shown in FIG.

  The synchronization client transmits a synchronization request packet in which the transmission time T1 is written to the synchronization server. The synchronization server receives the synchronization request packet and transmits a synchronization response packet to the synchronization client. In addition to the transmission time T1, the synchronization response packet includes a time T2 when the synchronization request packet is received and a time T3 when the synchronization response packet is transmitted.

The time when the synchronization client receives the synchronization response packet is T4. Here, T1 and T4 are times measured by the client time generation unit, and T2 and T3 are times measured by the server time generation unit. The effective round trip time RTT (Round Trip Time) required for the packet to reciprocate between the synchronous client and the synchronous server is RTT = (T4-T1)-(T3-T2) = T4-T3 + T2-T1
Given in. If the delay time of the synchronization request packet is D1, and the delay time of the synchronization response packet is D2,
D1 + D2 = RTT
It becomes. Assuming that the delay time of the sync request packet is equal to the delay time of the sync response packet,
D1 = D2 = RTT / 2
It becomes. In this case, the error between the client time generation unit and the server time generation unit is
dT = T4- (T3 + D2) = (T1-T2-T3 + T4) / 2
Can be estimated. By controlling the client time generation unit so that the value of dT becomes 0, time synchronization can be established.

D. W. Mills: “Network Time Protocol (Version 3) Specification, Implementation and Analysis”, RFC 1305, IETF. D. Mills: “Simple Network Time Protocol (SNTP) Version 4 for IPv4, IPv6 and OSI”, RFC2030, IETF.

When the delay time of the synchronization request packet and the delay time of the synchronization response packet change by ΔD1 and ΔD2 due to jitter in the network and path switching, respectively, the time difference between the client time generation unit and the server time generation unit is dT = (T1-T2- T3 + T4) / 2 + (ΔD2-ΔD1) / 2
That is, it appears that a change has occurred by (ΔD2−ΔD1) / 2. If this is used as it is and the client time generating unit is controlled, the client time generating unit will also be affected and fluctuate.

  The present invention has been made under such a background, and an object thereof is to provide a synchronization client and a synchronization system capable of removing the influence when the packet propagation delay time changes.

  In the present invention, in order to solve the above-mentioned problem, filtering for RTT is performed, packets that deviate significantly from the average value of delay time are removed, and control of the client time generation unit is performed using only data in the vicinity of the average value of delay time I do.

  Alternatively, RTT filtering is performed, and the client time generation unit is controlled using only data that falls within the error range estimated from the previous RTT.

  Alternatively, when the RTT or the one-side delay time changes abruptly, it is considered that path switching has occurred, and a transition is made to the holdover operation. That is, by controlling the client time generating unit using only data in the vicinity of the average value of the propagation delay time, fluctuations in the client time generating unit can be reduced.

  Alternatively, the fluctuation of the client time generation unit can be reduced by controlling the client time generation unit using only data that falls within the error range estimated from the previous RTT.

  Alternatively, when the RTT changes abruptly, it is considered that the path switching has occurred, and the transition to the holdover operation can be performed to prevent a sudden phase and frequency fluctuation of the output signal of the client time generation unit due to the path switching.

  Alternatively, if the RTT or the one-side delay time changes drastically, it is considered that path switching has occurred, and the change in dT at that time (the amount corresponding to (ΔD2−ΔD1) / 2 described above) is cancelled. By providing an offset value for, it is possible to prevent sudden phase and frequency fluctuations in the output signal of the client time generation unit due to path switching.

  That is, the present invention includes means for transmitting a synchronization request signal to a synchronization server that generates a reference time, means for receiving a synchronization response signal in response to the synchronization request signal from the synchronization server, and the received synchronization response. It is a synchronous client provided with a means (corresponding to the client time generator) that generates a frequency or time synchronized with a reference time based on a signal.

Here, the feature of the present invention is that the time when the synchronization client transmits the synchronization request signal to the synchronization server is T1, the time when the synchronization server receives the synchronization request signal is T2, and the synchronization server When the time at which the synchronization response signal in response to the synchronization request signal is transmitted is T3 and the time at which the own synchronization client receives the synchronization response signal is T4, the frequency in the own synchronization client or the reference time is referred to The time difference dT applied when setting the time is dT = (T1-T2-T3 + T4) / 2.
The round-trip delay time RTT between the synchronization client and the synchronization server when the synchronization request signal is the forward signal and the synchronization response signal is the return signal is RTT = T4-T3 + T2-T1
Means for applying the time difference dT to the frequency or time setting in the own synchronization client only for the time difference dT when the value of the RTT becomes a predetermined value (the client time generation unit has The predetermined value is calculated by calculating an average value RTT_mean of the RTT before receiving the synchronization response signal, and setting a preset jitter tolerance value as δ,
RTT_lim1 = RTT_mean + δ
RTT_lim2 = RTT_mean-δ
RTT_lim2 <RTT <RTT_lim1
There is a value of RTT that satisfies the above.

Alternatively, the predetermined value is the RTT of the synchronous response signal in which the RTT has reached the predetermined value immediately before receiving the synchronous response signal, RTT_prev, the preset allowable normalized frequency difference is y, and the synchronous response signal is received. The elapsed time from the application time of the immediately preceding time difference dT is T,
RTT_lim = RTT_prev + y × T
RTT <RTT_lim
RTT value satisfying

Alternatively, the predetermined value may be set such that the RTT of the synchronous response signal at which the RTT has reached the predetermined value immediately before receiving the synchronous response signal is RTT_prev, the preset allowable normalized frequency difference is y1 and y2, and the synchronous response signal is The elapsed time from the application time of the time difference dT immediately before reception is T,
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev−y2 × T
RTT_lim2 <RTT <RTT_lim1
RTT value satisfying

Alternatively, the predetermined value is a constant that is set in advance, with RTT_prev being the RTT of the synchronous response signal whose RTT has reached the predetermined value immediately before receiving the synchronous response signal, y1 and y2 being a preset allowable normalized frequency difference Is γ (0 <γ <1), and T is the elapsed time from the application time of the time difference dT immediately before receiving the synchronization response signal.
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev × (1-γ) −y2 × T
RTT_lim2 <RTT <RTT_lim1
RTT value satisfying

  The present invention can also be viewed from the viewpoint of a synchronization system. That is, the present invention is a synchronization system including a synchronization server that generates a reference time and the synchronization client of the present invention.

  The present invention can also be viewed from the viewpoint of a program. That is, the present invention is a program that, when installed in a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize functions corresponding to at least some of the functions of the synchronous client of the present invention.

  By recording the program of the present invention on a recording medium, the general-purpose information processing apparatus can install the program of the present invention using this recording medium. Alternatively, the program of the present invention can be directly installed on the general-purpose information processing apparatus via a network from a server that holds the program of the present invention.

  Thereby, the function corresponding to the function of at least one part of the synchronous client of this invention is realizable using a general purpose information processing apparatus.

  The program of the present invention includes not only a program that can be directly executed by a general-purpose information processing apparatus but also a program that can be executed by installing it on a hard disk or the like. Also included are those that are compressed or encrypted.

  According to the present invention, stable frequency / time synchronization can be realized even in a network environment where transmission delay fluctuations and path switching occur.

(First Example)
FIG. 2 shows the configuration of a frequency / time synchronization system using a synchronization client according to the present invention. The synchronization client 10 transmits the synchronization request packet in which the transmission time T1 is written from the synchronization request packet transmission unit 2 to the synchronization server 20.

  The synchronization server 20 receives the synchronization request packet at the synchronization request packet receiver 13 and transmits the synchronization response packet from the synchronization response packet transmitter 12 to the synchronization client 10.

  In addition to T1, the synchronization response packet includes time T2 when the synchronization request packet receiver 13 receives the synchronization request packet and time T3 when the synchronization response packet transmitter 12 transmits the synchronization response packet.

  The time when the synchronization client 10 receives the synchronization response packet at the synchronization response packet receiver 3 is T4. Here, T1 and T4 are times generated by the client time generating unit 1, and T2 and T3 are times generated by the server time generating unit 11.

The client time control unit 4 calculates the average value RTT_mean of the previous round-trip delay time based on T1, T2, T3, and T4, and sets a preset jitter tolerance value as δ.
RTTlim1 = RTTmean + δ
RTTlim2 = RTTmean−δ
RTTlim2 <RTT <RTTlim1
Only the time difference dT of the synchronization response signal satisfying the above condition is transmitted, and the RTT filtering process for applying the setting to the frequency or time of the synchronization client is performed only for the time difference dT.

The above algorithm is described in detail as follows.
S1 RTT_mean = {RTT [i-1] + RTT [i-2] + ... + RTT [i-M]} / M
S2 RTT_lim1 = RTT_mean + δ
RTT_lim2 = RTT_mean-δ
S3 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S4 if (RTT_mean-δ <RTT [i] <RTT_mean + δ) then S5 else S8
S5 dT [i] = (T1 [i] -T2 [i] -T3 [i] + T4 [i]) / 2
S6 i ++
S7 → S1
S8 i ++
S9 → S3
For example, a value such as 1/10 of the standard deviation of jitter is used as δ.

  The state of the above RTT filtering operation is shown in FIG. In FIG. 3 to FIG. 13, black circles indicate RTTs that are transmitted through the RTT filter (used for control), and white circles indicate RTTs that are not transmitted through the RTT filter (not used for control).

  The client time control unit 4 controls the phase and frequency of the client time generation unit 1 so that the value of dT approaches the target value.

  Note that the target value of dT may be 0 (in the case of time synchronization), and may be a value other than 0 (in the case of frequency synchronization). The same applies to the following embodiments.

(Second embodiment)
The configuration of the frequency and time synchronization system using the synchronization client in the second embodiment is the same as that in the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 sets RTT_prev as the round-trip delay time of the synchronization response signal transmitted immediately before, y as the preset allowable normalized frequency difference, and time difference dT as the synchronization response signal immediately before. The elapsed time from the time at which the
RTT_lim = RTT_prev + y × T
RTT <RTT_lim
RTT filtering processing for transmitting only the time difference dT of the synchronous response signal satisfying the above condition is performed. Since the frequency resolution of the oscillator is not zero, there is always a frequency error instantaneously. Therefore, a phase error corresponding to the frequency error (normalized frequency error) × elapsed time also occurs. If the jitter is smaller than the estimated phase error, it means that the data is determined to be valid.

The above algorithm is described in detail as follows.
S11 RTT_prev = RTT [i-1]
S12 T4_prev = T4 [i-1]
S13 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S14 RTT_lim = RTT_prev + y × (T4 [i] −T4_prev)
S15 if (RTT [i] <RTT_lim) then S16 else S19
S16 dT [i] = (T1 [i] -T2 [i] -T3 [i] + T4 [i]) / 2
S17 i ++
S18 → S11
S19 i ++
S20 → S13
As y, for example, a value such as a normalized frequency resolution of the oscillator used in the client time generation unit (a value obtained by normalizing the frequency resolution with the nominal frequency) is used.

  The state of the above RTT filtering operation is shown in FIG.

(Third embodiment)
The configuration of the frequency and time synchronization system using the synchronization client in the third embodiment is the same as that in the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 sets the round-trip delay time of the synchronization response signal transmitted immediately before to RTT_prev, the preset allowable normalized frequency difference to y1 and y2, and immediately before the synchronization response signal The elapsed time from the time when the time difference dT is transmitted is T,
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev−y2 × T
RTT_lim2 <RTT <RTT_lim1
RTT filtering processing for transmitting only the time difference dT of the synchronous response signal satisfying the above condition is performed. Since the frequency resolution of the oscillator is not zero, there is always a frequency error instantaneously. Therefore, a phase error corresponding to the frequency error (normalized frequency error) × elapsed time also occurs.

  If the jitter is smaller than the estimated phase error, it means that the data is determined to be valid.

The above algorithm is described in detail as follows.
S21 RTT_prev = RTT [i-1]
S22 T4_prev = T4 [i-1]
S23 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S24 RTT_lim1 = RTT_prev + y1 × (T4 [i] −T4_p rev)
RTT_lim2 = RTT_prev−y2 × (T4 [i] −T4_prev)
S25 if (RTT_lim2 <RTT [i] <RTT_lim1) then S26 else S29
S26 dT [i] = (T1 [i] -T2 [i] -T3 [i] + T4 [i]) / 2
S27 i ++
S28 → S21
S29 i ++
S30 → S23
As y1 and y2, for example, a value such as a normalized frequency resolution of the oscillator used in the client time generation unit (a value obtained by normalizing the frequency resolution with a nominal frequency) is used.

  In this embodiment, when the RTT decreases instantaneously, the dT value that passes through the RTT filter for a long period of time disappears, and an extremely small RTT packet is ignored in order to prevent the operation from becoming unstable. Like that.

The stochastic equilibrium point of the transmission RTT value is considered that the integral value (area) of the probability density function of the RTT of the average window having the smaller RTT than the equilibrium point is equal to the area of the average window having the larger RTT. . For a probability density function with a single peak, such as a Gaussian distribution,
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev−y2 × T
y2> y1
If the window with the smaller RTT is made larger, the equilibrium point can be controlled so that the RTT becomes smaller than the peak.

  The state of the above RTT filtering operation is shown in FIG.

  The client time control unit 4 controls the phase and frequency of the client time generation unit 1 so that the value of dT approaches the target value.

(Fourth embodiment)
The configuration of the frequency and time synchronization system using the synchronization client in the fourth embodiment is the same as that in the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 sets RTT_prev as the round-trip delay time of the synchronization response signal transmitted immediately before, y1 and y2 as the preset allowable normalized frequency difference, and γ as the preset constant. (0 <γ <1), where T is the time elapsed since the time when the time difference dT of the synchronization response signal was transmitted immediately before,
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev × (1-γ) −y2 × T
RTT_lim2 <RTT <RTT_lim1
RTT filtering processing for transmitting only the time difference dT of the synchronous response signal satisfying the above condition is performed.

  Since the frequency resolution of the oscillator is not zero, there is always a frequency error instantaneously. Therefore, a phase error corresponding to the frequency error (normalized frequency error) × elapsed time also occurs. If the jitter is smaller than the estimated phase error, it means that the data is determined to be valid.

The above algorithm is described in detail as follows.
S41 RTT_prev = RTT [i-1]
S42 T4_prev = T4 [i-1]
S43 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S44 RTT_lim1 = RTT_prev + y1 × (T4 [i] −T4_prev)
S45
S46 dT [i] = (T1 [i] -T2 [i] -T3 [i] + T4 [i]) / 2
S47 i ++
S48 → S41
S49 i ++
S50 → S43
As y1 and y2, for example, a value such as a normalized frequency resolution of the oscillator used in the client time generation unit (a value obtained by normalizing the frequency resolution with a nominal frequency) is used. Moreover, even if y1 and y2 are about the same,
RTT [i]> RTT_prev
RTT [i] <RTT_prev
Can be controlled so that the RTT becomes smaller.

In this embodiment, when the RTT instantaneously decreases, the dT value that passes through the RTT filter for a long period of time disappears, and an extremely small RTT packet is ignored in order to prevent the operation from becoming unstable. Like to do. The stochastic equilibrium point of the transmission RTT value is considered that the integral value (area) of the probability density function of the RTT of the average window having the smaller RTT than the equilibrium point is equal to the area of the average window having the larger RTT. . For a probability density function with a single peak, such as a Gaussian distribution,
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev × (1-γ) −y2 × T
(Γ is a positive value sufficiently smaller than 1)
If the window with the smaller RTT is made larger, the equilibrium point can be controlled so that the RTT becomes smaller than the peak.

  The state of the above RTT filtering operation is shown in FIG.

(Fifth embodiment)
The configuration of the frequency / time synchronization system using the synchronization client in the fifth embodiment is the same as that in the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 adds the round-trip delay time of the synchronous response signal transmitted immediately before to RTT_prev and the threshold value RTT_th in addition to the processing steps described in the first to fourth embodiments. As
| RTT-RTT_prev |> RTT_th
Is continued N times or more (N is the number of protection stages for sudden large jitter) or more, it is determined that a change in the round-trip delay time due to network path switching has occurred, and a holdover operation is performed. Also, let RTT_prev be the round-trip delay time before transition to the holdover state, y1 and y2 be a preset allowable normalized frequency difference, and T be the elapsed time from the transition to the holdover state.
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev−y2 × T
RTT_lim2 <RTT <RTT_lim1
When the above condition is satisfied, it is determined that the network path has returned to its original state, and a transition is made from the holdover operation to the client time control operation.

The above algorithm is described in detail as follows.
S51 RTT_prev = RTT [i-1]
S52 T4_prev = T4 [i-1]
S53 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T [i]
S54 if (| RTT [i] −RTT_prev | ≦ RTT_th) then S55 else S58
S55 dT [i] = (T1 [i] -T2 [i] -T3 [i] + T4 [i]) / 2
S56 i ++
S57 → S1
S58 K = 0
S59 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S60 if (| RTT [i] −RTT_prev | ≦ RTT_th) then S55 else S61
S61 K ++
S62 if (K <N) then S63 else S65
S63 i ++
S64 → S69
S65 Transition to holdover state S66 T_ho_start = T4 [i]
S67 i ++
S68 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S69 RTT_lim1 = RTT_prev + y1 × (T4 [i] −T_ho_start)
RTT_lim2 = RTT_prev−y2 × (T4 [i] −T_ho_start)
S70 if (RTT_lim2 <RTT [i] <RTT_lim1) then S55 else S67
As RTT_th, for example, a value of standard deviation of jitter × 3 is used. In S51, an average value can be used instead of the immediately preceding instantaneous value RTT [i-1].
S51 ′ RTT_prev = {RTT [i−M] + RTT [i− (M−1)] +... + RTT [i−1]} / M
The state of the above holdover operation is shown in FIG.

(Sixth embodiment)
The configuration of the frequency / time synchronization system using the synchronization client in the sixth embodiment is the same as that of the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 synchronizes the round-trip delay time of the synchronization response signal transmitted immediately before with the RTT_prev and the synchronization server in addition to the processing steps described in the first to fourth embodiments. Assuming that the client time difference is dT_prev and the threshold is RTT_th,
| RTT-RTT_prev |> RTT_th
Is continued N times or more (N is the number of protection steps against sudden large jitter), it is determined that a change in the round-trip delay time caused by network path switching has occurred, and the time difference dT is set to dT = (T1−T2−T3 + T4) ) / 2 + dT_offset
And offset. here,
dT_offset = dT_offset + dT_prev−dT
It is.

The above algorithm is described in detail as follows.
S81 dT_offset = 0
S82 RTT_prev = RTT [i-1]
S83 dT_prev = dT [i-1]
S84 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S85 if (| RTT [i] −RTT_prev | ≦ RTT_th) then S86 else S89
S86 dT [i] = (T1 [i] -T2 [i] -T3 [i] + T4 [i]) / 2 + dT_offset
S87 i ++
S88 → S82
S89 K = 0
S90 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S91 if (| RTT [i] −RTT_prev | ≦ RTT_th) then S86 else S92
S92 K ++
S93 if (K <N) then S94 else S96
S94 i ++
S95 → S90
S96 dT_offset = dT_offset + dT_prev−dT [i]
S97 → S86
As RTT_th, for example, a value of standard deviation of jitter × 3 is used. In S82 and S83, an average value can be used instead of the immediately preceding instantaneous values RTT [i-1] and dT [i-1].
S82′RTT_prev = {RTT [i−M] + RTT [i− (M−1)] +... + RTT [i−1]} / M
S83 ′ dT_prev = {dT [i−M] + dT [i− (M−1)] +... + DT [i−1]} / M
The state of the above offset control operation is shown in FIG. The hatched circles in FIGS. 8 to 13 indicate the RTT before the offset correction.

(Seventh embodiment)
The configuration of the frequency / time synchronization system using the synchronization client in the seventh embodiment is the same as that in the first embodiment.

In addition to the processing steps described in the first to fourth embodiments, the client time control unit 4 immediately before the processing steps described in the first to fourth embodiments based on T1, T2, T3, and T4. The round-trip delay time of the transmitted synchronization response signal is RTT_prev, the time difference between the synchronization server and the synchronization client is dT_prev, and the threshold is RTT_th.
| RTT-RTT_prev |> RTT_th
Is continued N times or more (N is the number of protection steps against sudden large jitter), it is determined that a change in the round-trip delay time caused by network path switching has occurred, and the time difference dT is set to dT = (T1−T2−T3 + T4) ) / 2 + max {min (dT_offset, dT_offset_max), dT_offset_min}
And offset. here,
dT_offset = dT_offset + dT_prev−dT
It is.

The above algorithm is described in detail as follows.
S101 dT_offset = 0
S102 RTT_prev = RTT [i-1]
S103 dT_prev = dT [i−1]
S104 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S105 if (| RTT [i] −RTT_prev | ≦ RTT_th) then S106 else S109
S106 dT [i] = (T1 [i] −T2 [i] −T3 [i] + T4 [i]) / 2 + max {min (dT_offset, dT_offset_max), dT_offset_min}
S107 i ++
S108 → S102
S109 K = 0
S110 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S111 if (| RTT [i] −RTT_prev | ≦ RTT_th) then S106 else S112
S112 K ++
S113 if (K <N) then S114 else S116
S114 i ++
S115 → S110
S116 dT_offset = dT_offset + dT_prev−dT [i]
S117 → S106
As RTT_th, for example, a value of standard deviation of jitter × 3 is used.

  As dT_offset_max, by using a value such as a delay time difference associated with the assumed maximum path switching, dT_offset_min = dT_offset_max, it is possible to prevent an offset value from becoming excessive due to erroneous path switching detection.

In S102 and S103, an average value can be used instead of the immediately preceding instantaneous values RTT [i-1] and dT [i-1].
S102′RTT_prev = {RTT [i−M] + RTT [i− (M−1)] +... + RTT [i−1]} / M
S103 ′ dT_prev = {dT [i−M] + dT [i− (M−1)] +... + DT [i−1]} / M
The state of the above offset control operation is shown in FIG.

(Eighth Example)
The configuration of the frequency / time synchronization system using the synchronization client in the eighth embodiment is the same as that in the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 adds the round-trip delay time of the synchronization response signal transmitted immediately before to the RTT_prev, the synchronization server, in addition to the processing steps described in the first to fourth embodiments. The time difference from the synchronous client is dT_prev and the threshold is RTT_th.
| RTT-RTT_prev |> RTT_th
Is continued N times or more (N is the number of protection stages for sudden large jitter) or more, it is determined that a change in the round-trip delay time associated with the network path switching has occurred, and the threshold is set as dT_offset_th,
| DT_offset |> dT_offset_th
In this case, the time difference dT is expressed as dT = (T1-T2-T3 + T4) / 2 + dT_offset
And offset. here,
dT_offset = dT_offset + dT_prev−dT
It is.

The above algorithm is described in detail as follows.
S121 dT_offset = 0
S122 RTT_prev = RTT [i-1]
S123 dT_prev = dT [i-1]
S124 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S125 if (| RTT [i] −RTT_prev | ≦ RTT_th) then S126 else S132
S126 if (| dT_offset | <dT_offset_th) then S127 else S129
S127 dT [i] = (T1 [i] −T2 [i] −T3 [i] + T4 [i]) / 2
S128 → S130
S129 dT [i] = (T1 [i] −T2 [i] −T3 [i] + T4 [i]) / 2 + dT_offset
S130 i ++
S131 → S122
S132 K = 0
S133 RTT [i] = T4 [i] -T3 [i] + T2 [i] -T1 [i]
S134 if (| RTT [i] −RTT_prev | ≦ RTT_th) then S126 else S135
S135 K ++
S136 if (K <N) then S137 else S139
S137 i ++
S138 → S130
S139 dT_offset = dT_offset + dT_prev−dT [i]
S140 → S126
As RTT_th, for example, a value of standard deviation of jitter × 3 is used.

  As dT_offset_th, for example, a value about the standard deviation of jitter is used.

In S122 and S123, an average value can be used instead of the immediately preceding instantaneous values RTT [i-1] and dT [i-1].
S122′RTT_prev = {RTT [i−M] + RTT [i− (M−1)] +... + RTT [i−1]} / M
S123 ′ dT_prev = {dT [i−M] + dT [i− (M−1)] +... + DT [i−1]} / M
The state of the above offset control operation is shown in FIG.

(Ninth Example)
The configuration of the frequency / time synchronization system using the synchronization client in the ninth embodiment is the same as that in the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 sets the one-side delay times for the synchronous response signal transmitted immediately before to ΔTq_prev and ΔTr_prev, in addition to the processing steps described in the first to fourth embodiments. When the time difference between the synchronization server and the synchronization client is dT_prev and the threshold is ΔT_th,
| ΔTq−ΔTq_prev |> ΔT_th
Or | ΔTr−ΔTr_prev |> ΔT_th
Is continued N times (N is the number of protection stages for sudden large jitter) or more, it is determined that a change in one-side delay time due to network path switching has occurred, and the time difference dT is set to dT = (T1-T2- T3 + T4) / 2 + dT_offset
And offset. here,
dT_offset = dT_offset + dT_prev−dT
It is.

The above algorithm is described in detail as follows.
S151 dT_offset = 0
S152 ΔTq_prev = ΔTq [i−1]
ΔTr_prev = ΔTr [i−1]
S153 dT_prev = dT [i−1]
S154 ΔTq [i] = T2 [i] −T1 [i]
ΔTr [i] = T4 [i] −T3 [i]
S155 if {(| ΔTq [i] −ΔTq_prev | ≦ ΔT_th) and (| ΔTr [i] −ΔTr_prev | ≦ ΔT_th)} then S156 else S159
S156 dT [i] = (T1 [i] −T2 [i] −T3 [i] + T4 [i]) / 2 + dT_offset
S157 i ++
S158 → S152
S159 K = 0
S160 ΔTq [i] = T2 [i] −T1 [i]
ΔTr [i] = T4 [i] −T3 [i]
S161 if {(| ΔTq [i] −ΔTq_prev | ≦ ΔT_th) and (| ΔTr [i] −ΔTr_prev | ≦ ΔT_th)} then S156 else S162
S162 K ++
S163 if (K <N) then S164 else S166
S164 i ++
S165 → S160
S166 dT_offset = dT_offset + dT_prev−dT [i]
S167 → S156
As ΔT_th, for example, a value of standard deviation of jitter × 3 is used.

In S152 and S153, an average value can be used instead of the immediately preceding instantaneous values ΔTq [i-1], ΔTr [i-1], and dT [i-1].
S152 ′ ΔTq_prev = {ΔTq [i−M] + ΔTq [i− (M−1)] +... + ΔTq [i−1]} / M
ΔTr_prev = {ΔTr [i−M] + ΔTr [i− (M−1)] +... + ΔTr [i−1]} / M
S153 ′ dT_prev = {dT [i−M] + dT [i− (M−1)] +... + DT [i−1]} / M
The state of the above offset control operation is shown in FIG.

(Tenth embodiment)
The configuration of the frequency / time synchronization system using the synchronization client in the tenth embodiment is the same as that in the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 sets the one-side delay times for the synchronous response signal transmitted immediately before to ΔTq_prev and ΔTr_prev, in addition to the processing steps described in the first to fourth embodiments. When the time difference between the synchronization server and the synchronization client is dT_prev and the threshold is ΔT_th,
| ΔTq−ΔTq_prev |> ΔT_th
Or | ΔTr−ΔTr_prev |> ΔT_th
Is continued N times (where N is the number of protection stages for sudden large jitter) or more, it is determined that a change in the round-trip delay time associated with network path switching has occurred, and the time difference dT is set to dT = (T1−T2−). T3 + T4) / 2 + max {min (dT_offset, dT_offset_max), dT_offset_min}
And offset. here,
dT_offset = dT_offset + dT_prev−dT
It is.

The above algorithm is described in detail as follows.
S171 dT_offset = 0
S172 ΔTq_prev = ΔTq [i−1]
ΔTr_prev = ΔTr [i−1]
S173 dT_prev = dT [i−1]
S174 ΔTq [i] = T2 [i] −T1 [i]
ΔTr [i] = T4 [i] −T3 [i]
S175 if {(| ΔTq [i] −ΔTq_prev | ≦ ΔT_th) and (| ΔTr [i] −ΔTr_prev | ≦ ΔT_th)} then S176 else S179
S176 dT [i] = (T1 [i] −T2 [i] −T3 [i] + T4 [i]) / 2 + max {min (dT_offset, dT_offset_max), dT_offset_min}
S177 i ++
S178 → S172
S179 K = 0
S180 ΔTq [i] = T2 [i] −T1 [i]
ΔTr [i] = T4 [i] −T3 [i]
S181 if {(| ΔTq [i] −ΔTq_prev | ≦ ΔT_th) and (| ΔTr [i] −ΔTr_prev | ≦ ΔT_th)} then S176 else S182
S182 K ++
S183 if (K <N) then S184 else S186
S184 i ++
S185 → S180
S186 dT_offset = dT_offset + dT_prev−dT [i]
S187 → S176
As ΔT_th, for example, a value of standard deviation of jitter × 3 is used.

  As dT_offset_max, by using a value such as a delay time difference associated with the assumed maximum path switching, dT_offset_min = dT_offset_max, it is possible to prevent an offset value from becoming excessive due to erroneous path switching detection.

In S172 and S173, an average value can be used instead of the immediately preceding instantaneous values ΔTq [i−1], ΔTr [i−1], and dT [i−1].
S172 ′ ΔTq_prev = {ΔTq [i−M] + ΔTq [i− (M−1)] +... + ΔTq [i−1]} / M
ΔTr_prev = {ΔTr [i−M] + ΔTr [i− (M−1)] +... + ΔTr [i−1]} / M
S173 ′ dT_prev = {dT [i−M] + dT [i− (M−1)] +... + DT [i−1]} / M
The state of the above offset control operation is shown in FIG.

(Eleventh Example)
The configuration of the frequency / time synchronization system using the synchronization client in the eleventh embodiment is the same as that in the first embodiment.

Based on T1, T2, T3, and T4, the client time control unit 4 sets the one-side delay times for the synchronous response signal transmitted immediately before to ΔTq_prev and ΔTr_prev, in addition to the processing steps described in the first to fourth embodiments. When the time difference between the synchronization server and the synchronization client is dT_prev and the threshold is ΔT_th,
| ΔTq−ΔTq_prev |> ΔT_th
Or | ΔTr−ΔTr_prev |> ΔT_th
Is determined to have occurred in the round-trip delay time associated with the network path switching, and the threshold is dT_offset_th,
| DT_offset |> dT_offset_th
In this case, the time difference dT is expressed as dT = (T1-T2-T3 + T4) / 2 + dT_offset
And offset. here,
dT_offset = dT_offset + dT_prev−dT
It is.

The above algorithm is described in detail as follows.
S191 dT_offset = 0
S192 ΔTq_prev = ΔTq [i−1]
ΔTr_prev = ΔTr [i−1]
S193 dT_prev = dT [i−1]
S194 ΔTq [i] = T2 [i] −T1 [i]
ΔTr [i] = T4 [i] −T3 [i]
S195 if {(| ΔTq [i] −ΔTq_prev | ≦ ΔT_th) and (| ΔTr [i] −ΔTr_prev | ≦ ΔT_th)} then S196 else S202
S196 if (| dT_offset | <dT_offset_th) then S197 else S199
S197 dT [i] = (T1 [i] −T2 [i] −T3 [i] + T4 [i]) / 2
S198 → S200
S199 dT [i] = (T1 [i] −T2 [i] −T3 [i] + T4 [i]) / 2 + dT_offset
S200 i ++
S201 → S192
S202 K = 0
S203 ΔTq [i] = T2 [i] −T1 [i]
ΔTr [i] = T4 [i] −T3 [i]
S204 if {(| ΔTq [i] −ΔTq_prev | ≦ ΔT_th) and (| ΔTr [i] −ΔTr_prev | ≦ ΔT_th)} then S196 else S205
S205 K ++
S206 if (K <N) then S207 else S209
S207 i ++
S208 → S200
S209 dT_offset = dT_offset + dT_prev−dT [i]
S210 → S196
As ΔT_th, for example, a value of standard deviation of jitter × 3 is used.

  As dT_offset_th, for example, a value about the standard deviation of jitter is used.

In S192 and S193, an average value can be used instead of the immediately preceding instantaneous values ΔTq [i−1], ΔTr [i−1], and dT [i−1].
S192 ′ ΔTq_prev = {ΔTq [i−M] + ΔTq [i− (M−1)] +... + ΔTq [i−1]} / M
ΔTr_prev = {ΔTr [i−M] + ΔTr [i− (M−1)] +... + ΔTr [i−1]} / M
S193 ′ dT_prev = {dT [i−M] + dT [i− (M−1)] +... + DT [i−1]} / M
FIG. 13 shows the state of the above offset control operation.

(Twelfth embodiment)
FIG. 14 shows the configuration of the client time generation unit in the twelfth embodiment.

  The numerically controlled oscillator 30 cumulatively adds the output clocks of the voltage controlled oscillator 31. The output phase of the numerically controlled oscillator 30 can be made variable by numerically controlling the accumulated value by dT.

  The output of the numerically controlled oscillator 30 is converted into time data by the time generator 32 and output. The time data can be used as a transmission time T1 of the synchronization request packet and a reception time T4 of the synchronization response packet, or can be used for an application (time authentication service or the like) using time.

  The output of the numerically controlled oscillator 30 and the output of the voltage controlled oscillator 31 are respectively divided by a frequency divider 33 and a frequency divider 34, phase-compared by a phase comparator 35, and a high frequency component removed by a low-pass filter 36. It is fed back as a frequency control signal of the voltage controlled oscillator 31. The output of the voltage controlled oscillator 31 can be used as a highly stable clock.

  The numerically controlled oscillator 30 can respond at high speed to a fast fluctuation of dT. On the other hand, if an oscillator having excellent frequency stability such as a temperature-compensated crystal oscillator or a temperature-stabilized crystal oscillator with a thermostatic chamber is used as the voltage-controlled oscillator 31, and the cut-off frequency of the low-pass filter 36 is set to a low value, dT can cope with medium- to long-term fluctuations. Stable operation is possible.

  In this way, by combining a plurality of control means having different response characteristics, stable operation over a short period and a medium to long period can be realized.

(Thirteenth embodiment)
The configuration of the synchronization system in the thirteenth embodiment is shown in FIG. Here, the operation of the client time control unit 4 uses the one described in any of the first to eleventh embodiments. As the client time generation unit 1, the one described in the twelfth embodiment can be used.

(14th embodiment)
An embodiment of a program that, when installed in a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize functions corresponding to at least some of the functions of the synchronization client 10 of the present embodiment will be described.

  By recording the program of this embodiment on a recording medium, the general-purpose information processing apparatus can install the program of this embodiment using this recording medium. Alternatively, the program of this embodiment can be installed directly on the general-purpose information processing apparatus via a network from a server holding the program of this embodiment.

  Thereby, a function corresponding to at least a part of the function of the synchronous client 10 of the present embodiment can be realized using a general-purpose information processing apparatus.

  The program of this embodiment includes not only a program that can be directly executed by a general-purpose information processing apparatus but also a program that can be executed by installing it on a hard disk or the like. Also included are those that are compressed or encrypted.

  According to the present invention, even if there is a fluctuation in the delay time of the network, the synchronization of the frequency or time in the synchronization client can be realized, so that the operation accuracy of the network system that requires synchronization between the client devices is improved. Can be improved.

The figure explaining the principle of the time synchronization of NTP or SNTP. The figure which shows the structure of the frequency and time synchronous system in this invention. The figure explaining the operation | movement of RTT filtering in a 1st Example. The figure explaining the operation | movement of RTT filtering in a 2nd Example. The figure explaining the operation | movement of RTT filtering in a 3rd Example. The figure explaining the operation | movement of RTT filtering in 4th Example. The figure explaining operation | movement of the holdover in 5th Example. The figure explaining operation | movement of the offset control in a 6th Example. The figure explaining operation | movement of the offset control in a 7th Example. The operation of offset control in the eighth embodiment will be described. The figure explaining operation | movement of the offset control in a 9th Example. The figure explaining operation | movement of the offset control in a 10th Example. The figure explaining the operation | movement of the offset control in 11th Example. The figure which shows the structure of the client time generation part in 12th Example.

Explanation of symbols

T1 synchronization request packet transmission time T2 synchronization request packet reception time T3 synchronization response packet transmission time T4 synchronization response packet reception time D1 synchronization request packet delay time D2 synchronization response packet delay time 1 client time generation unit 2 synchronization request packet transmission unit 3 synchronization response Packet receiver 4 Client time controller 10 Sync client 11 Server time generator 12 Sync response packet transmitter 13 Sync request packet receiver 20 Sync server 30 Numerically controlled oscillator 31 Voltage controlled oscillator 32 Time generators 33 and 34 Frequency divider 35 Phase comparator 36 Low pass filter

Claims (15)

Means for transmitting a synchronization request signal to a synchronization server for generating a reference time, means for receiving a synchronization response signal in response to the synchronization request signal from the synchronization server, and the reference based on the received synchronization response signal In a client device comprising a frequency synchronized with time or means for generating time,
The time when the client device transmits the synchronization request signal to the synchronization server is T1, the time when the synchronization server receives the synchronization request signal is T2, and the synchronization server responds to the synchronization request signal. Applied when setting the frequency or time in the client device with reference to the reference time when the time when the response signal is transmitted is T3 and the time when the client device receives the synchronization response signal is T4 The time difference dT is dT = (T1-T2-T3 + T4) / 2
The round-trip delay time RTT between the client device and the synchronization server when the synchronization request signal is an outbound signal and the synchronization response signal is a return signal is RTT = T4-T3 + T2-T1
The time difference dT is limited to the time difference dT when the RTT value is a predetermined value.
The predetermined value is calculated when an average value RTT_mean of RTT before receiving the synchronization response signal is calculated and a preset jitter allowable value is δ,
RTT_lim1 = RTT_mean + δ
RTT_lim2 = RTT_mean-δ
RTT_lim2 <RTT <RTT_lim1
An RTT value satisfying the requirement: A client device characterized by: Means for transmitting a synchronization request signal to a synchronization server for generating a reference time, means for receiving a synchronization response signal in response to the synchronization request signal from the synchronization server, and the reference based on the received synchronization response signal In a client device comprising a frequency synchronized with time or means for generating time,
The time when the client device transmits the synchronization request signal to the synchronization server is T1, the time when the synchronization server receives the synchronization request signal is T2, and the synchronization server responds to the synchronization request signal. Applied when setting the frequency or time in the client device with reference to the reference time when the time when the response signal is transmitted is T3 and the time when the client device receives the synchronization response signal is T4 The time difference dT is dT = (T1-T2-T3 + T4) / 2
The round-trip delay time RTT between the client device and the synchronization server when the synchronization request signal is an outbound signal and the synchronization response signal is a return signal is RTT = T4-T3 + T2-T1
The time difference dT is limited to the time difference dT when the RTT value is a predetermined value.
The predetermined value is defined as RTT_prev, where RTT of the synchronous response signal at which the RTT has reached the predetermined value immediately before receiving the synchronous response signal, y is a preset allowable normalized frequency difference, and the synchronous response signal is The elapsed time from the application time of the time difference dT immediately before reception is T,
RTT_lim = RTT_prev + y × T
RTT <RTT_lim
An RTT value satisfying the requirement: A client device characterized by: Means for transmitting a synchronization request signal to a synchronization server for generating a reference time, means for receiving a synchronization response signal in response to the synchronization request signal from the synchronization server, and the reference based on the received synchronization response signal In a client device comprising a frequency synchronized with time or means for generating time,
The time when the client device transmits the synchronization request signal to the synchronization server is T1, the time when the synchronization server receives the synchronization request signal is T2, and the synchronization server responds to the synchronization request signal. Applied when setting the frequency or time in the client device with reference to the reference time when the time when the response signal is transmitted is T3 and the time when the client device receives the synchronization response signal is T4 The time difference dT is dT = (T1-T2-T3 + T4) / 2
The round-trip delay time RTT between the client device and the synchronization server when the synchronization request signal is an outbound signal and the synchronization response signal is a return signal is RTT = T4-T3 + T2-T1
The time difference dT is limited to the time difference dT when the RTT value is a predetermined value.
The predetermined value is defined as RTT_prev, RTT_prev of the synchronous response signal in which RTT has reached the predetermined value immediately before receiving the synchronous response signal, y1 and y2 being a preset allowable normalized frequency difference, and the synchronous response signal. The elapsed time from the application time of the time difference dT immediately before receiving the signal is T,
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev−y2 × T
RTT_lim2 <RTT <RTT_lim1
An RTT value satisfying the requirement: A client device characterized by: Means for transmitting a synchronization request signal to a synchronization server for generating a reference time, means for receiving a synchronization response signal in response to the synchronization request signal from the synchronization server, and the reference based on the received synchronization response signal In a client device comprising a frequency synchronized with time or means for generating time,
The time when the client device transmits the synchronization request signal to the synchronization server is T1, the time when the synchronization server receives the synchronization request signal is T2, and the synchronization server responds to the synchronization request signal. Applied when setting the frequency or time in the client device with reference to the reference time when the time when the response signal is transmitted is T3 and the time when the client device receives the synchronization response signal is T4 The time difference dT is dT = (T1-T2-T3 + T4) / 2
The round-trip delay time RTT between the client device and the synchronization server when the synchronization request signal is an outbound signal and the synchronization response signal is a return signal is RTT = T4-T3 + T2-T1
The time difference dT is limited to the time difference dT when the RTT value is a predetermined value.
The predetermined value is set in advance by setting RTT_prev as the RTT of the synchronous response signal in which the RTT becomes the predetermined value immediately before receiving the synchronous response signal, and y1 and y2 as the preset allowable normalized frequency difference. The constant is γ (0 <γ <1), and the elapsed time from the application time of the time difference dT immediately before receiving the synchronization response signal is T,
RTT_lim1 = RTT_prev + y1 × T
RTT_lim2 = RTT_prev × (1-γ) −y2 × T
RTT_lim2 <RTT <RTT_lim1
An RTT value satisfying the requirement: A client device characterized by: When the threshold is RTT_th,
| RTT-RTT_prev |> RTT_th
5. The client device according to claim 1, further comprising a unit that transitions to a holdover state in which the frequency of the local client device is held constant when the frequency continues for a certain period or more. The RTT before transition to the holdover state is RTT_prev ′, the preset allowable normalized frequency difference is y1 and y2, and the elapsed time from the transition to the holdover state is T ′.
RTT_lim1 = RTT_prev ′ + y1 × T ′
RTT_lim2 = RTT_prev′−y2 × T ′
RTT_lim2 <RTT <RTT_lim1
The client apparatus according to claim 5, wherein the holdover state is canceled when the RTT value satisfying the condition is satisfied. Immediately before receiving the synchronization response signal, the RTT of the synchronization response signal at which the RTT becomes the predetermined value is RTT_prev, the time difference dT at that time is dT_prev, and the threshold is RTT_th.
| RTT-RTT_prev |> RTT_th
Is continued for a certain period or longer, the current time difference dT is set to dT_offset = dT_offset + dT_prev−dT
dT = (T1-T2-T3 + T4) / 2 + dT_offset
The client device according to claim 1, wherein the client device is offset. Immediately before receiving the synchronization response signal, the RTT of the synchronization response signal at which the RTT becomes the predetermined value is RTT_prev, the time difference dT at that time is dT_prev, and the threshold is RTT_th.
| RTT-RTT_prev |> RTT_th
Is continued for a certain period or longer, the current time difference dT is set to dT_offset = dT_offset + dT_prev−dT
dT = (T1-T2-T3 + T4) / 2 + max {min (dT_offset, dT_offset_max), dT_offset_min}
The client device according to claim 1, wherein the client device is offset. Immediately before receiving the synchronization response signal, the RTT of the synchronization response signal whose RTT has reached the predetermined value is RTT_prev, the time difference dT at that time is dT_prev, and the first threshold is RTT_th,
| RTT-RTT_prev |> RTT_th
Is continued for a certain period or longer, the current time difference dT is set to dT_offset = dT_offset + dT_prev−dT
And the second threshold is dT_offset_th,
| DT_offset |> dT_offset_th
In this case, the time difference dT is expressed as dT = (T1-T2-T3 + T4) / 2 + dT_offset
The client device according to claim 1, wherein the client device is offset. ΔTq = T2-T1 as one-side delay time
ΔTr = T4-T3
The one-side delay time of the synchronization response signal in which RTT has reached the predetermined value immediately before receiving the synchronization response signal is ΔTq_prev and ΔTr_prev, the time difference dT at that time is dT_prev, and the threshold is ΔT_th,
| ΔTq−ΔTq_prev |> ΔT_th
Or | ΔTr−ΔTr_prev |> ΔT_th
Is continued for a certain period or longer, the current time difference dT is set to dT_offset = dT_offset + dT_prev−dT
dT = (T1-T2-T3 + T4) / 2 + dT_offset
The client device according to claim 1, wherein the client device is offset. ΔTq = T2-T1 as one-side delay time
ΔTr = T4-T3
The one-side delay time of the synchronization response signal in which RTT has reached the predetermined value immediately before receiving the synchronization response signal is ΔTq_prev and ΔTr_prev, the time difference dT at that time is dT_prev, and the threshold is ΔT_th,
| ΔTq−ΔTq_prev |> ΔT_th
Or | ΔTr−ΔTr_prev |> ΔT_th
, The current time difference dT is changed to dT_offset = max {min (dT_prev−dT, dT_offset_max), dT_offset_min}
dT = (T1-T2-T3 + T4) / 2 + dT_offset
The client device according to claim 1, wherein the client device is offset. ΔTq = T2-T1 as one-side delay time
ΔTr = T4-T3
ΔTq_prev and ΔTr_prev are the one-side delay times of the synchronization response signal in which the RTT has reached the predetermined value immediately before receiving the synchronization response signal, the time difference dT at that time is dT_prev, and the first threshold is As ΔT_th,
| ΔTq−ΔTq_prev |> ΔT_th
Or | ΔTr−ΔTr_prev |> ΔT_th
Is continued for a certain period of time,
dT_offset = dT_offset + dT_prev−dT
And the second threshold is dT_offset_th,
| DT_offset |> dT_offset_th
In this case, the current time difference dT is set to dT = (T1−T2−T3 + T4) / 2 + dT_offset
The client device according to claim 1, wherein the client device is offset. A numerically controlled oscillator capable of controlling the phase by the time difference dT and a variable frequency oscillator;
The phase difference between the output of the numerically controlled oscillator and the output of the frequency variable oscillator is detected, and means for controlling the frequency of the frequency variable oscillator so that the phase difference becomes a constant value is provided. The client device according to claim 1.   A synchronization system comprising: a synchronization server that generates a reference time; and the client device according to claim 1.   14. A program that, when installed in a general-purpose information processing device, causes the general-purpose information processing device to realize functions corresponding to at least some of the functions of the client device according to any one of claims 1 to 13.

Images