CN102752843A - Time synchronization method - Google Patents

Abstract

The invention relates to a time synchronization method. The time synchronization method is applied to a wireless sensor network with discontinuous communication, and comprises the steps of carrying out opportunity synchronization on sensor nodes in the wireless sensor network during each synchronous cycle, and carrying out prediction calibration on the sensor node without being synchronized to correct time after the opportunity synchronization is finished, thus achieving the synchronization. According to the time synchronization method, the nodes of the whole wireless sensor network are synchronized in a mode combining the opportunity synchronization and the prediction calibration without dependence on a specified network topology structure, thereby well adapting to network topology change caused by the moving of the nodes or sleeping schedule, reducing the energy consumption of the nodes, and saving the resources of wireless channels.

Description

A kind of method for synchronizing time

Technical field

The present invention relates to the communications field, relate in particular to a kind of method for synchronizing time.

Background technology

Wireless sensor network is a distributed system, because ardware feature and operational environment is different, the time of sensor node tends to depart from the standard time.Many functions in the wireless sensor network are closely related with the time with service, normally work in an orderly manner for guaranteeing wireless sensor network, need a kind of effective method for synchronizing time that the time of the whole network node is carried out synchronously.

In some wireless sensor network application scenarioss, need utilize the mobility of sensor node to increase the monitoring range and the ability of network.In addition, in order to improve the energy utilization efficiency of network, the operating state that needs employing dormancy dispatching mechanism to come management node usually is to reduce unnecessary energy expense.Yet node motion or dormancy meeting destroy the connectedness of network, cause part of nodes in some periods can't with node communication on every side, thereby make network present intermittently connected characteristics.Existing time synchronization protocol and method are primarily aimed at the stable sensor network of topological structure and design; RBS (Reference Broadcast Synchronization for example;) agreement, TPSN (Timing-sync Protocol for Sensor Networks;) agreement and FTPS (Flooding Time Synchronization Protocol) agreement etc., all do not consider the intermittently connected problem of network that node motion or dormancy cause.In intermittently connected sensor network, if adopt existing method for synchronizing time to carry out synchronously, part of nodes possibly can't receive the synchronizing information in the network and lose the synchronous motor meeting term of execution of synchronizing process.Though can deal with this problem through the execution frequency that increases synchronizing process in theory, carry out the energy that synchronizing process not only can the lot of consumption node continually, also can tie up the limited wireless channel resource, influence the operate as normal of sensor network.

Summary of the invention

Technical problem to be solved by this invention provides a kind of method for synchronizing time, solves the time synchronization problem in the intermittently connected sensor network, and reduces node energy consumption, practices thrift radio channel resource.

For solving the problems of the technologies described above, the present invention proposes a kind of method for synchronizing time, be applied to intermittently connected wireless sensor network, comprising:

When arrived each synchronizing cycle, it is synchronous that the sensor node in the said wireless sensor network carries out chance;

After the synchronous process of said chance was ended, the synchronous sensor node of being unrealized was predicted and is calibrated the time of revising oneself, to realize synchronously.

Further, above-mentioned method for synchronizing time also can have following characteristics, and said chance comprises synchronously:

Aggregation node in the said wireless sensor network comprises the synchronizing information of self temporal information to said wireless sensor network broadcasting with the synchronizing cycle set;

The sensor node that directly or indirectly is communicated with said aggregation node; Receive and realize the time synchronized with said prime node according to the synchronizing information that the prime node of this sensor node sends; The synchronizing information that said prime node sends comprises the temporal information of said prime node; Send the synchronizing information comprise self temporal information then; It is synchronous until the afterbody node deadline to repeat this step, and the prime node of said sensor node is meant prior to said sensor node realizes synchronous sensor node.

Further; Above-mentioned method for synchronizing time also can have following characteristics; Said prime node is called sending node; With receiving and realizing being called receiving node with the synchronous back level node of this prime node time according to the synchronizing information that this prime node sends, the synchronizing process between then said sending node and the said receiving node comprises:

Said sending node when detecting channel idle outwards one of broadcasting only comprise the brief control messages of self identification number, and write down the local zone time tsend of this brief control messages delivery time;

Brief control messages and the recorder that said receiving node the receives said sending node local zone time t constantly that finishes _Receive

Said sending node is with the time t of said brief control messages delivery time _SendWrite in the synchronization message, send to said receiving node;

Said receiving node receives said synchronization message and according to the logical time of its temporal information correction that comprises oneself, correction formula is L (t)=L (t) '-(t _Receive-t _Send-n τ), wherein L (t) ' is the logical time before the said receiving node correction, and n is the bit number of said brief control messages, and τ is for sending the time of unit Bit data.

Further, above-mentioned method for synchronizing time also can have following characteristics, and said brief control messages comprises sequence number, node serial number; Said synchronization message comprises start-up time, the maximum duration of chance synchronizing process, the synchronizing cycle of transmitting time, the epicycle synchronizing cycle of sequence number, node serial number, brief control messages.

Further, above-mentioned method for synchronizing time also can have following characteristics, and in the said chance synchronizing process, said sending node is confirmed the quantity forwarded of said synchronization message according to following steps:

With the time slot that is divided into equal length each synchronizing cycle, said prime node judges whether the epicycle chance is overtime synchronously, if not overtime then with P (k)=1-β ^kProbability judgement current time slots in whether send synchronization message, wherein, β is for regulating parameter, k is the sequence number of current time slots;

If said prime node decision is not sent synchronization message in current time slots, then intercept of the arrival of the state of channel until next time slot;

If said prime node does not receive brief control messages and synchronization message that other node sends in intercepting the process of channel; Then in next time slot, repeat preceding two steps; Otherwise calculate the Duplication λ of two node communication scopes according to the node location information that synchronization message is carried, and adjudicate next time slot with the probability of P=1-λ and whether repeat preceding two steps.

Further, above-mentioned method for synchronizing time also can have following characteristics, and said prediction calibration comprises:

The synchronous not synchronization node of in said chance synchronizing process, being unrealized, according to record from the hardware time H (t when preceding n realizes that chance is synchronous in the synchronizing cycle ₀), H (t ₁) ..., H (t _n) and revised logical time L (t ₀), L (t ₁) ..., L (t _n), calculate certainly in the hardware time drift rate of preceding n in the synchronizing cycle ${\mathrm{\ρ}}_i=\frac{H\left(t_i\right)-H\left(t_{i-1}\right)}{L\left(t_i\right)-L\left(t_{i-1}\right)},$ (i=1,2 ..., n);

To the hardware time drift rate sequence { ρ in preceding n synchronizing cycle of said not synchronization node ₁, ρ _N-1..., ρ _nCarry out prediction and calculation, and obtaining the hardware time drift rate of said not synchronization node in n+1 synchronizing cycle, the formula of prediction and calculation does ${\widehat{\mathrm{\ρ}}}_{n+1}={\▿}^2{\widehat{\mathrm{\ρ}}}_{n+1}+{\▿\mathrm{\ρ}}_n+{\mathrm{\ρ}}_n;$

According to the hardware time drift rate of said not synchronization node in n+1 synchronizing cycle the logical time of said not synchronization node is revised, correction formula does $L\left(t\right)=L\left(t_n\right)+{\widehat{\mathrm{\ρ}}}_{n+1}[H\left(t\right)-H\left(t_n\right)],$ Wherein t representes the Current Standard time, and H (t) representes current hardware time of said not synchronization node, H (t _n) and L (t _n) represent last hardware time and the logical time of realizing when synchronous of said not synchronization node respectively.

Further, above-mentioned method for synchronizing time also can have following characteristics, and chance is synchronous if said not synchronization node was unrealized in i synchronizing cycle, then in prediction

The time, said hardware time drift rate sequence { ρ ₁, ρ _N-1..., ρ _nIn ρ _iObtained by prediction in the last synchronizing cycle

Replace.

The mode that method for synchronizing time employing chance of the present invention is synchronous and the prediction calibration combines is carried out synchronously the whole network node of wireless sensor network; Do not rely on the particular network topological structure; Can adapt to the network topology change that node motion or dormancy dispatching cause well; Reduce the consumption of node energy, practiced thrift radio channel resource.And method for synchronizing time of the present invention adopts the message mode of " send earlier, then write down, redispatch ", has eliminated the influence of sync message transmission lag to synchronization accuracy, has improved the precision of time synchronized.

Description of drawings

Fig. 1 is the sketch map of chance synchronizing process in the embodiment of the invention;

Fig. 2 is the sketch map of BTS message;

Fig. 3 is the sketch map of TSM message;

Fig. 4 confirms the flow chart of synchronization message quantity forwarded for node in the wireless sensor network;

Fig. 5 is the Duplication algorithm sketch map of two node communication scopes.

Embodiment

Main design of the present invention is: adopt chance to come the whole network node of intermittently connected wireless sensor network is carried out time synchronized with the mode that the prediction calibration combines synchronously.Wherein, chance is meant that synchronously synchronizing process does not rely on the particular network topological structure, and the communication opportunity that utilizes node motion or dormancy awakening to bring carries out synchronously; Prediction calibration is meant that node calibrates the time of oneself through predicting the hardware time drift of oneself, is consistent with the standard time as far as possible.

Method for synchronizing time of the present invention is a kind of method for synchronizing time towards intermittently connected sensor network (A Time Synchronization Method for Intermittently Connected Sensor Networks is called for short TSIC).

Sensor network of the present invention is a wireless network.

Below in conjunction with accompanying drawing principle of the present invention and characteristic are described, institute gives an actual example and only is used to explain the present invention, is not to be used to limit scope of the present invention.

In intermittently connected sensor network, network topology presents the characteristics of dynamic change, so the whole network synchronizing process can not depend on the particular network topological structure.

Method for synchronizing time of the present invention at first adopts the synchronous mode of chance to come the whole network is carried out synchronously, and this chance synchronizing process can comprise the steps:

Step 1, the synchronizing information that synchronizing cycle to wireless sensor network broadcasting comprise the temporal information of Sink node of Sink node (being aggregation node) to set;

Wherein, synchronizing information can be a sync message.

The Sink node can be with the synchronizing information of regular time interval △ t (being synchronizing cycle) broadcasting oneself.

Synchronizing process is initiated by the Sink node, because have only the time of Sink node and standard time to be consistent in the sensor network.

Step 2; With the back level node that the Sink node directly or indirectly is communicated with, receive and realize and the time synchronized of prime node, wherein according to the synchronizing information that the prime node of this back level node sends; The synchronizing information that the prime node sends comprises the temporal information of this prime node; Deadline synchronously after, back level node sends the synchronizing information that comprises self temporal information, it is only synchronous to the afterbody node deadline to repeat this step.

Need to prove that among this paper, the rank of sensor node is meant the time rank, realize that earlier synchronous sensor node is called the prime node, realize that according to this prime node synchronous sensor node is called the back level node of this prime node.Therefore; The fixedly hierarchical relationship that does not have the one-level one-level between the sensor node; The time rank of sensor node just is provided with at receiving synchronous information and after realizing synchronously; Its effect is as follows: if certain not synchronization node receive more than two the synchronizing information of synchronization node, this node that synchronization node is can the select time rank not low is as synchronization object, with the higher synchronization accuracy of acquisition so.

The node of receiving synchronous information is called receiving node; The node that sends synchronizing information is called sending node; Then first order node (being the Sink node) is a sending node; The afterbody node is a receiving node, other nodes except that first order node and afterbody node be sending node be again receiving node.Receiving node is at first synchronous with the sending node of synchronizing information, and the follow-up moment according to network and self state, dynamically the synchronizing information of broadcasting oneself to other not synchronization node carry out synchronously.

The synchronous process of chance is as shown in Figure 1.Among Fig. 1, the black round dot is represented working node, and white round dot is represented the dormancy node, and the dotted line of band arrow is represented the transmission direction of synchronizing information in the chance synchronizing process.

In the step 2,, receive step by step and realize to comprise the steps: with the time synchronized of prime node according to the synchronizing information that its prime node sends with the back level node that the Sink node directly or indirectly is communicated with

Prime node (here for sending node) when detecting channel idle outwards one of broadcasting only comprise the brief control messages BTS (Begin To Synchronization) of self identification number (ID number), and write down the local zone time t of this BTS message delivery time _Send

BTS message and the recorder that back level node (being receiving node here) receives the prime node local zone time t constantly that finishes _Receive

The prime node is with the time t of BTS message delivery time _SendWrite among the synchronization message TSM (Time Sync Message), send to back level node;

Level node in back receives synchronization message and the logical time of the temporal information correction oneself that comprises according to this synchronization message is L (t)=L (t) '-(t _Receive-t _Send-n τ), wherein L (t) ' is the logical time before the level node correction of back, and n is the bit number of BTS message, and τ is for sending the time of unit Bit data.

The form of BTS message is as shown in Figure 2, and the form of TSM message is as shown in Figure 3.As shown in Figure 2, BTS message comprises sequence number (SEQ) and node serial number (Node ID).Wherein, sequence number accounts for 1byte (byte), and node serial number accounts for 2byte.As shown in Figure 3, TSM message comprises sequence number (SEQ), node serial number (Node ID), synchronous round (Sync Cycle), transmitting time (Sending Time), start-up time (Starting Time), duration (Duration), sync interval (Sync Interval) and node location (Node Location).Wherein, sequence number accounts for 1byte (byte), and node serial number accounts for 2byte, and round accounts for 2byte synchronously, and transmitting time accounts for 4byte, accounts for 4byte start-up time, and the duration accounts for 4byte, and sync interval accounts for 4byte, and node location accounts for 4byte.Wherein, transmitting time is meant the transmitting time t of BTS message _Send, being meant epicycle start-up time of synchronizing cycle start-up time, the duration is meant the maximum duration t of chance synchronizing process _Duration, sync interval is meant △ t blanking time of cycle synchronisation, just synchronizing cycle.

Whether do not set up any network topology structure in the chance synchronizing process, therefore, node is difficult to obtain the state information of other node in its communication range of current time, for example whether be in resting state, realized synchronously.In order to prevent that too much redundant synchronization message from appearring in node density when big; Avoid simultaneously node density hour to influence the convergence rate of synchronizing process again because of the synchronization message lazy weight; In the chance synchronizing process; The prime node is confirmed the quantity forwarded of synchronization message according to flow process shown in Figure 4, and this confirms that the strategy of synchronization message quantity forwarded is called probability and keeps out of the way strategy.

As shown in Figure 4, in the chance synchronizing process, the prime node is confirmed the quantity forwarded of synchronization message according to following steps:

Step 401, the realization of prime node is synchronous with other nodes;

Step 402 judges whether chance is overtime synchronously, if execution in step 412 then, otherwise execution in step 403;

With the time slot that is divided into equal length each synchronizing cycle, all carry out the judgement of step 402 at each time slot.

Step 403 is with P (k)=1-β ^kThe probabilistic determination current time slots whether send synchronizing information;

Wherein, β is for regulating parameter, and k is the sequence number of current time slots.

Step 404 judges whether that decision sends synchronizing information, if execution in step 407 then, otherwise execution in step 405;

Step 405 is sent BTS message and TSM message, execution in step 412;

Step 406 gets into next time slot, execution in step 402;

Step 407 is intercepted channel;

Step 408 judges whether to receive TSM message, if execution in step 409 then, otherwise execution in step 406;

Step 409, the Duplication λ of computing node scope;

The Duplication lambda definition of communication range is the overlapping area of two node communication scopes and the ratio of the area of individual node communication range between the node.As shown in Figure 5, establish node N _iWith node N _jDistance be d _Ij, communication radius all is R, the Duplication of two node communication scopes calculates according to following formula (1) so:

$\mathrm{\λ}=\frac{1}{{\mathrm{\πR}}^2}[2R^2\arccos (\frac{d_{\mathrm{ij}}}{2}/R)-2d_{\mathrm{ij}}\sqrt{R^2-{\left(\frac{d_{\mathrm{ij}}}{2}\right)}^2}]---\left(1\right)$

Whether step 410 continues with the next time slot of the probabilistic determination of P=1-λ,

Step 411, if the judged result of step 410 is for continuing then execution in step 406, otherwise execution in step 412;

Step 412, shut-down operation.

Since node move or the dormancy meeting causes part of nodes can't receive synchronizing information, therefore, the chance synchronizing process does not guarantee that all nodes in the wireless sensor network can both realize synchronously.In the chance synchronizing process, this part owing to do not receive synchronizing information lose synchronous motor can node waiting for t in the preamble period _DurationDo not receive BTS message and TSM message yet after (the maximum duration of chance synchronizing process) time, judge that then epicycle chance synchronizing process ends, and start the prediction alignment mechanism and revise own logical time, realize synchronous with the standard time.

The important prerequisite of prediction calibration is to estimate when the preamble period interior nodes drift rate of hardware time.Drift rate equals the ratio of hardware time and the timing length of standard time.If node in the synchronizing cycle, is t in the standard time at preceding n respectively ₀, t ₁..., t _nThe time and other node realize synchronously, and note the hardware time H (t of oneself ₀), H (t ₁) ..., H (t _n) and revised logical time L (t ₀), L (t ₁) ..., L (t _n), node can use following formula (2) to represent in the hardware time drift rate of preceding n in the synchronizing cycle so:

${\mathrm{\ρ}}_i\frac{H\left(t_i\right)-H\left(t_{i-1}\right)}{t_i-t_{i-1}},(i=\mathrm{1,2},\·\·\·,n)---\left(2\right)$

Because node is difficult to obtain each standard time constantly, and very approaching through revised logical time and standard time, L (t is promptly arranged _i) ≈ t _i, (i=0,1 ..., n), if ignore small synchronous error, then formula (2) can change following formula (3) into:

${\mathrm{\ρ}}_i=\frac{H\left(t_i\right)-H\left(t_{i-1}\right)}{L\left(t_i\right)-L\left(t_{i-1}\right)},(i=\mathrm{1,2},\·\·\·,n)---\left(3\right)$

The drift of node hardware time is to cause that both deviations are big more because the actual operating frequency of crystal oscillator departs from nominal frequency, and the drift rate of hardware time is also just big more.The operating frequency of crystal oscillator mainly receives the influence of ambient temperature, and has the characteristics of nonlinear change.Because ambient temperature is a continually varying in time, this just means that the hardware time drift rate in each synchronizing cycle is a nonlinear time series.If the hardware time drift rate of preceding n in the synchronizing cycle is { ρ ₁, ρ _N-1..., ρ _n, among the present invention, can adopt the second differnce exponential smoothing to come this time series is carried out prediction and calculation, obtain the hardware time drift rate in n+1 synchronizing cycle.The second differnce exponential smoothing is a kind of combination forecasting method towards Nonlinear Time Series, and its fundamental formular is following:

$\left\{\begin{array}{c}{\▿\mathrm{\ρ}}_n={\mathrm{\ρ}}_n-{\mathrm{\ρ}}_{n-1}\\ {\▿}^2{\mathrm{\ρ}}_n={\▿\mathrm{\ρ}}_n-{\▿\mathrm{\ρ}}_{n-1}\\ {\▿}^2{\widehat{\mathrm{\ρ}}}_{n+1}={\mathrm{\α}\▿}^2{\mathrm{\ρ}}_n+(1-\mathrm{\α}){\▿}^2{\widehat{\mathrm{\ρ}}}_n\end{array}\right.---\left(4\right)$

In the formula (4); α ∈ (0; 1) is weight coefficient; The reaction speed of representing forecast model that time series data is changed, theoretical value is 2/ (n+1), is the difference mark.After obtaining the result of calculation of formula (4), the hardware time drift rate that just can predict n+1 synchronizing cycle by following formula (5):

${\widehat{\mathrm{\ρ}}}_{n+1}={\▿}^2{\widehat{\mathrm{\ρ}}}_{n+1}+{\▿\mathrm{\ρ}}_n+{\mathrm{\ρ}}_n---\left(5\right)$

After doping the hardware time drift rate of node in next synchronizing cycle, just can revise in real time the logical time of node, concrete formula is following:

$L\left(t\right)=L\left(t_n\right)+{\widehat{\mathrm{\ρ}}}_{n+1}[H\left(t\right)-H\left(t_n\right)]---\left(6\right)$

In the formula (6), t representes the Current Standard time, and H (t) representes current hardware time of node, H (t _n) and L (t _n) represent last hardware time and the logical time of realizing when synchronous of node respectively.

Need to prove that node has several times at preceding n and fails to realize that chance is synchronous in the synchronizing cycle, following to the processing method of this situation: chance is synchronous if node was unrealized in i synchronizing cycle, causes ρ _iBy formula calculate (6), so in prediction and calculation

The time, sequence { ρ ₁, ρ _N-1..., ρ _nIn the i item by having obtained

Replace.

The mode that method for synchronizing time employing chance of the present invention is synchronous and the prediction calibration combines is carried out synchronously the whole network node of wireless sensor network; Do not rely on the particular network topological structure; Can adapt to the network topology change that node motion or dormancy dispatching cause well; Reduce the consumption of node energy, practiced thrift radio channel resource.

In addition, the message mode of the synchronous employing " send record, back earlier " in the existing method for synchronous between two nodes, promptly sending node is stamped timestamp earlier on sync message, and then sync message is sent to receiving node.Because there are certain deviation in the actual transmitting time of sync message and the time of timestamp mark, the message mode of " send record, back earlier " can make synchronization accuracy receive influence to a certain degree.And method for synchronizing time of the present invention adopts the message mode of " send earlier, then write down, redispatch ", and sending node can accurately write down the transmitting time of BTS message, and meanwhile, receiving node can accurately write down the concluding time of BTS message; Subsequently, sending node writes sync message with the transmitting time of BTS message, and sends to receiving node, has eliminated the influence of sync message transmission lag to synchronization accuracy.Therefore, method for synchronizing time of the present invention has improved the precision of time synchronized.

The above is merely preferred embodiment of the present invention, and is in order to restriction the present invention, not all within spirit of the present invention and principle, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (7)

1. a method for synchronizing time is applied to intermittently connected wireless sensor network, it is characterized in that, comprising:

When arrived each synchronizing cycle, it is synchronous that the sensor node in the said wireless sensor network carries out chance;

After the synchronous process of said chance was ended, the synchronous sensor node of being unrealized was predicted and is calibrated the time of revising oneself, to realize synchronously.

2. method for synchronizing time according to claim 1 is characterized in that, said chance comprises synchronously:

Aggregation node in the said wireless sensor network comprises the synchronizing information of self temporal information to said wireless sensor network broadcasting with the synchronizing cycle set;

The sensor node that directly or indirectly is communicated with said aggregation node; Receive and realize the time synchronized with said prime node according to the synchronizing information that the prime node of this sensor node sends; The synchronizing information that said prime node sends comprises the temporal information of said prime node; Send the synchronizing information comprise self temporal information then; It is synchronous until the afterbody node deadline to repeat this step, and the prime node of said sensor node is meant prior to said sensor node realizes synchronous sensor node.

3. method for synchronizing time according to claim 2; It is characterized in that; Said prime node is called sending node; With receiving and realizing being called receiving node with the synchronous back level node of this prime node time according to the synchronizing information that this prime node sends, the synchronizing process between then said sending node and the said receiving node comprises:

Said sending node when detecting channel idle outwards one of broadcasting only comprise the brief control messages of self identification number, and write down the local zone time t of this brief control messages delivery time _Send

Brief control messages and the recorder that said receiving node the receives said sending node local zone time t constantly that finishes _Receive

Said sending node is with the time t of said brief control messages delivery time _SendWrite in the synchronization message, send to said receiving node;

Said receiving node receives said synchronization message and according to the logical time of its temporal information correction that comprises oneself, correction formula is L (t)=L (t) '-(t _Receive-t _Send-n τ), wherein L (t) ' is the logical time before the said receiving node correction, and n is the bit number of said brief control messages, and τ is for sending the time of unit Bit data.

4. method for synchronizing time according to claim 3 is characterized in that, said brief control messages comprises sequence number, node serial number; Said synchronization message comprises start-up time, the maximum duration of chance synchronizing process, the synchronizing cycle of transmitting time, the epicycle synchronizing cycle of sequence number, node serial number, brief control messages.

5. method for synchronizing time according to claim 3 is characterized in that, in the said chance synchronizing process, said sending node is confirmed the quantity forwarded of said synchronization message according to following steps:

With the time slot that is divided into equal length each synchronizing cycle, said prime node judges whether the epicycle chance is overtime synchronously, if not overtime then with P (k)=1-β ^kProbability judgement current time slots in whether send synchronization message, wherein, β is for regulating parameter, k is the sequence number of current time slots;

If said prime node decision is not sent synchronization message in current time slots, then intercept of the arrival of the state of channel until next time slot;

If said prime node does not receive brief control messages and synchronization message that other node sends in intercepting the process of channel; Then in next time slot, repeat preceding two steps; Otherwise calculate the Duplication λ of two node communication scopes according to the node location information that synchronization message is carried, and adjudicate next time slot with the probability of P=1-λ and whether repeat preceding two steps.

6. method for synchronizing time according to claim 1 is characterized in that, said prediction calibration comprises:

The synchronous not synchronization node of in said chance synchronizing process, being unrealized, according to record from the hardware time H (t when preceding n realizes that chance is synchronous in the synchronizing cycle ₀), H (t ₁) ..., H (t _n) and revised logical time L (t ₀), L (t ₁) ..., L (t _n), calculate certainly in the hardware time drift rate of preceding n in the synchronizing cycle ${\mathrm{\ρ}}_i=\frac{H\left(t_i\right)-H\left(t_{i-1}\right)}{L\left(t_i\right)-L\left(t_{i-1}\right)},$ (i=1,2 ..., n);

To the hardware time drift rate sequence { ρ in preceding n synchronizing cycle of said not synchronization node ₁, ρ _N-1..., ρ _nCarry out prediction and calculation, and obtaining the hardware time drift rate of said not synchronization node in n+1 synchronizing cycle, the formula of prediction and calculation does ${\widehat{\mathrm{\ρ}}}_{n+1}={\▿}^2{\widehat{\mathrm{\ρ}}}_{n+1}+{\▿\mathrm{\ρ}}_n+{\mathrm{\ρ}}_n;$

According to the hardware time drift rate of said not synchronization node in n+1 synchronizing cycle the logical time of said not synchronization node is revised, correction formula does $L\left(t\right)=L\left(t_n\right)+{\widehat{\mathrm{\ρ}}}_{n+1}[H\left(t\right)-H\left(t_n\right)],$ Wherein t representes the Current Standard time, and H (t) representes current hardware time of said not synchronization node, H (t _n) and L (t _n) represent last hardware time and the logical time of realizing when synchronous of said not synchronization node respectively.

7. method for synchronizing time according to claim 6 is characterized in that, chance is synchronous if said not synchronization node was unrealized in i synchronizing cycle, then in prediction

The time, said hardware time drift rate sequence { ρ ₁, ρ _N-1..., ρ _nIn ρ _iObtained by prediction in the last synchronizing cycle

Replace.

Images