CN103699455B - In a kind of multinuclear real-time fault tolerance system, obtain accurately the method for deadline the latest - Google Patents
In a kind of multinuclear real-time fault tolerance system, obtain accurately the method for deadline the latest Download PDFInfo
- Publication number
- CN103699455B CN103699455B CN201310739053.0A CN201310739053A CN103699455B CN 103699455 B CN103699455 B CN 103699455B CN 201310739053 A CN201310739053 A CN 201310739053A CN 103699455 B CN103699455 B CN 103699455B
- Authority
- CN
- China
- Prior art keywords
- task
- node
- deadline
- latest
- tasks
- 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.)
- Expired - Fee Related
Links
Landscapes
- Debugging And Monitoring (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention proposes in a kind of multinuclear real-time fault tolerance system and obtain accurately the method for deadline the latest, comprise the steps: according to given task scheduling, on the basis that keeps legacy data to rely on, for increasing dispatching sequence, the adjacent task of scheduling on same core rely on; Increase the virtual task node that two times of implementation are 0, make in task scheduling, one of them dummy node is prior to all tasks carryings, after another one in all tasks carryings; Suppose to occur at most X soft error in task implementation, on the basis of original scheduling, on same core, re-execute immediately to realize fault-tolerant by the task of makeing mistakes, and the mission critical of definite set of tasks, deadline the latest accurately of obtaining set of tasks. If comprise N task in set of tasks, and occur at most in the process of implementation X soft error, the present invention can determine at O (n^2) deadline the latest accurately of set of tasks in the time, ensure fault-tolerant, efficiently quick.
Description
Technical field
The present invention relates to field, be specifically related to obtain accurately in a kind of multinuclear real-time fault tolerance system deadline the latestMethod.
Background technology
The real-time application of multinuclear real-time system is made up of a lot of tasks conventionally, and requires these tasks need to be in when cut-offBetween before complete. For example, in a multinuclear real-time system, between task, there is dependence, generally represent with directed acyclic graphThese tasks.
Fig. 1 (a) is a task scheduling figure, has expressed the scheduled for executing order of task in figure, and task is assigned toOn which core, carry out, and Starting Executing Time. On some core, in implementation, likely there is soft error in task, forEnsure obtain correct result, must provide fault-tolerant. A kind of simple and common method is to allow this task on same coreRe-execute immediately, as shown in Fig. 1 (b), in figure, shade is making a mistake of task.
In existing method, in multinuclear real-time fault tolerance system, the computational methods of in real time application deadline are to allow tool on each coreHave maximum execution time task occur wrong to obtain the deadline the latest of task on each core, if between taskThere is no dependence, this computational methods are accurately. For example, in Fig. 1 (b), on each core, there is respectively maximum execution timeThere are two mistakes in task, be 19 chronomeres the deadline that twice mistake occurs task T4, and mistake occurs twice task T2Be 15 chronomeres the deadline of mistake, and be 14 chronomeres the deadline that twice mistake occurs task T3, therefore shouldBe 19 chronomeres the optimum deadline of multinuclear real time system application operation. But, if there is dependence between task,This method is just inaccurate. For example, in Fig. 1 (a), 6 tasks are distributed on three cores. Arrow in figure represents between taskData dependence relation, the communication delay between the task on same core and in different IPs is assumed that respectively 0 and 1. Fig. 1 (c)In situation about showing, there is fault twice in T4, and be 19 chronomeres deadline. And in Fig. 1 (d), processor P 0 is carried outThere is fault twice in task T2, be 20 chronomeres deadline. Clearly, worse off cake shown in Fig. 1 (d). Therefore,The computational methods of existing deadline may have been underestimated truth.
In addition, because the number that each task makes a mistake is uncertain, existing method is exhaustive all possible situation, entersRow compares, the then selective cut-off time. The shortcoming of this method is: after determining that how many mistakes occur each task, obtainThe time complexity that obtains the deadline the latest of set of tasks is O (N2), but because needs are considered all situations: there is X in N taskIndividual mistake is totalThe situation of kind, therefore total time complexity isLose time very much.
Summary of the invention
In order to overcome the defect existing in above-mentioned prior art, the object of this invention is to provide a kind of multinuclear real-time fault tolerance systemThe present invention in system, obtains accurately the method for deadline the latest, when can efficiently obtain the cut-off the latest of set of tasks rapidlyBetween.
In order to realize above-mentioned purpose of the present invention, the invention provides in a kind of multinuclear real-time fault tolerance system and obtain accuratelyThe method of deadline the latest, it comprises the steps:
S1, according to the task scheduling of multiple nucleus system, on the basis that keeps legacy data to rely on, for scheduling is on same coreThe adjacent task of carrying out increases dispatching sequence and relies on, and sets up new directed acyclic graph;
S2, on the basis of new directed acyclic graph, increases by two virtual task nodes, when the execution of described virtual taskBetween be 0, in the task scheduling taking this directed acyclic graph as model, one of them virtual task node is carried out at first, other oneIndividual virtual task node is finally carried out, and because its time of implementation is 0, does not increase deadline the latest;
S3, supposes in task implementation, to occur at most X soft error, on the basis of original scheduling, appoints by makeing mistakesBusiness re-executing immediately on same core realized fault-tolerant, ensures the correct execution of set of tasks, and definite set of tasksMission critical, obtains set of tasks deadline the latest accurately;
Whether S4, meet the work need in current multinuclear real-time fault tolerance system the deadline the latest of measurement set of tasksAsk, if met, exit, if do not met, adjust scheduling strategy, return to step S1.
In a kind of preferred embodiment of the present invention, described step S1 specifically comprises the steps:
S11, obtain the number N of task in set of tasks, and in set of tasks implementation, contingent maximum is softThe number X of mistake
S12, represents the data dependence between task with directed acyclic graph;
S13, according to given task scheduling, if two tasks are scheduled on same core carries out and dispatching sequence's phaseNeighbour, the dispatching sequence who increases between these two tasks relies on, and on the basis of original directed acyclic graph, increases expression schedulingThe limit of order dependent, obtains new directed acyclic graph;
S14, the time of implementation of obtaining each task node deposits in the data structure of task node, obtains the weight on limit and depositsEnter in array.
In another kind of preferred embodiment of the present invention, described step S2 concrete steps are:
On new directed acyclic graph basis, increase by two nodes, source node and aggregation node, and establish these two nodesTime of implementation is 0; For all task nodes that there is no father node, increase a limit from source node to this node; For instituteWhether there is the task node of child node, increase a limit from this node to aggregation node. To and converge joint with source node simultaneouslyThe weight of putting connected limit is made as 0.
In a kind of preferred embodiment of the present invention, described step S3 specifically comprises the steps:
S31, be all tasks the completing of this task while not making a mistake the deadline the earliest that initializes each taskTime; Be 0 the deadline the latest that initializes each task;
S32, using a task node I as input, this task can be any one node; If this task node IDeadline be the latest greater than 0, illustrate that this task node was calculated, directly return to its deadline and key thereof the latestTask; If the father node set-inclusion source node of this task node, the mission critical of this task is itself, by allowing thisTask occur the wrong deadline the latest that obtains this task; Otherwise enter next step;
S33, the traversal of recurrence solves deadline the latest and the corresponding key of all father nodes of this task node ITask;
S34, supposes that task I has m father node, and is respectively F (I) their deadline the latest1,F(I)2,…,F(I)m. Respectively at F (I)1,F(I)2,…,F(I)mBasis on deadline of calculation task I, obtain I1,I2,…,Im. SimultaneouslyCalculating other tasks does not make a mistake and only has task I that the deadline I of all X soft error occurs0And and I1,I2,…,ImCompare, choose the latest deadline of maximum as task I, determine that the mission critical of this task node I is simultaneouslyThe mission critical of trying to achieve in itself or S33.
The invention has the beneficial effects as follows:
Whether meet in current multinuclear real-time system the optimum deadline that the present invention can weigh set of tasks fastWork requirements.
Be accurately the deadline the latest that the present invention obtains. If the deadline of default is less than this time,Have be greater than that 0 probability causes within deadline can not fault-tolerant all X mistake; If the deadline of default is than thisThe individual time is large, can cause temporal waste.
The present invention can determine at O (n^2) set of tasks deadline the latest accurately in the time, than prior artFast efficient.
Additional aspect of the present invention and advantage in the following description part provide, and part will become from the following descriptionObtain obviously, or recognize by practice of the present invention.
Brief description of the drawings
Above-mentioned and/or additional aspect of the present invention and advantage from conjunction with below accompanying drawing to becoming the description of embodimentObviously and easily understand, wherein:
Fig. 1 is task scheduling figure of the prior art, wherein, and appointing when Fig. 1 (a) does not make a mistake for all tasksBusiness scheduling graph; There is respectively the fault-tolerant task scheduling when wrong twice in the task on the each core of Fig. 1 (b) with maximum execution timeFigure, does not have data dependence relation between task; Fig. 1 (c) is twice of T4 (having the longest time of implementation) generation in set of tasksFault-tolerant task scheduling figure when mistake; Fig. 1 (d) is fault-tolerant task scheduling figure when twice mistake occurs T2 in set of tasks;
Fig. 2 is task scheduling figure and the directed acyclic thereof to step S1 and S2 explanation in a kind of preferred embodiment of the present inventionFigure, wherein, Fig. 2 (a) represents the directed acyclic graph of original data dependence in set of tasks; Fig. 2 (b) is task scheduling figure; Fig. 2(c) for the present invention is oriented after increasing dispatching sequence and relying on of the adjacent task of scheduling on same core on the basis of Fig. 2 (a)Acyclic figure; Fig. 2 (d) increases the directed acyclic after source node Tsr and aggregation node Tsk for the present invention on the basis of Fig. 2 (c)Figure;
Fig. 3 is the directed acyclic graph of in the another kind of preferred embodiment of the present invention, step S3 being explained.
Detailed description of the invention
Describe embodiments of the invention below in detail, the example of described embodiment is shown in the drawings, wherein from start to finishSame or similar label represents same or similar element or has the element of identical or similar functions. Attached below by referenceThe embodiment that figure describes is exemplary, only for explaining the present invention, and can not be interpreted as limitation of the present invention.
The invention provides in a kind of multinuclear real-time fault tolerance system and obtain accurately the method for deadline the latest, it comprisesFollowing steps:
The first step: represent the data dependence between task with directed acyclic graph, as shown in Fig. 2 (a);
In Fig. 2 (b), express task scheduling figure, the time of implementation of task T1, T2, T3, T4, T5, T6 is respectively 4,10,3,4,5,3 chronomeres.
According to the task scheduling of the multiple nucleus system shown in Fig. 2 (b), on the basis that keeps legacy data to rely on, be schedulingAdjacent task on same core increases dispatching sequence and relies on, and sets up new directed acyclic graph, as shown in dotted line in Fig. 2 (c), increasesThe dispatching sequence who adds between the upper task T2 of core P2 and task T5 relies on, and the scheduling between the upper task T3 of increase core P3 and task T6 is suitableOrder relies on, and sets up new directed acyclic graph. In the present embodiment, specifically can adopt following steps:
A, sorts and forms Task-list task according to the Starting Executing Time of each task, carries out from chained list headScanning, carrying out of the task that is scheduled at first from this core scans, and establishes index and is designated as i, makes i=1;
B, makes j=i+1, if j > N stops, if otherwise task Ti and task Tj be assigned on same core, formerHave on the basis of data dependence and increase a limit that represents dispatching sequence from node Ti to node Tj, make i=i+1, return to stepB。
Finally, obtain the time of implementation C of each task node TiTiDeposit in the data structure of task node, obtain limitWeight deposits in array.
Second step increases by two nodes on new directed acyclic graph basis, and source node Tsr and aggregation node Tsk, as figureShown in 2 (d), and the time of implementation of establishing these two nodes be 0; For all task nodes that there is no father node, increase by one fromSource node, to the limit of this node, in present embodiment, increases the limit from source node Tsr to task node T1, T2 and T3; ForAll task nodes that there is no child node, increase a limit from this node to aggregation node, in the present embodiment, increaseLimit from task node T5 to aggregation node Tsk. The weight of simultaneously setting all limits of this step increase is 0.
The 3rd step, determines the mission critical of set of tasks, obtains optimum deadline. Specifically comprise the steps:
S31, should when BCFT deadline the earliest (Ti) that initializes each task Ti does not make a mistake for all tasksThe deadline of task; WCFT deadline the latest (Ti) that initializes each task Ti is 0; Initialize the pass of each task TiKey task is CT (Ti)=Tnull; Initialize possible maximum soft error number X. In the present embodiment, i=1,2 ..., N.
S32, using a task node Ti as input, this task can be any one node; If this task nodeBe greater than 0 the deadline the latest of Ti, illustrates that this task node was calculated, and directly returns to its deadline and pass thereof the latestKey task; If the father node set-inclusion source node of this task node, the mission critical of this task is itself, by allowingThis task occur the wrong deadline the latest that can obtain this task; Otherwise enter next step S33;
S33, the traversal of recurrence solves deadline the latest and the corresponding pass of all father nodes of this task node TiKey task;
S34, supposes that task Ti has m father node, and is respectively F (Ti) their optimum deadline1,F(Ti)2,…,F(Ti)m. Respectively at F (Ti)1,F(Ti)2,…,F(Ti)mBasis on deadline of calculation task Ti, obtainI1,I2,…,Im. There are all X wrong deadline I in calculation task Ti simultaneously0And and I1,I2,…,ImCompare choosingGet the latest deadline of maximum as task Ti, the mission critical of simultaneously determining this task node Ti be itself orThe mission critical of trying to achieve in S33.
In the present embodiment, as shown in Figure 3, be that all tasks are not sent out the deadline the earliest that initializes each taskThe deadline of this task when raw mistake, i.e. BCFT (T1)＝CT1、BCFT(T2)＝CT2、BCFT(T3)＝max{CT1+CT3+W(T1+T3),CT2+CT3+W(T2+T3), wherein CTiFor the time of implementation of task Ti; Be the deadline the latest that initializes each task0, i.e. WCFT (T1)＝WCFT(T2)＝WCFT(T3)=0; Initialize the mission critical CT (T of each task1)＝CT(T2)＝CT(T3)＝Tnull; Initialize possible maximum soft error number X.
Make task node Ti as input, this task can be any one node, in the present embodiment, preferably makes i=1; If be greater than 0 the deadline the latest of this task node Ti, illustrate that this task node was calculated, and directly returned to itDeadline and mission critical thereof the latest; If the father node set-inclusion source node of this task node, the key of this taskTask is itself, for example, for task node T1And T2, by allow this task occur wrongly can obtain this taskDeadline in evening,
WCFT(T1)＝CT1+X*CT1CT(T1)＝T1
WCFT(T2)＝CT2+X*CT2CT(T2)＝T2
If the father node set of this task node does not comprise source node, enter next step S33;
S33, incoming task node Ti, recurrence traversal solves the optimum deadline of all father nodes of this task node,Get and in father node, there is the maximum father node of deadline the latest and corresponding mission critical, and record this father node to this taskThe weight on the limit of node. For example,, for task node T3, recurrence traversal solves the father node T of this task node1And T2The latestDeadline, add respectively task node T1And T2To T3The weights W (T on limit1,T3)、W(T2,T3) and T3Time of implementationCT3, be respectively:
WCFT(T1)+W(T1,T3)+CT3
WCFT(T2)+W(T2,T3)+CT3
In the present embodiment, weight table is shown the propagation delay time of data dependence, for example W (T1,T3) refer to task T1Carry out knotRestraint to task T3Start the time delay of carrying out.
Allow this task T3Occur the deadline BCFT (T that arrives of wrong gained3)+X*CT3With the WCFT obtaining above(T1)+W(T1,T3)+CT3、WCFT(T2)+W(T2,T3)+CT3Compare, that maximum is task T3Deadline the latest, simultaneouslyCan determine the mission critical of this task. For example,, if max{WCFT is (T1)+W(T1,T3)+CT3,WCFT(T2)+W(T2,T3)+CT3,BCFT(T3)+X*CT3}＝WCFT(T1)+W(T1,T3), task T3Deadline be the latest WCFT (T1)+W(T1,T3)+CT3, task T3Mission critical be CT (T3)＝T1。
Determining after deadline the latest of set of tasks, the present invention also comprises the steps: to weigh set of tasksWhether meet the work requirements of multinuclear real-time fault tolerance system evening deadline, if met, exit, if do not met, adjustThe whole scheduling strategy to this set of tasks, returns to the first step. The adjustment mode of specific tasks scheduling can be definite according to actual conditions,Not key protection point of the present invention, do not do too much and repeat herein.
It should be noted that, in the present embodiment, for different multiple nucleus systems, the maximum soft error number that can occurAmount X can be different, and concrete numerical value can obtain according to experiment measuring.
In the description of this description, reference term " embodiment ", " some embodiment ", " example ", " specifically showExample " or the description of " some examples " etc. mean the specific features, structure, material or the spy that describe in conjunction with this embodiment or examplePoint is contained at least one embodiment of the present invention or example. In this manual, to the schematic statement of above-mentioned term notNecessarily refer to identical embodiment or example. And specific features, structure, material or the feature of description can be anyOne or more embodiment or example in suitable mode combination.
Although illustrated and described embodiments of the invention, those having ordinary skill in the art will appreciate that: notDepart from the situation of principle of the present invention and aim and can carry out multiple variation, amendment, replacement and modification to these embodiment, thisScope of invention is limited by claim and equivalent thereof.
Claims (3)
1. in multinuclear real-time fault tolerance system, obtain accurately the method for deadline the latest, it is characterized in that, comprise as followsStep:
S1, according to the task scheduling of multiple nucleus system, on the basis that keeps legacy data to rely on, is the phase of carrying out on same coreAdjacent task increases dispatching sequence and relies on, and sets up new directed acyclic graph;
S2, on the basis of new directed acyclic graph, increases by two virtual task nodes, and the time of implementation of described virtual task is0, in the task scheduling taking this directed acyclic graph as model, one of them virtual task node is carried out at first, another one voidIntending task node finally carries out;
S3, supposes in task implementation, to occur at most X soft error, on the basis of original scheduling, exists by the task of makeing mistakesRe-executing immediately on same core realized fault-tolerant, ensures the correct execution of set of tasks, and the key of definite set of tasksTask, deadline the latest accurately of obtaining set of tasks, specifically comprises the steps:
S31, be the deadline of all tasks this task while not making a mistake the deadline the earliest that initializes each task;Be 0 the deadline the latest that initializes each task;
S32, using a task node I as input, this task can be any one node; If this task node IBe greater than 0 deadline in evening, illustrates that this task node was calculated, and directly returns to its deadline and key times thereof the latestBusiness; If the father node set-inclusion source node of this task node, the mission critical of this task is itself, by allowing thisBusiness occur the wrong deadline the latest that obtains this task; Otherwise enter next step;
The traversal of S33, recurrence solves deadline the latest and the corresponding key of all father nodes of this task node I and appointsBusiness;
S34, suppose that task I has m father node, and be respectively F (I) their deadline the latest1,F(I)2,…,F(I)m，Respectively at F (I)1,F(I)2,…,F(I)mBasis on deadline of calculation task I, be respectively I1,I2,…,Im, meter simultaneouslyCalculating other tasks does not make a mistake and only has task I that the deadline I of all X soft error occurs0, by I0With I1,I2,…,ImCompare, choose the maximum deadline the latest as task I, determine that the mission critical of this task node I is simultaneouslyThe mission critical of trying to achieve in itself or S33;
S4, weigh set of tasks whether meet the latest the work in current multinuclear real-time fault tolerance system deadline accurately timeLimit demand, if met, exits, if do not met, adjusts scheduling strategy, returns to step S1.
2. in multinuclear real-time fault tolerance system as claimed in claim 1, obtain accurately the method for deadline the latest, its featureBe, described step S1 specifically comprises the steps:
S11, obtains the number N of task in set of tasks and contingent maximum soft error in set of tasks implementationNumber X;
S12, represents the data dependence between task with directed acyclic graph;
S13, according to given task scheduling, if two tasks on same core, be scheduled carry out and dispatching sequence adjacent,The dispatching sequence who increases between these two tasks relies on, and on the basis of original directed acyclic graph, increases expression dispatching sequenceThe limit relying on, obtains new directed acyclic graph;
S14, the time of implementation of obtaining each task node deposits in the data structure of task node, and the weight of obtaining limit deposits number inIn group.
3. in multinuclear real-time fault tolerance system as claimed in claim 1, obtain accurately the method for deadline the latest, its featureBe, described step S2 concrete steps are:
On new directed acyclic graph basis, increase by two nodes, source node and aggregation node, and the execution of establishing these two nodesTime is 0; For all task nodes that there is no father node, increase a limit from source node to this node; Do not have for allThere is the task node of child node, increase a limit from this node to aggregation node, simultaneously by all and source node with converge jointThe weight of putting connected limit is made as 0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310739053.0A CN103699455B (en) | 2013-12-27 | 2013-12-27 | In a kind of multinuclear real-time fault tolerance system, obtain accurately the method for deadline the latest |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310739053.0A CN103699455B (en) | 2013-12-27 | 2013-12-27 | In a kind of multinuclear real-time fault tolerance system, obtain accurately the method for deadline the latest |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103699455A CN103699455A (en) | 2014-04-02 |
| CN103699455B true CN103699455B (en) | 2016-05-25 |
Family
ID=50360994
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310739053.0A Expired - Fee Related CN103699455B (en) | 2013-12-27 | 2013-12-27 | In a kind of multinuclear real-time fault tolerance system, obtain accurately the method for deadline the latest |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103699455B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105138401B (en) * | 2015-08-07 | 2019-08-30 | 黑龙江大学 | Real-time task schedulability based on linear linked list tests direct analogy method |
| CN105224400B (en) * | 2015-08-07 | 2019-03-22 | 黑龙江大学 | Using the method for linear linked list record real-time task scheduling process |
| US10523760B2 (en) | 2015-10-16 | 2019-12-31 | Tttech Computertechnik Ag | Method and computer system for quickly transmitting time-controlled real-time messages |
| CN105260188A (en) * | 2015-10-24 | 2016-01-20 | 北京航空航天大学 | Time characteristic model and modeling method thereof |
| CN110019986B (en) * | 2018-09-27 | 2020-12-29 | 福建福链科技有限公司 | Method and terminal for sequencing messages in directed acyclic graph |
| CN110659480B (en) * | 2019-09-24 | 2021-11-05 | 重庆邮电大学 | Trojan circuit detection and error recovery method during equipment operation |
| CN112416591B (en) * | 2020-11-25 | 2023-10-13 | 广州虎牙科技有限公司 | Distributed task processing method, device, equipment, storage medium and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102170396A (en) * | 2011-05-06 | 2011-08-31 | 浙江大学 | QoS control method of cloud storage system based on differentiated service |
| CN102207883A (en) * | 2011-06-01 | 2011-10-05 | 华中科技大学 | Transaction scheduling method of heterogeneous distributed real-time system |
| CN103412794A (en) * | 2013-08-08 | 2013-11-27 | 南京邮电大学 | Dynamic dispatching distribution method for stream computing |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110246823A1 (en) * | 2010-04-05 | 2011-10-06 | Et International, Inc. | Task-oriented node-centric checkpointing (toncc) |
-
2013
- 2013-12-27 CN CN201310739053.0A patent/CN103699455B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102170396A (en) * | 2011-05-06 | 2011-08-31 | 浙江大学 | QoS control method of cloud storage system based on differentiated service |
| CN102207883A (en) * | 2011-06-01 | 2011-10-05 | 华中科技大学 | Transaction scheduling method of heterogeneous distributed real-time system |
| CN103412794A (en) * | 2013-08-08 | 2013-11-27 | 南京邮电大学 | Dynamic dispatching distribution method for stream computing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103699455A (en) | 2014-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103699455B (en) | In a kind of multinuclear real-time fault tolerance system, obtain accurately the method for deadline the latest | |
| Ciraci et al. | FNCS: A framework for power system and communication networks co-simulation | |
| Yu et al. | Distributed adaptive control of synchronization in complex networks | |
| CN110348571A (en) | A kind of neural network model training method, device, chip and system | |
| CN107562066B (en) | Multi-target heuristic sequencing task planning method for spacecraft | |
| WO2019104713A1 (en) | Machine learning method, master node, work node, and system | |
| Chan et al. | Mobiliti: Scalable transportation simulation using high-performance parallel computing | |
| JP6523497B1 (en) | Master controller and synchronous communication system using the same | |
| CN102664700B (en) | Nodal clock synchronous planning method in mesh network | |
| CN106354566B (en) | A kind of method and server of command process | |
| CN113033800A (en) | Distributed deep learning method and device, parameter server and main working node | |
| CN107239468A (en) | Task node management method and device | |
| CN102420797B (en) | Topology mapping method and system | |
| WO2021159929A1 (en) | Topology diagram conversion system and method | |
| Ke et al. | Aggregation on the fly: Reducing traffic for big data in the cloud | |
| CN107103360B (en) | Reliable application distribution distributed genetic method based on mixed cloud | |
| WO2017106293A3 (en) | Dynamic design of complex system-of-systems for planning and adaptation to unplanned scenarios | |
| CN105808340B (en) | Load-balancing method and system | |
| CN109451557A (en) | A kind of wireless sensor network dynamic clustering method for routing and device | |
| US9014202B2 (en) | Least disruptive AF assignments in TRILL LAN adjacencies | |
| CN105634963A (en) | Conformance route generation and implementation method and device | |
| Adacher et al. | A distributed approach for traffic signal synchronization problem | |
| CN103236894B (en) | Clock synchronizing method and device in a kind of seamless redundant network | |
| CN109002908A (en) | One kind is arranged an order according to class and grade dispatching method, system and electronic equipment | |
| CN114039936A (en) | Virtual power plant communication service flow scheduling method and device based on time sensitivity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160525 Termination date: 20171227 |