CN106533602A - ATS-protocol-based time synchronization method of hierarchical network - Google Patents
ATS-protocol-based time synchronization method of hierarchical network Download PDFInfo
- Publication number
- CN106533602A CN106533602A CN201610984696.5A CN201610984696A CN106533602A CN 106533602 A CN106533602 A CN 106533602A CN 201610984696 A CN201610984696 A CN 201610984696A CN 106533602 A CN106533602 A CN 106533602A
- Authority
- CN
- China
- Prior art keywords
- node
- network
- synchronization
- clock
- time
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04J—MULTIPLEX COMMUNICATION
- H04J3/00—Time-division multiplex systems
- H04J3/02—Details
- H04J3/06—Synchronising arrangements
- H04J3/0635—Clock or time synchronisation in a network
- H04J3/0638—Clock or time synchronisation among nodes; Internode synchronisation
- H04J3/0658—Clock or time synchronisation among packet nodes
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Synchronisation In Digital Transmission Systems (AREA)
Abstract
The invention relates to an average-time-synchronization(ATS)-protocol-based time synchronization method of a hierarchical network, wherein the method is used for realizing high-precision time synchronization of a distributed network. The method mainly comprises: a master clock and P synchronization node clocks are provided; on the basis of improvement of a Booststrap statistic method, collected node information is filtered; Z-layer layering is carried out on a network and time pair information of nodes in layers are collected; data fusion is carried out on the filtered node information in the layers by using a natural selection particle swarm algorithm to generate a virtual main block and the like. A method for layered synchronization in a distributed network is provided, so that the original ATS protocol advantages are kept and the network convergence speed increases and thus network synchronization time is shortened. Moreover, the virtual main clock obtained by using the natural selection particle swarm algorithm is introduced into the layers to realize consistent synchronization with the nodes in layers. Therefore, the network synchronization precision is improved.
Description
Technical field
The present invention relates to a kind of distributed network realizes the synchronous hierarchical network clock based on ATS agreements of high precision clock
Synchronous method.
Background technology
It is multi-hop that distributed wireless networks are made up of a series of mobile node of the complete equality in status, provisional from group
System is knitted, and does not rely on any static infrastructure, each node is realized between node by Radio Link direction communication
Information it is shared.The technologies such as intelligent office of family, Internet of Things are all based on distributed network, and the clock synchronization in distributed network
Technology is important research content.Clock is synchronously that a key technology of wireless sensor network is supported, and it is in data fusion, mesh
Mark is followed the trail of, and is played a significant role in the application such as event monitoring.
Clock synchronization mode mainly has 3 kinds：First kind of way is realized synchronous using global position system GPS；Second side
Formula is code synchronism；The third mode is message synchronization.Network topology structure of wireless sensor has centralized and distributed,
Some clock synchronization protocol have two kinds of the clock synchronization protocol based on reference mode and distributed clock synchronous protocol,
Reference Broadcast Synchronization (RBS) clock synchronization protocol, Flooding Time
Synchronization Protocol (FTSP) clock synchronization protocol and the Time- by propositions such as Ganeriwal
Synchronization Protocol for Sensor Networks (TPSN) clock synchronization protocol is based on reference mode
Clock synchronization protocol representative.Have algorithm simple based on the clock synchronization protocol of reference mode, amount of calculation is little, convergence rate
The advantages of fast, but these algorithms need to refer to node and have degree of precision, its to child node transmission packet information it is accurate
Degree also has higher requirements.Its change to network topology structure simultaneously is very sensitive, and the increase of node or loss can all make entirely
Network restarts work, and may cause worse synchronization accuracy.It is in order to improve the stability and extensibility of network, numerous
Scholar proposes distributed clock synchronous protocol.The Distributed Time Synchronization of the propositions such as Solis
Protocol (DTSP) clock synchronization protocol, Average Time Synchronization (ATS) clock synchronization protocol are
The representative of distributed clock synchronous protocol.Distributed clock synchronous protocol does not rely on any specific node, and network does not limit to
In specific topological structure.Thus distributed clock synchronously has more preferable robustness and extensibility.Meanwhile, consistency synchronization
Agreement can be while slope and the skew of compensating clock, the synchronization accuracy in raising network between neighbor node.It is existing consistent
Property synchronous protocol be all based on greatly that average homogeneity is theoretical, these algorithms are all asymptotic convergences, it is impossible to ensure finite time convergence control,
The synchronous energy consumption of real-time performance clock is larger, so as to limit the use of algorithm.
The content of the invention
Solve current distributed network and realize that synchronous precision is low, convergence rate object of the present invention is to provide a kind of
The hierarchical network clock synchronizing method based on ATS agreements of the problems such as slowly, energy consumption is big.
The object of the present invention is achieved like this：
(1) master clock and P synchronization node clock are provided；
(2) process is filtered to the node time information collected by improving Bootstrap statistical methods；
(3) a point Z layers, and the time of collecting layer interior nodes are carried out to network to information；Pass through natural selection particle in layer
Group's algorithm carries out data fusion to filtered nodal information, produces virtual master clock；
(4) the virtual master clock for obtaining is added in hierarchical network, then per layer contains n+1 node, n=P/Z；
(5) per the node in layer network according to ATS agreements and the time collected to information realization group internal clock synchronization, together
When obtain the independent logical timers of Z；
(6) this Z independent logical timer is considered as into Z independent logical node, while which is with the local logic of itself
On the basis of clock, the temporal information of itself is periodically broadcasted to its logical node；
(7) network again according to ATS agreements and the time of logical node collected to information realization whole network when
Clock synchronization.
The beneficial effects of the present invention is：The present invention under condition of small sample to the node crystal oscillator of wireless network when partially and
Frequency deviation is filtered process, ensure frequency deviation and when inclined amendment precision under conditions of shorten the sampling time.Existing layering
Network and its improved method are all based on centralized synchronous protocol, and the clock synchronization protocol based on reference mode is tied to network topology
The change of structure is very sensitive, and the increase of node or loss can all make whole network restart work, and may cause worse
Synchronization accuracy.And the present invention proposes to realize that in distributed network the method for hierarchical synchronization avoids the problems referred to above, keeping former
The convergence rate of network is improve while ATS agreement advantages, the time for making real-time performance synchronous shortens.And will be by natural selection
The virtual master clock that particle cluster algorithm is formed is included in layer, is allowed to realize consistency synchronization with layer interior nodes, is the method increased
The synchronization accuracy of network.
Description of the drawings
Fig. 1 is interaction between master-salve clock；
Fig. 2 is linear clock models；
Fig. 3 is the flow chart of the inventive method；
Fig. 4 is the topology diagram of hierarchical network.
Specific embodiment
Below in conjunction with the accompanying drawings the present invention is described further.
Present invention firstly provides a master clock and P synchronization node clock；By improving Bootstrap statistical methods pair
The node time information collected is filtered process, improves the reliability of node time information；Network is layered (it is assumed that
It is divided into Z groups), and the time of collecting layer interior nodes is to information.By natural selection particle cluster algorithm to filtered section in layer
Point information carries out data fusion, produces virtual master clock；The virtual master clock for obtaining is added in hierarchical network, then per layer contains
There is n+1 node (n=P/Z)；Per the node in layer network according to ATS agreements and the time collected in information realization group when
Clock synchronization, while obtaining Z independent logical timer；This Z independent logical timer is considered as into Z independent logical node,
Simultaneously on the basis of its local logical timer by oneself, the temporal information of oneself is periodically broadcasted to its logical node；Network
Again according to the clock synchronization of ATS agreements and the time of logical node collected to information realization whole network.
The present invention includes：
Step 1：One master clock and P synchronization node clock are provided；
Step 2：Bootstrap statistic algorithms are improved.The bootstrap of Bootstrap algorithms is straight by original sample
Deliver a child into, when raw sample data hour, bootstrap can be very similar to former sample, it will cause result of calculation much to deviate
Truth；And when continuous distributed is truly distributed as, the distribution character at the sample observation station cannot not obtained, so as to not
The distribution function for being closer to true distribution can be estimated.Original sample is ranked up by the Bootstrap algorithms after improvement
And its corresponding neighborhood is made, regeneration sample sheet is extracted according to neighborhood index.Bootstrap methods after improvement extend observation
Data, it is to avoid this case that only extract initial data, make estimation distribution as close as true distribution, improve the standard of estimation
True property.
Step 3：Kalman filtering algorithm to several node crystal oscillators when be partially corrected with frequency deviation, realize in crystal oscillator fast
While speed is stable, the stability and accuracy of node crystal oscillator are improved.
Fig. 1 is the interaction schematic diagram between master-salve clock, it is assumed that trip path is symmetrical, i.e., packet is in communication process
In time delay it is equal, then the time delay in communication process is equal, then：
Wherein offset be master-salve clock between when it is inclined；TTrans1And TTrans2Respectively receive master clock from clock to send out
The propagation delay and master clock of Sync packets is sent to receive from clock the propagation delay for sending Delay-Req packets；tm1Based on
Clock sends the time of Sync packets；ts2It is the time that Delay-Req packets are received and sent from clock；ts1It is to connect from clock
Receive the time of the Sync packets of master clock transmission；tm2The Delay-Req packets for sending are received from clock for master clock
Time.
By above-mentioned formula discretization, and think time deviation measure error size be 0, then to from its time deviation of clock
There is relationship below with frequency departure：
θ (k+1)=θ (k)-uθ(k)+[s(k)-us(k)]*τ+wθ(k) (3)
S (k+1)=s (k)-us(k)+ws(k) (4)
When inclined and frequency deviations of the wherein θ (k) and s (k) for k moment clocks；uθ(k) and usK () is that master-salve clock realizes that clock is same
During step, when partially and the secondary correction value of frequency deviation kth；τ is the sampling time of k-th sample, because sample frequency is certain, is then sampled
Constant period.Under conditions of small sample (i.e. the when inclined and frequency deviation sampled point of node is less than 10), using improved Bootstrap
Method carries out resampling to original sample, obtains new Bootstrap samples, and which is new observation data, assists in IEEE1588
Under view, new observation data are sent into into Kalman filter, correction is filtered to node crystal oscillator, improve the degree of accuracy of crystal oscillator.
Step 4：Level finds and node is collected：Layering is realized by the size sequence of node crystal oscillator clock slope, this
It is bright real-time performance to be layered in the form of dividing equally.Nodes P is had in assuming network, node is divided into Z groups, between group and group
It is separate, per group n containing node.Wherein n=P/Z.
Step 5：The node crystal oscillator in layer is merged using natural selection particle cluster algorithm per layer network, produce Z
Individual virtual master clock.
The search dimension and the total number of particles in space of population are set, the speed and position of particle are initialized.
L numbers are randomly selected respectively as the initial position of particle in the range of the inclined and estimated accuracy of frequency deviation when known, it is ordered
Entitled Xs、Xo, node number is used as space search dimension.The fitness function of particle is the important of the fusion accuracy of impact node
Factor, current invention assumes that fitness function is adaptive weighted function, shown in the weighted factor such as formula (5) of auto-adaptive function：
Wherein σdIt is the estimated standard deviation that each node filtered data are asked for as after valid data, D is node
Number,
Under the conditions of Monte Carlo simulation, the when inclined and offset estimation value N group after each node is stablized is randomly selected, per group
L, then the weighted factor of auto-adaptive function is：
Auto-adaptive function is defined as：
The position of particle and adaptive value are stored in into individual extreme value p of particlebestIn, all of individual extreme value is sorted simultaneously
Relatively, optimum individual extreme value is designated as global extremum gbest, gbestIt is then optimum search result.Colony is used in search procedure
The best half particle of middle adaptive value goes to replace worst half particle, as new displacement, and retains original per each and every one
The history optimal value remembered of body.Displacement, speed and the g of particle are constantly updated simultaneouslybest, all of particle is looked for newest
The globally optimal solution for arriving is close, stops search when the process reaches maximum search number of times M and output result.The displacement of particle and
It is as follows that speed updates relational expression：
xd,l(t+1)=xd,l(t)+vd,l(t+1) (8)
vd,l(t+1)=ω vd,l(t)+c1·r1·[pd,l-xd,l(t)]+c2·r2·[pg,l-xd,l(t)] (9)
Wherein i=1,2 ..., N；J=1,2 ..., D ω are inertia weight value；C1, c2 are Studying factors, generally take c1
=c2.r1And r2For the random number in the range of [0,1], t is iteration cycle.In maximum iteration time, each layer network is by upper
Stating process carries out effective integration by node, forms the virtual master clock with degree of precision.
Step 6：Understand n+1 node is had per layer network by above-mentioned steps.Per layer each node is with the local of oneself
On the basis of clock, the nodal information of oneself is periodically broadcasted to its neighbor node, the cycle is T, neighbor node receiving data bag
Information simultaneously records the clock readings at oneself current time.By the time of the packet information implementation level interior nodes of broadcast to letter
The collection of breath.
Step 7：Node in per layer network is according to ATS agreements and the time collected to information realization group internal clock synchronization
Process.Node j is in t0Moment is included hardware clock reading Hj(t0) packet be sent to its neighbor node i, node i
The local zone time H of oneself is recorded while the packet of node j transmissions is receivedi(t0), in the ideal case, what node j sent
Packet is instantaneously received.Node i is identical with the absolute reference time of node j, then node i record time to information be：(Hi
(t0),Hj(t0)), when a new synchronization packets are sent to node i by node j, node i records the new time to information
To (Hi(t1),Hj(t1)).A can be calculated according to Fig. 1ij：
ajiRepresent the relative time clock slope of node j and node i.Arbitrary node i with T as cycle, periodically to oneself
Neighbor node j broadcast the nodal information of oneself, its neighbor node j receives the packet information of the transmission of node i, and record is current
Clock readings, while record node i transmission packet information, if the node stored a group node i transmission
Packet information, then update Relative slope aij, logical timer slopeWith logical timer deviation
Wherein ρη, ρv, ρoFor initial given weight, its excursion for (0,1).
The maximum difference of the maximum difference and logical timer deviation of the logical timer slope between node and its neighbor node
It is defined as dsAnd do：
And if only if do(t)=dsDuring (t)=0, then realize between node i and node j synchronous.
Step 8：Z independent logical timer is obtained according to step 7, this Z independent logical timer is considered as into Z independently
Logical node, the linear clock models of cellular logic node are built by linear regression model (LRM), its expression formula is：
Sz(t)=szt+oz 1≤z≤Z(16)
Wherein, szFor the slope of z logical node clock, ozRepresent z-th logical node clock jitter.In the same manner, logic section
Point periodically broadcasts the temporal information of oneself to its neighbours' logical node on the basis of the local logical timer of oneself, collects
The step of, is assumed to reach in t per group node synchronous in group, and logical timer is equally with T as the logical timer of periodic broadcasting oneself
Information, the cycle is T, and neighbours' logical node receiving data package informatin simultaneously records the clock readings at oneself current time.Assisted using ATS
Time between negotiation logical node, to the clock synchronization between information realization logical node, finally realizes the clock synchronization of network
Process.
The present invention is filtered process to the when inclined and frequency deviation of the node crystal oscillator of wireless network under condition of small sample, is protecting
Card frequency deviation and when inclined amendment precision under conditions of shorten the sampling time.Existing hierarchical network and its improved method are all bases
In centralized synchronous protocol, change of the clock synchronization protocol based on reference mode to network topology structure is very sensitive, node
Increase or loss can all make whole network restart work, and worse synchronization accuracy may be caused.And the present invention is proposed
Realize that the method for hierarchical synchronization avoids the problems referred to above in distributed network, improve while original ATS agreement advantages are kept
The convergence rate of network, the time for making real-time performance synchronous shorten.And it is virtual by what is formed by natural selection particle cluster algorithm
Master clock is included in layer, is allowed to realize consistency synchronization with layer interior nodes, is the method increased the synchronization accuracy of network.
Claims (1)
1. a kind of hierarchical network clock synchronizing method based on ATS agreements, it is characterised in that comprise the following steps：
(1) master clock and P synchronization node clock are provided；
(2) process is filtered to the node time information collected by improving Bootstrap statistical methods；
(3) a point Z layers, and the time of collecting layer interior nodes are carried out to network to information；Calculated by natural selection population in layer
Method carries out data fusion to filtered nodal information, produces virtual master clock；
(4) the virtual master clock for obtaining is added in hierarchical network, then per layer contains n+1 node, n=P/Z；
(5) per the node in layer network according to ATS agreements and the time collected to information realization group internal clock synchronization, while
To Z independent logical timer；
(6) this Z independent logical timer is considered as into Z independent logical node, while which is with the local logical timer of itself
On the basis of, the temporal information of itself is periodically broadcasted to its logical node；
(7) network is same to the clock of information realization whole network again according to the time of ATS agreements and the logical node collected
Step.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610984696.5A CN106533602A (en) | 2016-11-09 | 2016-11-09 | ATS-protocol-based time synchronization method of hierarchical network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610984696.5A CN106533602A (en) | 2016-11-09 | 2016-11-09 | ATS-protocol-based time synchronization method of hierarchical network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106533602A true CN106533602A (en) | 2017-03-22 |
Family
ID=58350239
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610984696.5A Pending CN106533602A (en) | 2016-11-09 | 2016-11-09 | ATS-protocol-based time synchronization method of hierarchical network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106533602A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111416785A (en) * | 2020-02-25 | 2020-07-14 | 重庆邮电大学 | Method for estimating relative frequency offset of consistent clock synchronization based on weighted median |
| CN111812970A (en) * | 2020-07-22 | 2020-10-23 | 桂林电子科技大学 | Double-compensation clock synchronization method based on IEEE1588 protocol |
| CN113014344A (en) * | 2019-12-19 | 2021-06-22 | 中国科学院沈阳自动化研究所 | Time synchronization method of redundant network |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102196554A (en) * | 2011-06-10 | 2011-09-21 | 西安电子科技大学 | Method for realizing average clock synchronization of wireless sensor network through cluster averaging |
| CN105188126A (en) * | 2015-08-04 | 2015-12-23 | 西北工业大学 | Distributed multi-hop wireless network clock synchronization method based on mean field |
-
2016
- 2016-11-09 CN CN201610984696.5A patent/CN106533602A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102196554A (en) * | 2011-06-10 | 2011-09-21 | 西安电子科技大学 | Method for realizing average clock synchronization of wireless sensor network through cluster averaging |
| CN105188126A (en) * | 2015-08-04 | 2015-12-23 | 西北工业大学 | Distributed multi-hop wireless network clock synchronization method based on mean field |
Non-Patent Citations (3)
| Title |
|---|
| LUCA SCHENATO, GIOVANNI GAMBA: "A distributed consensus protocol for clock synchronization in wireless sensor network", 《IEEE》 * |
| YILIN JIANG,FANGYUAN ZHANG,MURAN GUO: "High-precision clock synchronization based on PSO algorithm under the small sample condition", 《IEEE》 * |
| 蒋伊琳，张芳园，郑辉: "Bootstrap方法对晶振稳定度估计", 《沈阳工业大学学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113014344A (en) * | 2019-12-19 | 2021-06-22 | 中国科学院沈阳自动化研究所 | Time synchronization method of redundant network |
| CN111416785A (en) * | 2020-02-25 | 2020-07-14 | 重庆邮电大学 | Method for estimating relative frequency offset of consistent clock synchronization based on weighted median |
| CN111812970A (en) * | 2020-07-22 | 2020-10-23 | 桂林电子科技大学 | Double-compensation clock synchronization method based on IEEE1588 protocol |
| CN111812970B (en) * | 2020-07-22 | 2022-11-11 | 桂林电子科技大学 | Double-compensation clock synchronization method based on IEEE1588 protocol |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102869084B (en) | For synchronizing the method and apparatus with compensating clock drift between communicator | |
| Swain et al. | A model for the classification and survey of clock synchronization protocols in WSNs | |
| CN104519563B (en) | One kind layering wireless sensor network time synchronization method | |
| CN106452650B (en) | Clock synchronizing frequency deviation estimating method suitable for multi-hop wireless sensor network | |
| CN101466142B (en) | Synchronization method for layered time comparative clock in wireless sensor network | |
| He et al. | Accurate clock synchronization in wireless sensor networks with bounded noise | |
| CN104053223A (en) | Low-power wireless sensor network time synchronization method | |
| CN105743598B (en) | A kind of Industrial Ethernet clock synchronizing method and system | |
| Huang et al. | Long term and large scale time synchronization in wireless sensor networks | |
| CN106533602A (en) | ATS-protocol-based time synchronization method of hierarchical network | |
| CN101917759A (en) | Wireless sensor network time synchronization method based on linear fit | |
| Huang et al. | An accurate on-demand time synchronization protocol for wireless sensor networks | |
| Tian et al. | Delay compensation-based time synchronization under random delays: Algorithm and experiment | |
| CN105188126A (en) | Distributed multi-hop wireless network clock synchronization method based on mean field | |
| CN102196554A (en) | Method for realizing average clock synchronization of wireless sensor network through cluster averaging | |
| CN106912100A (en) | Appliance network method for synchronizing time based on TPSN and TSync | |
| Gong et al. | Temperature compensated Kalman distributed clock synchronization | |
| CN103957591A (en) | Node pair based wireless sensor network time synchronization method | |
| CN103298099A (en) | Time synchronizing method based on bimodal clock frequency estimation | |
| CN106162858A (en) | Directintermination time synchronization method and device | |
| CN106604387A (en) | Wireless sensor time synchronization method based on game theory | |
| Tabassum et al. | Clock synchronization in wireless sensor networks using least common multiple | |
| CN107343314B (en) | Time synchronization method and device of wireless sensor network | |
| CN1965283A (en) | Decentralized time interval synchronization in distributed networks | |
| CN108712769A (en) | A kind of wireless sensor network synchronous method based on dynamic route and variable period |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170322 |
|
| WD01 | Invention patent application deemed withdrawn after publication |