CN104660361A - Filtering method for time synchronization of distributed network - Google Patents
Filtering method for time synchronization of distributed network Download PDFInfo
- Publication number
- CN104660361A CN104660361A CN201510072034.6A CN201510072034A CN104660361A CN 104660361 A CN104660361 A CN 104660361A CN 201510072034 A CN201510072034 A CN 201510072034A CN 104660361 A CN104660361 A CN 104660361A
- Authority
- CN
- China
- Prior art keywords
- time
- delay
- array
- master
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Synchronisation In Digital Transmission Systems (AREA)
Abstract
The invention relates to a filtering method for time synchronization of a distributed network. The method comprises the following steps: setting up a synchronous clock and determining a master clock node and a slave clock node in a network; carrying out synchronous communication for the master and the slave clock nodes to acquire one-way time delays in the master and slave communication; taking a certain number of the one-way time delays as a group to calculate the minimum time delay in the group, and recording the corresponding time value of the minimum time delay; taking the minimum time delays of many groups as a first array, and taking the corresponding time values of respectively minimum time delays as a second array; carrying out Kalman filter calculation for the first and the second arrays to acquire the difference of frequency of the master and the slave communication. According to the filtering method for the time synchronization of the distributed network, the jitter caused in the transmission of a synchronous message in the network is effectively filtered, the message with the minimum transmission delay is subjected to Kalman filter to finally output a stable and reliable frequency deviation, so that the accuracy and stability of the post frequency synchronization are effectively improved.
Description
Technical field
The present invention relates to network communication field, espespecially a kind of filtering method of distributed network time service.
Background technology
Along with the development of business IPization is accelerated, constantly there is essence change in the type of service of communication network carrying.The IPization of mobile operator core net is implemented in full, in mobile network, RAN is mainly based on SDH (Synchronous Digital Hierarchy) (SDH) technique construction, and 3G (Third Generation) Moblie (3G) network and forth generation communication (4G) network all need the transmission based on Packet Service.The carrying change of bandwidth demand and the evolution of technology, transport network also there occurs change.Therefore, packetizing has become one of of paramount importance feature of transmission network.Network communication has two ranks for the demand of clock synchronous---Frequency Synchronization and time synchronized.Time synchronized can transmit correct time information by gps satellite clock, although sync message can carry out Frequency Synchronization, but in a transport network, due to not fixing of network traffics, message is through intermediate transmission media (netting twine, optical fiber, switch, router etc.) impact that transmission delay and residence time postpone, the precision of Frequency Synchronization can be caused to occur very large deviation.
At present, although industry has employing E1, IRIG-B carries out frequency time service, and the Signal transmissions of picture E1, IRIG-B needs to lay client cables separately, and need material resources and the human cost of at substantial, cost is high, is unfavorable for popularization and application.
Summary of the invention
The object of the invention is to the defect overcoming prior art, a kind of filtering method of distributed network time service is provided, solve the frequency departure problem that produces in existing Internet Transmission and E1, IRIG-B to carry out frequency time service and need to lay separately the problems such as cable labor intensive material resources.
The technical scheme realizing above-mentioned purpose is:
The filtering method of a kind of distributed network time service of the present invention, comprising:
Assembly time service network environment, determines the master-salve clock node in network;
Carry out the synchronous communication of master-salve clock node, obtain the One Way Delay in master-slave communication;
Be one group with the described One Way Delay of some, calculate the minimal time delay in this group, record the corresponding time value of described minimal time delay simultaneously;
Form the first array with the minimal time delay organized more again, and form the second array with the corresponding time value of the minimal time delay in described first array;
Described first array and described second array are carried out Kalman filtering calculating, to obtain the frequency-splitting in master-slave communication.
A kind of filtering method of distributed network time service is provided, the shake that effective filtering sync message brings in Internet Transmission, message minimum for transmission delay is carried out Kalman filtering, the reliable frequency departure of final stable output, effectively improve precision and the stability of later stage Frequency Synchronization.The present invention utilizes existing internet platform, in conjunction with low-pass filtering and Kalman filtering algorithm, can remove network jitter and transmission delay under the network environment of complexity, improves sync message to the precision synchronous from nodal frequency at different levels.Because the method is a kind of time calibration in network agreement, without the need to laying separately dedicated network passage, solve E1, IRIG-B carries out the problem that frequency time service needs to lay separately cable labor intensive material resources.Directly utilize conventional network resources, greatly reduce input cost, have good practicality.
The further improvement of the filtering method of distributed network time service of the present invention is, described One Way Delay is the difference between the transmitting time of master clock node is stabbed and stabbed from the time of reception of clock node.
The further improvement of the filtering method of distributed network time service of the present invention is, described first array and described second array is carried out Kalman filtering calculating and comprises:
Utilize state equation and the observational equation of Kalman Filter Estimation algorithm, through linear Kalman filter algorithm, best One Way Delay is shown to described first array and described second array, and calculates principal and subordinate's frequency-splitting;
Described state equation is:
X
(k)···X
(k·1)·W
(k)·[X
(k)·H···X
(k·1)] (1)
Described observational equation is:
In formula (1) and formula (2): K is the marshalling sequence number of One Way Delay, X
(k)be the minimal time delay of K group, X
(k1) be the minimal time delay of K+1 group, be triangle state matrix, W
(k)for weight matrix, H is observing matrix, X
k () observesfor the measured value of principal and subordinate's unidirectional path time delay, I is unit matrix, and Q is plant noise.
The further improvement of the filtering method of distributed network time service of the present invention is, described linear Kalman filter algorithm comprises:
Tmp···P
(k，k)··
T·Q (5)
Formula (3) in formula (7), X
k1be the minimal time delay of K+1 group,
in matrix
for the optimal estimation value of principal and subordinate's One Way Delay after Kalman filtering, Freq
k1being the frequency-splitting after the K+1 time filtering between master-slave communication, is triangular matrix, W
(k)be the weight matrix of the K time, H is observing matrix, and Tmp is provisional matrix, and R is 1, Q is plant noise, and I is unit matrix, P
(k1, k1)it is the evaluated error of the K+1 time.
Accompanying drawing explanation
Fig. 1 is the filtering system structure chart of the filtering method adopting distributed network time service of the present invention;
Fig. 2 is the schematic diagram of One Way Delay in the filtering method of distributed network time service of the present invention; And
Fig. 3 is that sync message causes the schematic diagram of error by network delay shake.
Embodiment
Below in conjunction with the drawings and specific embodiments, the invention will be further described.
The invention provides a kind of filtering method of distributed network time service, have employed accurate clock synchronous protocol, network message is adopted to carry out clock synchronous, being one allows the multiple nodes in communication network by the propagation of clock message, thus reaches and reach synchronous agreement with the primary timing reference be connected in same network.The present invention is for solving the technical problem of existing clock clock synchronous difference under the network environment of complexity.Adopt in master-salve clock synchronisation of nodes communication process, postpone to measure from clock node to the one-way communication between master-salve clock, the master-slave communication time delay of acquisition is carried out filtering and finds out principal and subordinate's One Way Delay minimum in window, again Kalman filtering algorithm is carried out in the minimum principal and subordinate's one-way latency after process and moment of current record and carry out frequency-splitting calculating, finally obtain master-salve clock frequency departure accurately, ensure that the precision of carrying out master-salve clock Frequency Synchronization under the topological network of various complexity.Be described below in conjunction with the filtering method of accompanying drawing to distributed network time service of the present invention.
Consult Fig. 1, show the filtering system structure chart of the filtering method adopting distributed network time service of the present invention.Below in conjunction with Fig. 1, the filtering method of distributed network time service of the present invention is described.
As shown in Figure 1, the filtering method of distributed network time service of the present invention comprises:
Assembly time service network environment, puts up synchronised clock, determines the master-salve clock node in network; Build server 11 and be connected to form network communications environment, determine master clock node in network service and from clock node;
Carry out the synchronous communication of master-salve clock node, obtain the One Way Delay in master-slave communication, as shown in Figure 2, T
mfor master clock time axle, T
sfor from clock time axle, sync message P1 sends from master clock to from clock, and t11 is the transmitting time stamp of sync message P1 at master clock place, and t21 is that sync message P2 stabs at the time of reception from clock, the One Way Delay T in master-slave communication
mS, wherein T
mS1for the time difference of t21 and t11, computing formula is: T
mSn=t2n-t1n, n are natural number.In the synchronous communication of master-salve clock node, obtain the One Way Delay T in master-slave communication repeatedly
mS, such as, obtain the One Way Delay T of sync message P1, sync message P2 and sync message Pn
mS1, T
mS2, and T
mSn, n is natural number;
With the One Way Delay T of some
mSbe one group, calculate the minimal time delay [T in this group
mS] min, record the corresponding time value t of this minimal time delay simultaneously
n; By the One Way Delay T obtained
mSbe input in mini-value filtering 13, export minimal time delay [T
mS] min and corresponding time value t
n, the One Way Delay T that this mini-value filtering 13 pairs is N time
mScarry out filtering, find out N One Way Delay T
mSin minimum One Way Delay as minimal time delay [T
mS] min, wherein N is natural number, and computing formula is: [T
mS] min=MIN ([T
mS1, T
mS2, T
mS3t
mSn]), n is natural number;
Form the first array with the minimal time delay organized more again, and form the second array with the corresponding time value of each minimal time delay in the first array; Through the minimal time delay [T that mini-value filtering 13 obtains
mS] min, obtain multiple minimal time delay [T
mS] min forms the first array, result [T filtering obtained
mS] min0, [T
mS] min1, [T
mS] min2 ... [T
mS] minN as the first array, can [X0, X1, X2 be expressed as ... Xi, Xi+1], simultaneously by each filtering minimal time delay [T out
mS] the corresponding time value t of min
nmform the second array, this second array can be expressed as [t
0, t
1, t
2... t
k, t
k+1];
First array and the second array are carried out Kalman filtering calculating, to obtain the frequency-splitting in master-slave communication; By [X0, X1, X2 ... Xi, Xi+1] and [t
0, t
1, t
2... t
k, t
k+1] be input in Kalman filter 15 and calculate, to obtain the frequency-splitting 111 in master-slave communication.
Carry out Kalman filtering calculating to comprise, the minimum principal and subordinate's unidirectional path time delay exported by low-pass filtering window can draw optimum principal and subordinate's unidirectional path time delay through Discrete Linear Kalman filtering algorithm, and finally can calculate accurate principal and subordinate's frequency-splitting, implement state equation and observational equation required for Kalman Filter Estimation algorithm, utilize this state equation and observational equation, and then through linear Kalman filter algorithm, best One Way Delay is shown to the first array and the second array, and calculate principal and subordinate's frequency-splitting;
State equation is:
X
(k)···X
(k·1)·W
(k)·[X
(k)·H···X
(k·1)] (1)
Observational equation is:
In formula (1) and formula (2): K is the marshalling sequence number of One Way Delay, X
(k)be the minimal time delay of K group, X
(k1)being the minimal time delay of K+1 group, is triangle state matrix, W
(k)for weight matrix, H is observing matrix, X
k () observesfor the measured value of principal and subordinate's unidirectional path time delay, I is unit matrix, and Q is plant noise.
Linear Kalman filter algorithm comprises:
Tmp···P
(k，k)··
T·Q (5)
Formula (3) in formula (7), X
k1be the minimal time delay of K+1 group,
in matrix
for the optimal estimation value of principal and subordinate's One Way Delay after Kalman filtering, Freq
k1being the frequency-splitting after the K+1 time filtering between master-slave communication, is triangular matrix, W
(k)be the weight matrix of the K time, H is observing matrix, and Tmp is provisional matrix, and R is 1, Q is plant noise, and I is unit matrix, P
(k1, k1)it is the evaluated error of the K+1 time.Freq
k1for the last output of linear Kalman filter algorithm, this Kalman filtering algorithm is iterative algorithm, as long as have new X at every turn
k1, i.e. the K+1 time minimal time delay, will calculate principal and subordinate's frequency-splitting more accurately in real time.
Decompose in detail for Kalman filtering algorithm below and illustrate.
First initialization is carried out to the parameter in Karman equation formula, namely to formula (3) to formula (7), then carry out iteration, circular is as follows:
it is the input matrix of i-th time;
W
iit is the weight of i-th time;
P
(i, i)be the evaluated error of i-th time, this value is updated in an iterative process;
be the output matrix of i-th time,
be the optimal estimation of the t of the i-th+1 time, y
ilit is the frequency difference of i-th time;
Tmp···P
(i，i)··
T·Q (8)
W
(i·1)·Tmp·H
T·[H·Tmp·H
T·R]
·1(9)
P
(i·1，i·1)·(I·W
(i·1)·H)·Tmp (11)
Be input as P
(i, i)with
wherein during i=0,
in y
0=0;
Formula (8) is utilized to calculate temporary variable matrix T mp;
Temporary variable matrix is substituted in formula (9), obtains the weight matrix W of the i-th+1 time
(i1);
By Tmp and W
(i1)in substitution formula (10), obtain the Output rusults matrix of Kalman filtering
wherein X
(i1)the t in array is inputted for us
i1;
By weight matrix W
(i1) and temporary variable Tmp substitute in formula (11) and calculate P
(i1, i1), utilize this value to upgrade P
(i, i);
Will
and P
(i1, i1)as the input data of filtering next time; Above-mentioned y
i1be principal and subordinate's frequency-splitting that Kalman filtering exports.
As shown in Figure 3, describe traditional sync message synchronous inaccurate problem of the frequency of occurrences in the entire network, because message is subject to network traffics in Internet Transmission, the impact of the uncertain factors such as intermediate equipment such as route switching, the frequency departure that may lead to errors is estimated.T in figure
mfor master clock time axle, T
sfor from clock time axle, place's master-salve clock synchronization communication in 0 second, at 1 second place from the message generation delay variation of clock, create error, its clock 1ppm slower in opposite end should be estimated from clock from message arrival interval.
Filtering method of the present invention adds new filtering algorithm and strategy on the distributed networking protocol of standard, improves accordingly for original clock synchronous principle and supplements, and promotes the development of synchronization of distributed networks technology at internet arena greatly.First master-salve clock node is determined; Carry out the communication of main and subordinate node time synchronized, obtaining the stamp of the transmitting time in a series of principal and subordinate from clock node and stabbing from the time of reception of clock; Calculate the unidirectional delay of principal and subordinate often organizing corresponding timestamp; Carrying out window minimum filtering to often organizing principal and subordinate's One Way Delay, extracting minimum principal and subordinate's One Way Delay and corresponding moment; And then carry out the process of Discrete Linear Kalman filtering and calculating; Finally obtain reliable principal and subordinate's frequency departure.This method effectively can be removed sync message and shake large bag in transmitting procedure in real network, can be good at the problem that Suppression network shake causes frequency difference calculation deviation.
The beneficial effect of the filtering method of distributed network time service of the present invention is:
The invention provides a kind of filtering method of distributed network time service, the shake that effective filtering sync message brings in Internet Transmission, message minimum for transmission delay is carried out Kalman filtering, the reliable frequency departure of final stable output, effectively improve precision and the stability of later stage Frequency Synchronization.The present invention utilizes existing internet platform, in conjunction with low-pass filtering and Kalman filtering algorithm, can remove network jitter and transmission delay under the network environment of complexity, improves sync message to the precision synchronous from nodal frequency at different levels.Because the method is a kind of time calibration in network agreement, without the need to laying separately dedicated network passage, solve E1, IRIG-B carries out the problem that frequency time service needs to lay separately cable labor intensive material resources.Directly utilize conventional network resources, greatly reduce input cost, have good practicality.
Below embodiment is to invention has been detailed description by reference to the accompanying drawings, and those skilled in the art can make many variations example to the present invention according to the above description.Thus, some details in embodiment should not form limitation of the invention, the present invention by the scope that defines using appended claims as protection scope of the present invention.
Claims (4)
1. a filtering method for distributed network time service, is characterized in that, comprising:
Set up time service network environment, determine the master-salve clock node in network;
Carry out the synchronous communication of master-salve clock node, obtain the One Way Delay in master-slave communication;
Be one group with the described One Way Delay of some, calculate the minimal time delay in this group, record the corresponding time value of described minimal time delay simultaneously;
Form the first array with the minimal time delay organized more again, and form the second array with the corresponding time value of the minimal time delay in described first array;
Described first array and described second array are carried out Kalman filtering calculating, to obtain the frequency-splitting in master-slave communication.
2. the filtering method of distributed network time service as claimed in claim 1, is characterized in that, described One Way Delay is the difference between the transmitting time of master clock node is stabbed and stabbed from the time of reception of clock node.
3. the filtering method of distributed network time service as claimed in claim 1, is characterized in that, described first array and described second array is carried out Kalman filtering calculating and comprises:
Utilize state equation and the observational equation of Kalman Filter Estimation algorithm, through linear Kalman filter algorithm, best One Way Delay is shown to described first array and described second array, and calculates principal and subordinate's frequency-splitting;
Described state equation is:
X
(k)···X
(k·1)·W
(k)·[X
(k)·H···X
(k·1)] (1)
Described observational equation is:
In formula (1) and formula (2): K is the marshalling sequence number of One Way Delay, X
(k)be the minimal time delay of K group, X
(k1)being the minimal time delay of K+1 group, is triangle state matrix, W
(k)for weight matrix, H is observing matrix, X
k () observesfor the measured value of principal and subordinate's unidirectional path time delay, I is unit matrix, and Q is plant noise.
4. the filtering method of distributed network time service as claimed in claim 3, it is characterized in that, described linear Kalman filter algorithm comprises:
Tmp···P
(k，k)··T·Q (5)
Formula (3) in formula (7), X
k1be the minimal time delay of K+1 group,
in matrix
for the optimal estimation value of principal and subordinate's One Way Delay after Kalman filtering, Freq
k1being the frequency-splitting after the K+1 time filtering between master-slave communication, is triangular matrix, W
(k)be the weight matrix of the K time, H is observing matrix, and Tmp is provisional matrix, and R is 1, Q is plant noise, and I is unit matrix, P
(k1, k1)it is the evaluated error of the K+1 time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510072034.6A CN104660361A (en) | 2015-02-11 | 2015-02-11 | Filtering method for time synchronization of distributed network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510072034.6A CN104660361A (en) | 2015-02-11 | 2015-02-11 | Filtering method for time synchronization of distributed network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104660361A true CN104660361A (en) | 2015-05-27 |
Family
ID=53251104
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510072034.6A Pending CN104660361A (en) | 2015-02-11 | 2015-02-11 | Filtering method for time synchronization of distributed network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104660361A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112866758A (en) * | 2016-09-30 | 2021-05-28 | 网络洞察力有限公司 | Time synchronization in real-time content distribution systems |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101174912A (en) * | 2007-12-05 | 2008-05-07 | 武汉烽火网络有限责任公司 | Self-adapting clock method based on time stamp facing Ethernet circuit simulation service |
| EP2299613A1 (en) * | 2009-09-22 | 2011-03-23 | Siemens Aktiengesellschaft | A method for clock synchronization in a communication network |
| CN103166730A (en) * | 2013-03-13 | 2013-06-19 | 西北工业大学 | Method for synchronizing time in wireless ad hoc network based on protocol of institute of electrical and electronic engineers (IEEE) 1588 |
| CN103888237A (en) * | 2014-04-04 | 2014-06-25 | 瑞斯康达科技发展股份有限公司 | Method and device for achieving clock time synchronization |
-
2015
- 2015-02-11 CN CN201510072034.6A patent/CN104660361A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101174912A (en) * | 2007-12-05 | 2008-05-07 | 武汉烽火网络有限责任公司 | Self-adapting clock method based on time stamp facing Ethernet circuit simulation service |
| EP2299613A1 (en) * | 2009-09-22 | 2011-03-23 | Siemens Aktiengesellschaft | A method for clock synchronization in a communication network |
| EP2385641A1 (en) * | 2009-09-22 | 2011-11-09 | Siemens Aktiengesellschaft | A method for clock synchronization in a communication network |
| CN103166730A (en) * | 2013-03-13 | 2013-06-19 | 西北工业大学 | Method for synchronizing time in wireless ad hoc network based on protocol of institute of electrical and electronic engineers (IEEE) 1588 |
| CN103888237A (en) * | 2014-04-04 | 2014-06-25 | 瑞斯康达科技发展股份有限公司 | Method and device for achieving clock time synchronization |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112866758A (en) * | 2016-09-30 | 2021-05-28 | 网络洞察力有限公司 | Time synchronization in real-time content distribution systems |
| US11659215B2 (en) | 2016-09-30 | 2023-05-23 | Net Insight Ab | Time synchronization in a live content distribution system |
| CN112866758B (en) * | 2016-09-30 | 2024-01-05 | 网络洞察力有限公司 | Method for use in a network system capable of node-to-node time transfer |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2676389B1 (en) | Method of providing a path delay asymmetry for time synchronization between a master and a slave clock across a communciation network | |
| US9014282B2 (en) | Precision timing in a data over cable service interface specification (DOCSIS) system | |
| CN102265549B (en) | Network timing synchronization systems | |
| CN102332973B (en) | Real-time communication and clock synchronization method of chain network | |
| CN103929293B (en) | Asymmetrically-delayed time synchronization method and system | |
| CN101675614A (en) | Method for synchronizing a clock of a network component with a clock of further network component and network component therefor | |
| JP2013017178A (en) | Methods, system, communication device, and server device for performing synchronization and compensating clock drift among plurality of communication devices interconnected to form mesh communications network | |
| CN105959076A (en) | Device and method for making passive optical network possess ability of supporting time synchronization | |
| CN109150357A (en) | The method for synchronizing time of hybrid bus based on RS485 and Ethernet | |
| CN102833025B (en) | E1/Ethernet protocol conversion precise delay computational methods, module and transducer | |
| CN102843620A (en) | OTN (Optical Transport Network) device and method for realizing time synchronous transmission | |
| US11683150B2 (en) | Methods, apparatus and computer-readable media for synchronization over an optical network | |
| CN110784783B (en) | Clock synchronization method and device based on optical fiber network | |
| CN103168440A (en) | Method and device for compensating for time path | |
| CN103259640A (en) | Method and device for synchronizing time | |
| Lévesque et al. | Improving the PTP synchronization accuracy under asymmetric delay conditions | |
| CN103441833A (en) | Method and system for synchronizing frequency of master unit and frequency of slave unit | |
| CN104660361A (en) | Filtering method for time synchronization of distributed network | |
| Han et al. | First national high-precision time synchronization network with sub-microsecond accuracy over commercial optical networks for wireless applications | |
| CN104601269A (en) | Master and slave clock synchronization method and system | |
| CN105007134B (en) | A kind of method, apparatus for suppressing packet network PDV noises and from clockwork | |
| WO2016082369A1 (en) | Method, apparatus and system for synchronizing clock source attributes | |
| Rösel et al. | Current developments of ieee 1588 (precision time protocol) | |
| CN103634239A (en) | Packet clock network node frequency offset estimation method and device | |
| CN111510346A (en) | End-to-end network forwarding delay determination method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150527 |
|
| WD01 | Invention patent application deemed withdrawn after publication |