CN102624543A - Distribution simulation supporting environment implementation method based on multilayer distributed service - Google Patents
Distribution simulation supporting environment implementation method based on multilayer distributed service Download PDFInfo
- Publication number
- CN102624543A CN102624543A CN2012100144437A CN201210014443A CN102624543A CN 102624543 A CN102624543 A CN 102624543A CN 2012100144437 A CN2012100144437 A CN 2012100144437A CN 201210014443 A CN201210014443 A CN 201210014443A CN 102624543 A CN102624543 A CN 102624543A
- Authority
- CN
- China
- Prior art keywords
- rsc
- service
- service groups
- groups
- lsc
- 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
Images
Landscapes
- Computer And Data Communications (AREA)
Abstract
Disclosed is a distribution simulation supporting environment implementation method based on multilayer distributed service. A distribution simulation supporting environment adopts a center-layer service group as a core and outwards extends via the center-layer service group with multilayer tree-type relation, a fully-connected peer-to-peer network is arranged in each service group, and network efficiency is optimized and fault tolerance is improved in each group by a network load balance algorithm for processing duration on the basis of information. A bottom layer of the supporting environment consists of simulation members, and a central layer and various intermediate layers can also be directly connected with the simulation members. Each simulation member is created on the basis of a local server and consists of reusable modules. Only request information with an action scope and a period attribute needs to be proposed, is in fusion processing via the supporting environment and is issued to the various simulation members, and optimal request management and data distribution management are realized. By the aid of the method, the fault tolerance in large-scale distributed simulation and network balancing capacity are strengthened, the bottleneck affecting the efficiency of a system is eliminated, and reusability of the simulation members is enhanced.
Description
Technical field
The present invention relates to the implementation method of the distributed simulation back-up environment of the support large scale emulation on local area network (LAN) and the wide area network, comprise the implementation method of local the integration environment with the multi-level Distributed Services environment of emulation terminal.
Background technology
High Level Architecture HLA (High Level Architecture) provides a general technological frame and open standard for the modeling of distributed emulation with practice.The concrete realization of HLA is run time infrastructure RTI (RunTime Infrastructure); RTI provides one group of generic service, and these services comprise federation management, statement management, Object Management group, ownership management, time management, six aspects of data distribution management.RTI mainly contains following several types at present:
1, centralized.The eRTI that main representative is a MIT etc.
2, function is distributed.Main representative is the KDRTI of University of Science and Technology for National Defence, the pRTI of Sweden etc.
3, full distributed.FDK such as Georgia university.
4, hierarchy type.Such as the StarRTI of University of Science and Technology for National Defence, mainly there has been following problem in the present RTI of the BHRTI of BJ University of Aeronautics & Astronautics:
One, fault-tolerance need strengthen.
1, except full distributed RTI, there is a center RTI in present various RTI, is responsible for overall relevant issues and handles, if this center RTI fault will cause whole federal fault.The problem of full distributed RTI is that after simulation scale was increased to a certain degree, system complexity was excessive, and efficient and reliability are difficult to guarantee.
2, for hierarchy type RTI; If certain local RTI fault; One group membership's emulation that also will cause this RTI to be responsible for is interrupted even failure, although some fault-tolerant processing is also recovered through heartbeat mechanism detection failure RTI such as BHRTI; But during restoration, affected this group membership can't continue to participate in emulation.
Two, the NLB ability need strengthen.
Present all have the RTI of multiserver; Realize the NLB between server; Can only pass through the static configuration means, before emulation is initial, distribute each RTI server subordinate's member rationally, and can't when simulation run, dynamically realize the NLB between the RTI server.
Three, only till the member, other granularity of software repeated usage level is bigger than normal for the data distribution management level.
SOM has stipulated the demand and the publish information of member's level in the HLA standard, and till RTI was responsible for demand information only is distributed to the member, the member received after the information by oneself.The SOM of different members is different, in different Simulation Application, is difficult to run into the member of identical functional requirement, has caused the multiplexing difficulty of member bigger like this.Even improve reusability based on the BOM standard in the emulation, also multiplexing on simulation object model basis just, this multiplexing implication with software module is different.For software design, improve the software repeated usage ability, must reduce the granularity of software repeated usage module, granularity is more little, and durability is good more.
Four, the statement management should be stated ordering content, states published content again, and is more loaded down with trivial details in the use.
Five, present; The distributing emulation system that also has a large amount of non-HLA modes to develop; Its interactive information often is not to be divided into object class and interactive class, neither the object class property value be set one by one according to the interface method of RTI during transfer of data, but directly send the one piece of data bag.Old system like this is when carrying out adaptability reform, and all the elements except the inter-process flow process all will design and develop again, makes the workload of the old system reform bigger.
Six, the interface method of HLA standard code is careful concrete, and wherein the IEEE1516 version reaches 146, and this just possibly grasp after also having caused the training that utilizes RTI developer must accept system.But in general distributing emulation system exploitation; The developer often there is no need to grasp the total interface method and realizes comprehensive control; The function that actual needs initiatively calls concentrates on federal create, add and withdraw from federation, time propelling, statement of need, object and interaction data issue etc., and the callback method that needs concentrates on time synchronized, object and interaction data renewal and federation management incident etc.
The inventor is to above problem; A kind of distributed simulation back-up environment based on multi-level Distributed Services has been proposed; Provided one see from use angle simpler, but can support the implementation method of the back-up environment of the large-scale distributed emulation from the local area network (LAN) to the wide area network.The present invention possesses the six functions that HLA proposes, but is not devoted to compatible mutually with the HLA interface standard.
Summary of the invention
The objective of the invention is to improve fault-tolerant ability, the NLB ability of large-scale distributed analogue system, reduce simulated members software repeated usage granularity and simplify developer's work.
The technical scheme that the present invention adopts is: a kind of implementation method of the distributed simulation back-up environment based on multi-level Distributed Services; The organization and management method that has comprised local service assembly LSC (Local Service Component) and remote service assembly RSC (Remote Service Component); Simulated members provides service by local service assembly LSC; Each simulated members is made up of a plurality of reusable modules; Local service assembly LSC generates the agency for each module, and safeguards a local data environment of distinguishing by name; It is characterized in that: one group of simulated members under a plurality of local service assembly LSC support provides service by one group of remote service assembly RSC; The service groups of being made up of remote service assembly RSC is divided into intermediate layer RSC service groups and central core RSC service groups according to the status difference, and they are made up of the full remote service assembly RSC that connects of plurality of network respectively; A system has only a central core RSC service groups; Central core RSC service groups is managed some intermediate layers RSC service groups by tree; Can also extend the intermediate layer RSC service groups that makes new advances below each intermediate layer RSC service groups; The bottom be local service assembly LSC, simultaneously, central core RSC service groups and any one intermediate layer RSC service groups also can directly derive local service assembly LSC; Same simulated members disparate modules information interaction is independently accomplished by local service assembly LSC; Carry out information interaction between the different simulated members; Coordinate to accomplish by local service assembly LSC on the shortest network path between them and remote service assembly RSC, relate to global transaction and then participate in coordinating to accomplish through central core RSC service groups; Like this; Multi-level distributed remote serviced component RSC service groups is managed a plurality of simulated members groups; Form a distributed simulation system; And, comprise system management, time management, demand management, data distribution management, Object Management group and ownership management for it provides required service function; The realization of these service functions is based on the interoperability agreement of the inner definition of back-up environment, and the basis of this agreement is self-defining message.
Said back-up environment is to be core with central core RSC service groups, and through the service system that a plurality of intermediate layers RSC service groups is outwards constantly extended, each service groups inside is the full peer-to-peer network relation that connects, and is that tree concerns between group and the group; In any moment, as long as each service groups also has a remote service assembly RSC operate as normal at least, entire environment just can normally be moved.
Deserve to be called grade service groups and subordinate's service groups mutually between central core RSC service groups, intermediate layer RSC service groups or the local service assembly LSC of the adjacent two-stage of symbolic animal of the birth year in the tree of said back-up environment.Each group member in subordinate's service groups selects the higher level group member who connects to decide according to the Message Processing duration in higher level's service groups; Measure the time difference between the superior and the subordinate through periodicity time adjustment message between the interconnected group member between at first upper and lower level service groups; The higher level group member comes the detect-message handling duration through the message time stamp and the data circulation that statistics obtains is organized entirely then; When group member's load differs big; The high capacity group member sends migration message to its certain group member of subordinate, makes the group member of this subordinate be connected to the lighter group member of other loads in higher level's service groups again; This method is used to dispose the group member's failure condition that occurs in the service groups equally.
Said simulated members only proposes demand information and gets final product, and carries out demand information by back-up environment by level and merges, and finally is aggregated into central core RSC service groups, and then is issued to each simulated members after merging by level respectively; Demand information has action scope and cycle two kinds of attributes, can further optimize data distribution management and demand management, improves system-wide efficient.
The present invention compared with prior art has following advantage: the implementation method of this distributed simulation back-up environment based on multi-level Distributed Services is to be core with the central core service groups; Intermediate layer service groups through multi-level stretches out; Constitute tree type relation between group and the group; Each service groups inside all is to connect peer-to-peer network entirely, adopts the NLB algorithm based on the Message Processing duration in the group, is used to optimize network efficiency and improves Fault Tolerance.The bottom of back-up environment is the local service end that is provided support by the simulated members that reusable module constitutes, and central core also can directly be connected simulated members with any intermediate layer.Simulated members only proposes demand information, and demand information has action scope and cyclic attributes, through being issued to each simulated members after the back-up environment fusion treatment again, realizes demand management and the data distribution management more optimized.This method has strengthened fault-tolerant ability and the NLB ability in the large-scale distributed emulation; Improved the network interaction performance; Having eliminated influences the system effectiveness bottleneck, has improved the reusability and the flexibility of simulated members, has further optimized demand management and data distribution management pattern.
Description of drawings
Fig. 1 is local service assembly LSC and simulated members and remote service assembly RSC graph of a relation;
Fig. 2 is that local service assembly LSC realizes schematic diagram;
Fig. 3 is remote service assembly RSC and other RSC and LSC annexation figure;
Fig. 4 is the full annexation figure of RSC service groups;
Fig. 5 is multicast relations figure in the RSC service groups;
Fig. 6 is remote service assembly RSC and the superior and the subordinate's Communication Realization schematic diagram;
Fig. 7 is remote service assembly RSC and RSC Communication Realization schematic diagram on the same group;
Fig. 8 is the realization schematic diagram of multi-level Distributed Services;
Clock correction detected schematic diagram when Fig. 9 was;
Figure 10 be RSC fault or load when overweight network connect the migration sketch map;
Figure 11 is that the demand information fusion reports and assign the path sketch map.
Embodiment
Following the present invention being done with reference to accompanying drawing further describes.
Realization based on the distributed simulation back-up environment of multi-level Distributed Services is to rely on a plurality of local service assembly LSC and a plurality of remote service assembly RSC acting in conjunction, in network, forms a unified data environment.All simulated members position each other in this environment is transparent.The most basic communication unit is a message in the environment, on various message basis, realizes federation management, demand management, time management, data distribution management, Object Management group and the ownership management of distributed simulation back-up environment.
Fig. 1 has described local service assembly LSC and simulated members and remote service assembly RSC relation: the basic software elements of simulated members is a reusable module, for example the source code file of C++ class, dynamic link libraries or executable program etc.Each simulated members provides service by a local service assembly LSC through the plurality of modules agency.A local service assembly LSC is connected with a remote service assembly RSC through network, can move a plurality of local service assembly LSC in the computer simultaneously, and Different L SC distinguishes by name.
Fig. 2 has described local service assembly LSC and realized principle: the reusable module in the simulated members is at first filed an application to local service assembly LSC; Local service assembly LSC is that each module generates an agency separately then; The statement of need of module and message issue request under this module agency accepts, and the interface that callback module provides when the required Data Update of module.Local service assembly LSC merges all module agencies' conditions of demand, and the demand that generates local service assembly LSC is afterwards through reporting remote service assembly RSC to last connection agency.Local service assembly LSC safeguards that through unique last connection is acted on behalf of a pair of unique message of distinguishing by the LSC title of the local overall situation goes into formation and dequeue, and the module agency safeguards a message dequeue and a message for module under it and goes into formation.When remote service assembly RSC sent new information, local service assembly LSC fell in lines message earlier, woke all module agencies' observation thread then up.Module agency's observation thread is gone into the formation message of fetching data according to self-demand and joining self from the message of LSC and is gone in the formation, wakes its message threads again up.Message threads goes into the formation by the principle of FIFO to take out message from message, and the interface of module under the readjustment.When module has message to send; Its agency is with the message dequeue that mails to local service assembly LSC behind the message queueing; After local service assembly LSC presses FIFO principle taking-up message; On the one hand according to higher level's demand with data filter after through last connection agency is issued remote service assembly RSC, add message according to the LSC demand after with data filter on the one hand and go into formation.
Fig. 3 has described the annexation of remote service assembly RSC with other RSC and LSC: RSC a moment can only on be connected a RSC; Can connect a plurality of RSC on the same group; Also can accept being connected of a plurality of LSC and RSC; With respect to this RSC, other can be described as higher level RSC, RSC and the RSC/LSC of subordinate on the same group respectively.
Fig. 4 has described the interior fully-connected network relation of RSC service groups: a RSC service groups is formed by a plurality of RSC are individual, connects respectively between the RSC in the group is mutual, forms a full peer-to-peer network that connects.If initiate RSC is arranged, it must be grasped on the same group information and initiatively connect other RSC, other RSC Hui Lian more then, and after having set up stable full connection and initialization related data, the RSC service groups has just successfully been admitted the newcomer.Certain RSC initiatively withdraws from or breaks down in group, and other RSC can find after a sense cycle at the most that each RSC upgrades the connection of this group information and cleaning correspondence and receives the agency then.The difference of intermediate layer RSC service groups and central core RSC service groups is: central core RSC service groups does not have the upper management layer.
Fig. 5 has described multicast relations in the RSC service groups: when RSC service groups place network support network multicast; RSC in the RSC service groups adds same multicast group; Through multicast issue self networks address, higher level RSC information and load information; Initiate RSC at first adds this multicast group, receives on the same group initiatively to connect other RSC after the information, then other RSC Hui Lian again.
Fig. 6, Fig. 7 have described the implementation method of remote service assembly RSC: each RSC in the RSC service groups has same 26S Proteasome Structure and Function.See from the angle that network connects; Remote service assembly RSC can accept being connected of a plurality of local service assembly LSC and a plurality of remote service assembly RSC of subordinate; Also can connect higher level's remote service assembly RSC, can also interconnect with remote service assembly RSC on the same group simultaneously as the client.Remote service assembly RSC communicates by letter with a higher level RSC through the connection agency that makes progress, and message of maintenance was gone into formation and message dequeue during upwards connection was acted on behalf of, and preserved higher level's demand that the higher level is transmitted; Realize interconnection through many to connecting the agency and receiving each the remote service assembly RSC that acts on behalf of with on the same group; Has only the message dequeue among these connections agency; Be responsible for receiving the demand on the same group that the other side is transmitted of preserving simultaneously, have message among the reception agency and go into formation, simultaneously subordinate's fusion demand is sent; Handle the relevant issues of each subordinate through the agency of subordinate's reception one to one; These receive the agency and go into formation and a message dequeue for corresponding subordinate safeguards a message; And the demand of memory subordinate is reported to higher level and RSC on the same group after the demand of all subordinates is merged.When last connection agency being received the message of higher level RSC and connects that the reception agency of RSC receives message on the same group on the same group; They go into formation with the message that message adds separately at first respectively; The continuous Pop-up message of formation also adds corresponding message according to the conditions of demand that each subordinate receives the agency and goes into formation, and then mails to subordinate.After the reception agency of subordinate receives the message of uploading; On the one hand based on the higher level's demand among the last connection agency is added message to last connection agency and the message dequeue that is connected the agency on the same group with the demand on the same group that is connected the agency on the same group, one side joins the message that subordinate's reception of correspondence acts on behalf of based on the demand of other subordinates with message and goes in the formation.
Fig. 8 has described the network topology structure of multi-level Distributed Services: multi-level Distributed Services is meant that with central core RSC service groups be core; According to the stretch out intermediate layer RSC service groups of many levels of system scale demand; Whole network topology structure is tree; The end of tree is exactly the simulated members that is provided support by local service assembly LSC, and central core RSC service groups can both directly be connected LSC with any intermediate layer RSC service groups.
Each RSC node periodic statistics Message Processing duration in the RSC service groups is as loading index; If certain RSC node load is overweight; The Message Processing duration will increase so; Be equivalent to loading index and increase, its downstream site will select the lighter remote service assembly RSC of other loads in the same service groups as the higher level voluntarily; In addition, when remote service assembly RSC node failure, its subordinate's node also can be selected new remote service assembly RSC higher level voluntarily, has guaranteed the real-time and the reliability of whole simulation system.This multi-level Distributed Services has been guaranteed at any time; As long as each intermediate layer service groups or center service group all also have a remote service assembly RSC service end ability operate as normal at least; Just can guarantee the normal operation of whole system, so just promote system survivability and NLB ability.
Fig. 9 has described based on the time clock correction in the adaptive network load balancing algorithm of Message Processing duration and has detected principle.The final purpose that detects is to detect high relatively year and the low RSC that carries in the RSC service groups, does not need testing result very accurate.Basically satisfying under the accuracy requirement prerequisite, the detection of clock correction when the present invention realizes through following method under not having extraneous supplementary means situation: the time clock correction of establishing between the superior and the subordinate is fixed as δ, when each time difference sense cycle arrives, at T
1RSC of subordinate or LSC send clock correction message to higher level RSC constantly, and higher level RSC is through Network Transmission λ
1After time at T
2Constantly receive, and behind the processing time Δ with T
2Time information is included in the internal feedback of clock correction message and goes back to subordinate, and subordinate is through Network Transmission λ
2After time at T
3Constantly receive.According to accuracy requirement, can think λ
1=λ
2And Δ=0, so obtain detected value δ=T
2-(T
1+ T
3)/2.Because the time clock correction between two connected nodes of the superior and the subordinate is a constant steady state value in theory, incite somebody to action the measured value δ of n sense cycle after this
1δ
nCarry out recursive least-squares filtering, wherein get initial value
Can obtain the filter value of the n time detection
Along with time lengthening, accuracy of detection will improve constantly.
Figure 10 described certain RSC fault or load in the RSC service groups when overweight network connect transition process: realize that NLB and fault-tolerant processing are the adaptive network load balancing algorithm of foundation based on the Message Processing duration.All RSC service groups are all dynamically adjusted network through this algorithm and are connected, thereby realize the dynamic equilibrium of whole system.Concrete grammar is following:
1, detect-message handling duration.
(1) the time clock correction between detection the superior and the subordinate;
(2) each message all has time stab information, before message is about to send, the time stamp in the message is revised the time clock correction with the other side on the local clock basis, and it is the clock value of benchmark that time stamp is adapted to the message receiver clock;
(3) after the recipient receives and handles message, take out time stamp and local clock comparison in the message, just can obtain the handling duration of this message.The Message Processing duration comprises online transmission time and recipient's processing time.
2, duration data processing.The recipient handles the handling duration that can both obtain this message after every message; Long data carries out filtering and obtains relative smoothed data when at first adopting Kalman filtering algorithm to each; Then when a NLB sense cycle finishes; The mathematic expectaion of all Kalman filtering values in the computing cycle, and with this as final loading index.
3, NLB and fault-tolerant strategy.RSC in the RSC service groups regular load data of circular self in group, and the address information of all RSC is notified all downstream sites in when serving the group member and change, will organizing, and carries out NLB adjustment according to following step then:
(1) will organize interior all remote service assembly RSC and press the load value ordering;
(2) calculate the average of all remote service assembly RSC loads, each node is divided into high, medium and low three groups by its load and average departure degree;
(3) whether surpass detection threshold according to the departure degree of high capacity value size and high capacity and minimum load and determine whether to activate load balance adjustment, like uncomfortable turn over step (10); Otherwise change step (4);
(4) select interior all the remote service assembly RSC of high capacity group as being adjusted object, all RSC that are lower than the load average in the selection in load and the low load group receive migrating objects as target to be selected;
(5) when each selects the load migration target as the RSC that is adjusted object in the lower RSC of load to be selected, at first calculate the load of RSC to be selected and the bias of average, pick out the migration target according to calculated value by the roulette algorithm then;
(6) in controlled remote service assembly RSC, each subordinate receives the size of message that agency's statistics receives separately, and presses size of message rank order from less to more, receives the corresponding subordinate of the minimum agency of size of message and pays the utmost attention to migration.After selecting subordinate to be migrated, the subordinate through correspondence receives the agency and will have the migration order of move target and be dealt into subordinate;
(7) after this subordinate receives the migration order, at first preserve the scene, break off the migration target that current connection also is connected to appointment again, recover on-the-spot at last;
(8) each RSC restarts load detecting with the load initialization in the RSC service groups;
When (9) the quilt adjustment object of repetition in finding, having occurred, explain thrashing to have occurred that this moment, this RSC need raise detection threshold than short-term;
(10) next adjustment cycle is carried out (1)～(10) step again.
Figure 11 has described the demand information fusion and reported and assign process: one of characteristics of the present invention are: all emulation terminals are only stated desired content; Do not state published content; And all be provided with action scope and these two kinds of attributes of cycle in every demand information, come demand for control informational influence scope and degree through these attributes.Local service assembly LSC and intermediate layer RSC send subordinate step by step and report after all demands merge and to circular on the same group, finally be reported to central core RSC service groups.Central core RSC service groups at first merges demand, issues step by step respectively then.Issue in the process, the demand that remote service assembly RSC at different levels send toward each subordinate all is the higher level, does not belong to the result that this tripartite demand of other subordinates of same RSC service groups merges on the same group and with this subordinate.Concrete principle is in the transmission:
1, demand information by tree bottom end in central core direction communication process, whenever just subtract 1 voluntarily through one deck its action scope, when action scope reduced to 0, this demand information can not be again to uploading; Demand information is not limited by action scope when at the same level and subordinate propagate, and this attribute does not change yet.Same type demand scope attribute when merging is got maximum wherein.For example to order the action scope of demand be 1 to certain reusable module; This demand information is only uploaded one-level till the local service assembly LSC; Remote service assembly RSC does not receive this demand information; Like this, this module receives only the message that inner other modules of simulated members are sent, and does not receive the influence of other nodes; If action scope is enough big, can arrive center remote service assembly RSC service groups, this demand information just can influence system's overall situation so, and this message that is equivalent to take place in the system-wide all will send in this module.
2,, be not that 0 subordinate connects the agency and sends to this demand action scope from higher level's message; Then unrestricted from the message with other subordinates at the same level.
3, periodic quantity is used for the frequency of demand for control information response.For example the cycle is 2 o'clock, and the every generation of this demand information just can be sent for 2 times, and periodic quantity is big more, and back-up environment is just slow more to the speed of this demand information response.Same type demand periodic quantity when merging is got minimum value wherein.
4, whether each local service assembly LSC is consistent with original demand information with the demand information that remote service assembly RSC at first detects new fusion, only when changing, just reports or issues.
Core of the present invention has 3 points: the one, adopted the multi-level Distributed Services of load balancing algorithm Network Based; The 2nd, simulated members further is divided into the reusable module that possesses the unified interface standard; The 3rd, having adopted can the self-defined action scope and the demand management pattern in cycle.Adopt the present invention can reach following beneficial effect:
1, make distributing emulation system possess extremely strong fault-tolerant ability and good NLB ability.As long as all also have a RSC ability operate as normal in each intermediate layer RSC service groups and the central core RSC service groups at least, recovery is normal voluntarily after of short duration adjustment just to guarantee system.
2, because multi-level remote service assembly RSC can play the data relaying action, thus can be so that all be unified under the distributed simulation back-up environment at computer through all participation emulation of multimeshed network address transition management multi-level local area network (LAN) down in the wide area network scope.
3, owing to adopt multi-level Distributed Services, effectively enlarged the supported scale degree of distributing emulation system.
4, improve the efficient of analogue system, solved the bottleneck problem in the analogue system of centralized in the past or layered subset Chinese style.
5, simulated members further is divided into the reusable module that possesses the unified interface standard of loose coupling, module supported by local service assembly LSC, make emulation to reuse granularity littler, strengthened reusability and flexibility.
6, compare the HLA standard, only order statement, exempted publication states, thereby simplified operation, and demand information can be provided with action scope and cycle, further optimize demand management and data distribution management pattern, improve the efficient of system.
Claims (4)
1. implementation method based on the distributed simulation back-up environment of multi-level Distributed Services; The organization and management method that has comprised local service assembly LSC (Local Service Component) and remote service assembly RSC (Remote Service Component); Simulated members provides service by local service assembly LSC; Each simulated members is made up of a plurality of reusable modules, and local service assembly LSC generates the agency for each module, and safeguards a local data environment of distinguishing by name; It is characterized in that: one group of simulated members under a plurality of local service assembly LSC support provides service by one group of remote service assembly RSC; The service groups of being made up of remote service assembly RSC is divided into intermediate layer RSC service groups and central core RSC service groups according to the status difference, and they are made up of the full remote service assembly RSC that connects of plurality of network respectively; A system has only a central core RSC service groups; Central core RSC service groups is managed some intermediate layers RSC service groups by tree; Can also extend the intermediate layer RSC service groups that makes new advances below each intermediate layer RSC service groups; The bottom be local service assembly LSC, simultaneously, central core RSC service groups and any one intermediate layer RSC service groups also can directly derive local service assembly LSC; Same simulated members disparate modules information interaction is independently accomplished by local service assembly LSC; Carry out information interaction between the different simulated members; Coordinate to accomplish by local service assembly LSC on the shortest network path between them and remote service assembly RSC, relate to global transaction and then participate in coordinating to accomplish through central core RSC service groups; Like this; Multi-level distributed remote serviced component RSC service groups is managed a plurality of simulated members groups; Form a distributed simulation system; And, comprise system management, time management, demand management, data distribution management, Object Management group and ownership management for it provides required service function; The realization of these service functions is based on the interoperability agreement of the inner definition of back-up environment, and the basis of this agreement is self-defining message.
2. based on the implementation method of the described distributed simulation back-up environment based on multi-level Distributed Services of claim 1; It is characterized in that: said back-up environment is to be core with central core RSC service groups; Through the outwards continuous service system of extending of a plurality of intermediate layers RSC service groups; Each service groups inside is the full peer-to-peer network relation that connects, and is the tree relation between group and the group; In any moment, as long as each service groups also has a remote service assembly RSC operate as normal at least, entire environment just can normally be moved.
3. the implementation method of the distributed simulation back-up environment based on multi-level Distributed Services according to claim 1 is characterized in that: deserve to be called grade service groups and subordinate's service groups mutually between central core RSC service groups, intermediate layer RSC service groups or the local service assembly LSC of the adjacent two-stage of symbolic animal of the birth year in the tree of said back-up environment.Each group member in subordinate's service groups selects the higher level group member who connects to decide according to the Message Processing duration in higher level's service groups; Measure the time difference between the superior and the subordinate through periodicity time adjustment message between the interconnected group member between at first upper and lower level service groups; The higher level group member comes the detect-message handling duration through the message time stamp and the data circulation that statistics obtains is organized entirely then; When group member's load differs big; The high capacity group member sends migration message to its certain group member of subordinate, makes the group member of this subordinate be connected to the lighter group member of other loads in higher level's service groups again; This method is used to dispose the group member's failure condition that occurs in the service groups equally.
4. based on the implementation method of the described distributed simulation back-up environment based on multi-level Distributed Services of claim 1; It is characterized in that: said simulated members only proposes demand information and gets final product; Carrying out demand information by back-up environment by level merges; Finally be aggregated into central core RSC service groups, and then be issued to each simulated members after merging by level respectively; Demand information has action scope and cycle two kinds of attributes, can further optimize data distribution management and demand management, improves system-wide efficient.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012100144437A CN102624543A (en) | 2012-01-17 | 2012-01-17 | Distribution simulation supporting environment implementation method based on multilayer distributed service |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012100144437A CN102624543A (en) | 2012-01-17 | 2012-01-17 | Distribution simulation supporting environment implementation method based on multilayer distributed service |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102624543A true CN102624543A (en) | 2012-08-01 |
Family
ID=46564234
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012100144437A Pending CN102624543A (en) | 2012-01-17 | 2012-01-17 | Distribution simulation supporting environment implementation method based on multilayer distributed service |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102624543A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1391162A (en) * | 2002-07-26 | 2003-01-15 | 中国人民解放军国防科学技术大学 | Method for implementing hierarchical distributed simulation operation support environment based on interoperation protocol |
| US20030125915A1 (en) * | 2001-11-30 | 2003-07-03 | International Business Machines Corporation | Count data access in a distributed simulation environment |
| CN101055530A (en) * | 2006-12-31 | 2007-10-17 | 中国人民解放军63791部队 | Embedded type distributed simulation platform based on HLA facing to space flight measurement and control and its realization method |
-
2012
- 2012-01-17 CN CN2012100144437A patent/CN102624543A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030125915A1 (en) * | 2001-11-30 | 2003-07-03 | International Business Machines Corporation | Count data access in a distributed simulation environment |
| CN1391162A (en) * | 2002-07-26 | 2003-01-15 | 中国人民解放军国防科学技术大学 | Method for implementing hierarchical distributed simulation operation support environment based on interoperation protocol |
| CN101055530A (en) * | 2006-12-31 | 2007-10-17 | 中国人民解放军63791部队 | Embedded type distributed simulation platform based on HLA facing to space flight measurement and control and its realization method |
Non-Patent Citations (3)
| Title |
|---|
| 刘斌: "基于分布式对象的统一RTI接口设计与实现", 《计算机工程》 * |
| 李铁: "基于消息的分布式仿真支撑环境的研究", 《系统仿真学报》 * |
| 李铁: "多Agent运行环境的软件框架与数据机制研究", 《系统仿真学报》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104461740B (en) | A kind of cross-domain PC cluster resource polymerization and the method for distribution | |
| Kamburugamuve et al. | Survey of distributed stream processing for large stream sources | |
| CN103458033B (en) | Event-driven, service-oriented Internet of Things service provider system and method for work thereof | |
| CN106934497B (en) | Intelligent community power consumption real-time prediction method and device based on deep learning | |
| CN100366001C (en) | Method of constructing cluster management and monitoring system in resilient system architecture | |
| CN107453929B (en) | Cluster system self-construction method and device and cluster system | |
| CN105677836A (en) | Big data processing and solving system simultaneously supporting offline data and real-time online data | |
| CN106339509A (en) | Power grid operation data sharing system based on large data technology | |
| Ekmefjord et al. | Scalable federated machine learning with fedn | |
| CN104239555B (en) | Parallel data mining system and its implementation based on MPP | |
| CN103188345A (en) | Distributive dynamic load management system and distributive dynamic load management method | |
| CN105094982A (en) | Multi-satellite remote sensing data processing system | |
| CN104008007A (en) | Interoperability data processing system and method based on streaming calculation and batch processing calculation | |
| CN103441918A (en) | Self-organizing cluster server system and self-organizing method thereof | |
| CN101860564A (en) | Protocol-based service combination system and method | |
| CN101778002A (en) | Large-scale cluster system and building method thereof | |
| CN102088475B (en) | System and method for executing combined service with centralized control flow and distributed data flow | |
| CN103473848A (en) | Network invoice checking frame and method based on high concurrency | |
| Siddiqui et al. | Microservices based architectures for IoT systems-State-of-the-art review | |
| CN102082737A (en) | Method for replacing Web services based on service priority | |
| Liu et al. | Super-peer-based coordinated service provision | |
| CN109002350A (en) | Way of Process Communication and device in a kind of group system | |
| CN102624543A (en) | Distribution simulation supporting environment implementation method based on multilayer distributed service | |
| Vasconcelos et al. | Autonomous load balancing of data stream processing and mobile communications in scalable data distribution systems | |
| Abdullah et al. | Reliable and efficient hierarchical organization model for computational grid |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20120801 |