JP6337760B2 - Time progress control device, simulator, and distributed simulation system - Google Patents

Description

  The present invention relates to a time progress control device and simulator in a distributed simulation system, and a distributed simulation system.

  A large-scale system in which a large number of subsystems are connected by a network is called a SoS (System of Systems) system. The evaluation of the function and performance of the SoS system is often carried out using a SoS simulator constructed as a distributed simulation system in which different types of simulators such as a network simulator and a subsystem simulator are connected, because the target is large and complex. A typical SoS system includes a defense system in which satellites, radar sites, aircraft, ships, vehicles, ground equipment, and the like cooperate with each other via a network such as a data link. The SoS simulator includes a distributed system described in Non-Patent Document 1. A defense system simulator in which a radar simulator, a platform simulator, a network simulator, etc. are linked based on HLA (High Level Architecture) which is a standard architecture of a simulation system.

  In the SoS simulator, an important issue in evaluating the function / performance of the SoS system to be simulated with high accuracy is the time synchronization of each simulator constituting the SoS simulator. Since the simulators operate independently of each other, correct simulation results cannot be obtained unless the contents simulated by the individual simulators and the input / output data times of the simulators are synchronized. In such distributed simulations, if real-time time values are used for synchronization between simulators, correct synchronization cannot be achieved due to the effects of clock errors and load status differences among multiple computers running the simulator. The logical time expressing the progress of simulation as a variable is used for. In HLA, Time Management Service is defined as a function for synchronizing logical time between simulators. Here, the time synchronization function based on HLA will be described as background art.

  First, the distributed simulation system architecture by HLA is shown in FIG. In HLA, the entire simulation system is called federation (901), and a program such as simulation processing that is executed in a distributed manner is called federation (902). A distributed simulation infrastructure called Run-Time Infrastructure (RTI) (903) provides common functions such as execution control of the entire federation and message communication control that the federation sends to other federations, which are performed during the execution of the simulation. .

HLA's Time Management Service is specified as a function that RTI provides for federation, and the current time of federation is based on the possibility of receiving Time Stamp Ordered (TSO) messages with time sent from other federations. It is a function to control. The Time Management Service mainly defines the following two functions.
(1) Control of time advance request requested by federation to RTI and progress time permission control permitted by RTI to federate (2) Progress time permission is permitted after federation requests time advance request from RTI Control of TSO messages from other federations that are delivered from RTI to that federation

  In the HLA, with respect to the functions (1) and (2), the federation can be selected from three types of time progression methods: Time Step method, Event Base method, and Optimistic method. Note that federations using different time progression methods may be mixed in one federation. These three types of time advancement methods will be described with reference to the examples of FIGS. In addition, in the time progression method in HLA, in addition to the parameters shown in FIG. 10, FIG. 11, and FIG. 12, Lookahead (time that ensures that the federation will not transmit a TSO message from the current time: for example, the current time is 10 and In the case of 5, a parameter that guarantees that a TSO message with a time less than 15 will not be transmitted is used. However, in FIG. 10, FIG. 11, and FIG. .

  First, an example of time progression by the Time Step method will be described with reference to FIG. The Time Step method is a method in which the time progress of federation is advanced by a simulated period (ΔT). In Step 1 of FIG. 10, the current time of federate 1 and federate 2 is Ti, and ΔT is 10. Federate 1 uses Send Interaction (SI), which is an RTI interface, to send Ti + 5 TSO message m1 and Ti + 15 TSO message m2 to Federate 2, and to use Time Advance Request, which is an RTI interface. (TAR) is used to make a time advance request to the next cycle time Ti + 10. This time advance request by TAR ensures that federate 1 does not transmit a TSO message of Ti + 10 or less. Next, in Step 2, when Federate 2 makes a time advance request to the next cycle time Ti + 10 using TAR, it is guaranteed that Federate 2 will not transmit a TSO message of Ti + 10 or less. RTI confirms the state of each federation, and determines that there is no possibility that a TSO message having a request time advance of each federation that has made a request for time advance by TAR will be generated in other than m1 sent by federation 1. Based on the determination result, in Step 3, the time advance grant (TAG) that is the RTI interface notifies the federation 1 and the federation 2 of the time advance permission to Ti + 10 that is the request time. For federate 2, the TSO message sent from federation 1 before notification of TAG is the TSO message at the Ti + 5 time below Ti + 10 allowed to proceed by TAG. m1 is delivered by Receive Interaction (RI) which is RTI interface. Since the TSO message m2 is a TSO message at a time Ti + 15 that is later than Ti + 10 permitted to proceed, delivery is suspended until the time of federation 2 becomes Ti + 15 or more. This is because there is still a possibility that future TSO messages will be generated from Ti + 10 at the present time, and the simulation process is performed when TSO messages from other federations in the future from Ti + 10 are indeterminate. This is to prevent a contradiction between federations that occurs. As described above, in the Time Step method, the federation obtains time progression permission at the request time when the time progression request is made, and the TSO message reception at a time later than the permitted time is deferred. To prevent inconsistencies.

  Next, an example of time progression in the Event Base method will be described with reference to FIG. The Event Base method is a method of advancing the time of federation by the smallest time among “current time / request time of other federate”, “minimum time of received TSO message”, and “request time”. Step 1 and Step 2 in FIG. 11 are the same as Step 1 and Step 2 in FIG. 10 except for Next Message Request (NMR), which is an RTI interface, when the federate 1 and federate 2 make a time advance request. Due to the time advance request by NMR, Federate 1 and Federate 2 guarantee that a TSO message of Ti + 10 or less is not transmitted during the time advance request. In Step 3, RTI checks the status of each federation, and federation 2 determines that there is no possibility of receiving a TSO message with a time less than Ti + 5, which is the time of TSO message m1, and determines the federation 2 permission time. It is determined that Ti + 5 is “the minimum time of the received TSO message”. Federate 1 determines that there is no possibility of receiving a TSO message at a time equal to or lower than Ti + 5 because the permitted time of Federate 2 is Ti + 5, and the permitted time is set to `` Other federated rate It is determined that the current time is Ti + 5. Based on the determination result, the TAG notifies the federation 1 and federation 2 of the time advance permission to Ti + 5. For federate 2, m1 of the TSO messages transmitted from federate 1 is delivered by RI before TAG is notified. Since the TSO message m2 is a TSO message at a time Ti + 15 in the future from Ti + 5 that is allowed to proceed, delivery is suspended until the time of federation 2 becomes Ti + 15 or more. As described above, in the Event Base method, the federation obtains time advance permission at the smallest time among “current time / request time of other federate”, “minimum time of received TSO message”, and “request time”. The reception of TSO messages at a time later than the permitted time is suspended, thereby preventing inconsistencies between federations.

  Next, an example of time progression of the Optimistic method, which is an extension of the Event Base method, will be described with reference to FIG. Like the Event Base method, the Optimistic method advances the time progress of the federation by the smallest time among the “current time / request time of other federates”, “minimum time of received TSO message”, and “request time”. Unlike the Event Base method, the TSO message is received at a time later than the permitted time. Steps 1 and 2 in FIG. 12 are the same as those in FIGS. 10 and 11 except that when the federation 1 and federation 2 make a time advance request, the time advance request is made using the flush queue request (FQR) that is the RTI interface. Same as Step1 and Step2. Due to the time progression request by FQR, Federate 1 and Federate 2 guarantee that TSO messages of Ti + 10 or less will not be transmitted during the time progression request. In Step 3, RTI checks the status of each federation, determines that federation 2 is no longer able to receive a TSO message at a time below Ti + 5 of m1, and determines the permitted time of federation 2 as `` Receiving TSO message Ti + 5, which is the “minimum time”. Federate 1 determines that there is no possibility of receiving a TSO message at a time equal to or lower than Ti + 5 because the permitted time of Federate 2 is Ti + 5, and the permitted time is set to `` Other federated rate It is determined that the current time is Ti + 5. Based on the determination result, the TAG notifies the federation 1 and federation 2 of the time advance permission to Ti + 5. For federate 2, all TSO messages transmitted from federate 1 are delivered by RI before TAG is notified. The TSO message m2 is a TSO message at a time Ti + 15 in the future from Ti + 5 that is allowed to proceed, and other TSO messages from Ti + 5 to Ti + 15 can be sent to federate 2 from now on However, Federate 2 can speculatively simulate Ti + 15 using TSO message m2 as input information. If a TSO message from Ti + 5 to Ti + 15 is received after speculative execution of Ti + 15 by TSO message m2, speculatively executed simulation processing and its TSO messages sent to other federates can be canceled by the simulation process. As described above, in the Optimistic method, the federation obtains time advance permission at the smallest time among the “current time / request time of other federate”, “minimum time of received TSO message”, and “request time”, It is possible to simulate a speculative future time by receiving a TSO message at a time later than the permitted time.

  Next, the problems of the Optimistic method, which is an extension of the Time Step method, Event Base method, and Event Base method in the SoS simulator, will be described. In the Time Step method, simulation processing is fixedly executed based on ΔT, and therefore the temporal accuracy of events occurring in the simulation (hereinafter abbreviated as time accuracy) depends on the magnitude of ΔT. For example, in order to simulate the effects of events that occur in a relatively short time, such as network communication delays, within a certain accuracy, it is necessary to reduce ΔT according to the accuracy. Since the simulation process is executed along with ΔT, useless processes increase and the simulation speed decreases. For this reason, the Time Step method is generally not used in simulators that require detailed time accuracy.

  On the other hand, in the Event Base method and the Optimistic method, since the simulation processing time is determined by the minimum time of the TSO message received by the simulator, the simulator time accuracy is not rounded by ΔT and is simulated with high time accuracy. It is possible. However, if the frequency of receiving TSO messages at different times is high, the progress of the time is delayed and the simulation speed is reduced. For this reason, Event Base and Optimistic methods are generally used in single-function simulators that require high time accuracy, such as network simulators, but simulation speeds are reduced in simulators that simulate complex events. Because it is not used.

  Therefore, in order to improve the simulation speed, methods that use the Time Step method, the Event Base method, and the Optimistic method according to the content simulated by the simulator are being studied. Patent Document 1 describes a simulation execution method in which a character (object) simulated by a simulator switches a time progression method by a Time Step type Event Base type. FIG. 13 is an example of a distributed simulation system configured by the method described in Patent Document 1. In FIG. 13, four federations of FED1 to FED4 are controlled by RTI. FED1, FED2, and FED3 simulate the Time Step type object, and FED3 from the command station object that FED4 simulates in addition to the Time Step type It is shown that the Event Base type is simulated by receiving the communication (however, here, the time progression of the Optimistic method by FQR is used for the Event Base type object). In other words, FED3 uses the time progression of the Time Step method by TAR and the time progression of the Optimistic method by FQR separately to simulate more precise time accuracy than ΔT by Time Step method for Event Base type objects. . FIG. 14 shows a processing flow in federation and RTI regarding the proper use of TAR and FQR in Patent Document 1. In step 721 in FIG. 14, the federation 3 determines whether or not an event base type object exists in the object that it simulates. If true, the federation 3 makes a time advance request by FQR in step 722. After the time progress by FQR, when federation 3 obtains time progress permission in step 723, it is possible to receive future TSO messages from the permitted time. If the TSO message is managed and it is confirmed in step 725 that the TSO message has been received from all target objects, in step 726, the time is advanced by the Time Step method with the smallest time of the received TSO message as the next request time. Change to a Time Step type object. This process is a loop process, and in the next execution, this object is changed to a Time Step type. Thereafter, in step 730, the object is changed to the Event Base type according to the state. In this way, when an object that advances time by the Event Base method is included in the federation, a TSO message at a future time is received by the Optimistic method, and the TSO message for that object is received from all objects When it becomes clear, the time speed is advanced by the Time Step type, so that the simulation speed can be improved.

IEEE 1516.1-2010-Standard for Modeling and Simulation High Level Architecture (HLA)-Federate Interface Specification Patent No. 5414332

  As described in the background art, in the conventional distributed simulation apparatus, when there is only one federation that performs time progression by the Optimistic method or Time Step method for the Event Base type object, the federation time progression method is used. The simulation speed can be increased by changing from the Event Base method to the Time Step method.

  However, if there are multiple federations that perform time progression by switching between Optimistic method or Time Step method, which is an extension of Event Base method, the permitted time of one federation that performs time progression by Optimistic method or Time Step method is The time is limited to the time according to the current time / requested time of other federations that perform time progression by the Optimistic method or the Time Step method. The federation permitted time by the Optimistic method is a time that guarantees that there is no possibility of receiving a TSO message below the permitted time for that federation, but whether or not a TSO message at a future time is generated. Is not finalized. Therefore, it is not possible to confirm that TSO messages have been received from all target objects in step 725 in FIG. 14, and the change to the time step type object in step 726 is not performed. As a result, a state in which the processing of the Time Step method after Step 728 is not performed, and a state in which the simulation speed can be improved by changing to the Time Step method cannot be realized. In this way, in the conventional distributed simulation apparatus, in the environment where there are multiple federations that switch the time progression of the Time Step method and the Event Base method in the federation, the time progression of each federation is smoothly switched from the Event Base method to the Time Step method. There is a problem that simulation speed cannot be improved.

  The present invention relates to a time progress control device and simulator in a distributed simulation system in which a plurality of federations can smoothly switch the time progress between the time step method and the event base method in an environment where there are a plurality of federations in the federation, and the distributed simulation system The purpose is to obtain.

The time progression control device according to the present invention is a time progression control device included in one simulator in a distributed simulation system in which a plurality of simulators cooperate based on logical time, and detects a communication start message of the own simulator and another simulator A communication start detection unit, a message list for registering a communication start message detected by the communication start detection unit, and when the communication start message is registered in the message list , the Event Base method is used as a time progression method. When the communication start message is not registered in the message list, a progress method selection unit that selects a time step method as a time progress method, and a communication result message that is a response to the communication start message is received Communication start message corresponding to the communication result message. And a progress control unit for deleting messages from the message list .

  According to the time progress control device of the present invention, even when there are a plurality of federates for switching the time progress between the Time Step method and the Event Base method, the time progress method of the own simulator is set between the Time Step method and the Event Base method. Can be switched smoothly.

1 is a basic configuration diagram of a distributed simulation system according to Embodiment 1 of the present invention. FIG. An example of message communication between the network simulator and the subsystem simulator according to the first embodiment of the present invention. The block diagram of the SoS simulator which concerns on Embodiment 1 of this invention. The processing flow which shows the operation | movement which the subsystem simulator 300 switches a time progress method in Embodiment 1 of this invention. A part of processing flow which shows the operation | movement which the subsystem simulator 300 switches the time progress method in Embodiment 1 of this invention. A part of processing flow which shows the operation | movement which the subsystem simulator 300 switches the time progress method in Embodiment 1 of this invention. An example of message communication between a network simulator and a subsystem simulator according to Embodiment 2 of the present invention. The block diagram of the SoS simulator which concerns on Embodiment 2 of this invention. The basic block diagram of the distributed simulation system architecture in a prior art. Time progression method by Time Step method in the prior art. Time progression method by Event Base method in the prior art. Time progression method using the Optimistic method in the prior art. An example of a distributed simulation system configured according to the prior art. Processing flow at federation and RTI in the prior art.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Embodiment 1 FIG.
FIG. 1 shows the configuration of an SoS simulator 101 that is a distributed simulation system according to Embodiment 1 of the present invention. Here, the distributed simulation system is a distributed simulation system in which a plurality of simulators cooperate based on logical time. In addition, the SoS simulator 101 is configured as a federation based on HLA, with one or more network simulators and subsystem simulators as federations, respectively. Here, the network simulator performs time progress by the Event Base method in consideration of time accuracy, and the subsystem simulator receives a relatively short time event such as a communication delay time simulated by the network simulator. Only the time progress is performed by the Event Base method, and the time step is performed by the Time Step method in other cases.

  FIG. 2 shows an example of message communication between the network simulator and the subsystem simulator in which the time progress request and time progress permission states are omitted for the sake of simplicity. In Step 1 of FIG. 2, the network communication trigger is first simulated at the time Ti of the subsystem simulator, and the simulation result is a TSO message to the network simulator and transmitted as a communication start message indicating the start of message communication. The time of the TSO message needs to be set to a time that is later than the current time in order to maintain consistency of the time progression of the federation and the delivery order of the TSO message. Also, in order to calculate the communication delay time after the network simulator receives the communication start message, the time of the communication start message is set to the current time plus a time smaller than the communication delay time calculated by the network simulator. There must be. Therefore, as the time of the communication start message, the time obtained by adding epsilon (a minimum value of logical time and not 0) defined by the HLA to the current time is used. In the example of FIG. 2, a positive integer value is taken, and a communication start message is transmitted as a Ti + 1 time message using an epsilon value “1”. Since the network simulator advances the time according to the Event Base method, the Ti + 1 time when the communication start message is received becomes the current time. Next, in Step 2, simulation of the network simulator at Ti + 1 time is performed, and in this example, communication delay time L is calculated as a result of communication success as a result of communication success / failure determination. Finally, in Step 3, the network simulator transmits a communication result message that is a TSO message to the subsystem simulator as a simulation result of the Ti + 1 time. At this time, the time of the communication result message is set to Ti + L, which is a reception time obtained by adding L, which is a communication delay time, to Ti that simulates the trigger of network communication. Note that when the communication failure is determined as a result of the communication success / failure determination, it is not necessary to calculate the communication delay time, and thus the time of the communication result is set to an immediate time such as Ti + 2 time, for example. FIG. 2 illustrates a case in which the subsystem simulator and the network simulator each have one case. However, in the case where there are a plurality of cases, the communication start message and the communication result message are transmitted to all the simulators.

  Here, the subsystem simulator only needs to obtain the time of the communication result message from the network simulator as the permitted time by the Event Base method. Necessary if the subsystem simulator normally advances the time using the Time Step method and switches the time progression method to the Event Base method only when it detects a communication start message that triggers the generation of a communication result message by the network simulator. Switching to the Event Base method only in such cases can improve the overall simulation speed. Also, if the time progress method is switched to the Event Base method only when a communication start message that triggers the generation of a communication result message by the network simulator is detected, the Time Step method and the Event Base method can be switched smoothly. Can do. Note that the invention described in the first embodiment can also be applied to a case where another simulator operating in the Time Step method exists in the distributed simulation system.

  FIG. 3 shows an outline of a distributed simulation apparatus according to the present invention. In the figure, a subsystem simulator 300 performs a simulation processing unit 301 that simulates a simulation target object or event in the subsystem simulator 300, a time progression control device 302 that switches a time progression method according to the situation, and interface processing for RTI. The RTI interface processing unit 303 is configured to perform. The time progress control device 302 of the subsystem simulator 300 includes a communication start detection unit 304 that detects a communication start message, a message list 305 that stores a communication start message detected by the communication start detection unit 304, and a message list 305. It comprises a progress method selection unit 306 that selects a time progression method according to the contents, and a progress control unit 307 that controls the time progression.

  Next, the operation will be described. 4, 5, and 6 are process flows showing the operation of the subsystem simulator 300 switching the time progression method according to the situation. 4, 5 and 6, the first part of the flowchart, the part following the first part, and the part following the first part are described. First, in step 401 in FIG. 4, the simulation processing unit 301 of the subsystem simulator 300 simulates Ti that is the current time. Next, in step 402, the communication start detection unit 304 detects the communication start message transmitted by the simulation processing unit 301 of the own simulator and another simulator, and registers it in the message list 305. A method for detecting a communication start message transmitted by another simulator will be described in the second embodiment. In step 403, the progress method selection unit 306 determines whether a communication start message is registered in the message list 305.

  When the communication start message is not registered in the message list 305 (the determination result of step 403 is No), the progress method selection unit 306 selects the time step method as the time progress method of the own simulator in step 404. In step 405, the RTI interface unit 303 makes a time advance request to Ti + ΔT by TAR, and in step 406, the RTI interface unit 303 receives time advance permission to Ti + ΔT by TAG. Thereafter, at 420, the simulation processing unit 301 simulates Ti + ΔT at the next cycle time. Thus, when the determination result in step 403 is No, the time is advanced by the Time Step method.

  When the communication start message is registered in the message list 305 (Yes in Step 403), the progress method selection unit 306 selects the Event Base method as the time progress method of the own simulator in Step 407, and FIG. In step 408, the RTI interface unit 303 makes a time advance request to Ti + ΔT by NMR, and in step 409, the RTI interface unit 303 receives time advance permission to Tj by TAG. Here, since Tj is the time progression permission time according to the Event Base method, Tj ≧ Ti and Tj ≦ Ti + ΔT.

  Next, at step 410 in FIG. 5, before the RTI interface unit 303 receives the TAG at step 409, the progress control unit 307 determines whether or not the communication result message has been delivered from the RTI. When the determination result in step 410 is No, the progress control unit 307 determines whether Tj = Ti + ΔT in the RTI interface unit 303 is true in step 411. If the determination result in step 411 is Yes, the communication result message from the network simulator is determined to be transmitted at a time later than Ti + ΔT, which is the next cycle time of the own simulator. The control unit 307 instructs the simulation processing unit 301, and the simulation processing unit 301 simulates Ti + ΔT at the next cycle time. On the other hand, if the determination result in step 411 is No, it is determined that there is still a possibility that the communication result message from the network simulator is transmitted at a time smaller than Ti + ΔT, which is the next cycle time of the own simulator. Therefore, the progress control unit 307 instructs the RTI interface unit 303 again in step 408, and the RTI interface unit 303 makes a time progress request to Ti + ΔT by NMR.

  If the determination result in step 410 is Yes, first, in step 412, the RTI interface unit 303 deletes the communication start message that is the basis of the received communication result message from the message list 305. Next, in step 413, the progress control unit 307 determines whether or not Tj = Ti + ΔT is true in the RTI interface unit 303. If the determination result in step 413 is Yes, it is determined that the communication result message from the network simulator has been transmitted at the same time as Ti + ΔT, which is the next cycle time of the own simulator. Instructs the simulation processing unit 301, and the simulation processing unit 301 simulates Ti + ΔT at the next cycle time. If the determination result in step 413 is No, first, the progress control unit 307 instructs the simulation processing unit 301 in step 414, the simulation processing unit 301 simulates Tj at the next time, and communication starts in step 415. In the detection 304, the communication start message transmitted by the simulation processing unit 301 of the own simulator and the other simulator is detected and registered in the message list 305. However, if a communication start message is not detected in step 415, a new communication start message is not registered in the message list 305. Next, in order to determine whether or not to proceed with the time based on the Event Base method, in step 416, the progress method selection unit 306 determines whether or not a communication start message is registered in the message list 305.

  If the determination result in step 416 is Yes, it is determined that the simulator still has a possibility of receiving a communication result message from the network simulator. The interface unit 303 makes a time progress request to Ti + ΔT by NMR.

  If the determination result in step 416 is No, all communication result messages based on the communication start message already transmitted to the network simulator have been received, and communication result messages with a time smaller than the next cycle time Ti + ΔT are displayed. It is determined that there is no possibility of receiving. Therefore, in step 417 of FIG. 6, the progress method selection unit 306 selects the Time Step method as the time progress method of the simulator, and in step 418, the RTI interface unit 303 uses the TAR to generate Ti The time advance request is made until + ΔT, and in step 419, the RTI interface unit 303 receives the time advance permission to Ti + ΔT by TAG. Thereafter, in step 420, the simulation processing unit 301 simulates Ti + ΔT at the next cycle time.

  As described above, in the first embodiment of the present invention, the communication start detection unit 304 of the time progress control device 302 of the subsystem simulator 300 is simulated by the simulation processing unit 301 of the own subsystem simulator and the other subsystem simulator. When the start message is detected and the communication start message is detected, the progress method selection unit 306 selects the Event Base method as the time progress method, and the Event Base method until all the communication result messages corresponding to the detected communication start message are received. After the time progress is performed and all communication result messages corresponding to the detected communication start message are received, the progress method selection unit 306 selects the Time Step method as the time progress method. With the above operation, it is possible to simulate the reception event of communication that requires detailed simulation of time accuracy in detail by the Event Base method, and other simulations can be executed in the time period by the Time Step method. Is possible. Since the switching between the Event Base method and the Time Step method is executed according to the detected communication start message, unlike the method of Patent Document 1 cited in the prior art, it is not based on the state of other federations. Even when there are multiple federates that switch the time progression between the method and the Event Base method, the time progression method of the simulator can be smoothly switched between the Time Step method and the Event Base method.

  That is, the time progress control device 302 according to the first embodiment of the present invention is the time progress control device 302 included in the subsystem simulator 300 that is one simulator in a distributed simulation system in which a plurality of simulators cooperate based on logical time. Then, a communication start detection unit 304 that detects communication start messages of the own simulator and other simulators that are proceeding in time by the Time Step method, and a message list 305 that registers the communication start messages detected by the communication start detection unit 304 And a progress method selection unit 306 that selects either the Time Step method or the Event Base method as the time progress method based on whether or not the communication start message is registered in the message list 305, and a response to the communication start message. When a communication result message is received And a progress control unit 307 that deletes a communication start message corresponding to the communication result message from the message list 305. With such a configuration, the time progression method of the simulator can be switched to the Event Base method only when a communication start message that requires the Event Base method is detected. Also, unlike the method of Patent Document 1 cited in the prior art, it does not depend on other federation states, so it operates even when there are multiple federations that switch the time progression between the Time Step method and the Event Base method. Is possible.

  Further, the progress method selection unit 306 of the time progress control device 302 according to the first embodiment of the present invention selects the Event Base method as the time progress method when a communication start message is registered in the message list 305. And The progress method selection unit 306 of the time progress control device 302 selects the Time Step method as the time progress method when the communication start message is not registered in the message list. With such a configuration, a communication start message that requires the Event Base method is transmitted, and the Event Base method can be used only in a section in which processing for the communication start message is not completed.

  The simulator according to the first embodiment of the present invention includes a simulation processing unit 301 that simulates a simulation target object or event, an interface processing unit 303 that performs interface processing with an RTI that is a distributed simulation base, and a time progression control device 302. And. With such a configuration, in a distributed simulation system in which a plurality of simulators cooperate based on logical time, a simulator that can be used by appropriately switching between the Time Step method and the Event Base method can be obtained.

  In addition, the distributed simulation system according to the first embodiment of the present invention includes a subsystem simulator 300 that is a simulator that switches time progress between the Time Step method and the Event Base method, and a network that is a simulator that performs time progress using the Event Base method. And a system simulator. With such a configuration, in a distributed simulation system in which multiple simulators cooperate based on logical time, even when multiple simulators use the Event Base method for time progression, the Time Step method and the Event Base method are switched smoothly and used. A distributed simulation system can be obtained.

Embodiment 2. FIG.
In the first embodiment, the method for detecting the communication start message transmitted from the own simulator is described. In the second embodiment, the method for detecting the communication start message transmitted from the other simulator to the network simulator is described.

  Embodiment 2 shows a method of detecting a communication start message using a communication start notification message by a Received Ordered (RO) message, which is a message for which the subsystem simulator does not receive delivery control based on logical time. FIG. 7 shows an example of message communication between the network simulator and the subsystem simulator in the second embodiment. FIG. 8 shows an outline of a distributed simulation apparatus according to Embodiment 2 of the present invention. (1) in FIG. 7 is an example of delivery of a normal communication start message. Subsystem simulator 2, which is another simulator, simulates a network communication trigger at the current time Ti, and transmits a Ti + 1 communication start message to the network simulator and subsystem simulator 1 as a result of the simulation. When the current time of the subsystem simulator 1 is Ti, since the time of the communication start message is a future time, the communication start message is delivered until the subsystem simulator 1 advances the time at a time later than the time of the communication start message. Is put on hold. For this reason, the communication start detection unit 304 of the subsystem simulator 1 cannot detect the normal communication start message delivery as shown in (1) of FIG.

  In view of this, the present embodiment has devised a method for notifying that a communication start message has been transmitted using an RO message that is delivered regardless of the current time of federation. (2) in FIG. 7 is an example of processing for notifying that a communication start message has been transmitted by an RO message. First, the subsystem simulator 2 simulates the trigger of network communication at the current time Ti, and as a result, a communication start message of Ti + 1 is transmitted to the network simulator and the subsystem simulator 1, and notification that the communication start message has been transmitted is sent. A communication start notification message is transmitted as an RO message. Since the communication start notification message is an RO message, the subsystem simulator 1 can receive this message regardless of the current time and is delivered by RTI. The communication start detection unit 304 of the subsystem simulator 1 detects that the communication start message has been transmitted by receiving the communication start notification message. Note that the detected communication start message is registered in the message list 305 until a communication result message based on the communication start message is received in time progress according to the Event Base method, as in the first embodiment. Further, after the communication start message registered in the message list 305 disappears, the subsystem simulator 1 switches the time progression method from the Event Base method to the Time Step method (the processing of Step 415 and Step 416 in FIG. 5). Therefore, in the situation of (2) in FIG. 7, the subsystem simulator 1 has to detect all the communication start messages at the time between Ti which is the current time and Ti + ΔT which is the next cycle time of itself. . Therefore, when the subsystem simulator 2 makes a time advance request to Ti + ΔT, the subsystem simulator 2 sends a time advance request notification message of the RO message to the subsystem simulator 1 so that the subsystem simulator 2 newly adds up to Ti + ΔT. The subsystem simulator 1 is notified that no message transmission is performed. The subsystem simulator 1 determines that a communication start message at a time up to Ti + ΔT is not generated any more by receiving the time progress request notification message.

  As described above, in the second embodiment of the present invention, the communication start detection unit 304 of the time progress control device 302 of the subsystem simulator detects the communication start message simulated by the simulation processing unit 301 of the other subsystem simulator. As a method of using a communication start notification message by an RO message and a method of using a time progress request notification message by an RO message as a method of determining that all communication start messages up to the current time + ΔT have been detected. Such a method can be realized by the subsystem simulator 300 including the message transmission unit 308 that transmits the RO message, and the communication start detection unit 304 having a function of receiving the RO message. With the above operation, the communication start detecting unit 304 can detect that a communication start message that is a TSO message at a future time from the current time and that all communication start messages up to the current time + ΔT have been detected. is there. Further, by using the subsystem simulator 300 having both the configurations described in the first embodiment and the second embodiment, it is possible to detect both the communication start message of the own simulator and the other simulator.

  As described above, the time progress control device 302 according to the second embodiment of the present invention includes the message transmission unit 308 that transmits the communication start message of its own simulator to another simulator in a delivery format that does not depend on the logical time. And With such a configuration, a communication start message can be notified to other simulators regardless of the logical time, and even when multiple simulators switch the progress time between the Time Step method and the Event Base method. The simulator can smoothly switch the progress time between the Time Step method and the Event Base method.

  In addition, the time progress control device 302 according to the second embodiment of the present invention includes a message receiving unit that receives a communication start message transmitted from another simulator in a delivery format that does not depend on logical time. With this configuration, communication start messages can be received from other simulators regardless of the logical time, and even when multiple simulators switch the progress time between the Time Step method and the Event Base method, each simulator It is possible to smoothly switch the progress time between the Time Step method and the Event Base method.

  Further, the delivery format not depending on the logical time is a time progress notification message by a Received Ordered message. With this configuration, a communication start message can be notified to other simulators regardless of the logical time.

Embodiment 3 FIG.
In the second embodiment, the method for detecting the communication start message using the RO message has been described, whereas in the third embodiment, the Management Object Model (MOM) that can be applied to a simulator that cannot be modified with a commercial product. ) A method for detecting a communication start message using the function will be described.

  The method of detecting a communication start message using the RO message described in the second embodiment is a form in which a message transmission function for controlling execution of the simulator is added to the subsystem simulator. It cannot be applied to a simulator that cannot. As a method for enabling such a simulator to detect a communication start message, a method using the Management Object Model (MOM) function defined as the RTI function in the HLA protocol will be described.

  The MOM function is a function for monitoring federation by RTI and notifying the federation of the monitoring information. HLAobjectRoot.HLAmanager.HLAfederate.HLAreport.HLAreportServiceInvocation of MOM Interaction Class defined as the MOM function (The middle "." Indicates that the Interaction Class on the right inherits the Interaction Class on the left. (Omitted) is a function that, when a federation to be monitored uses an RTI interface, delivers information on the interface and arguments as an RO message to the monitoring federate by RI. By subscribing to the HLAreportServiceInvocation, the subsystem simulator can know the execution status of other federations calling to the RTI and the interface called from the RTI. By monitoring the transmitted communication start message transmission and the time progress request, the communication start message that is a TSO message with detailed time from the current time is detected, and the current time + ΔT until the current time + ΔT due to the time progress request to the current time + ΔT It is possible to determine whether to detect all communication start messages.

  However, the HOMreportServiceInvocation of the MOM function can limit the federation that you want to monitor and the HLA function that you want to monitor, but because it sends a large amount of information on the RTI interface used by the monitored federate, it performs a simulation process that is not dedicated to federation monitoring When the Federate subscribes to HLAreportServiceInvocation, there is an adverse effect that the processing load of the federation increases and the execution speed of the simulation process decreases. For this reason, both communication start message detection using the MOM function is used only for simulators that cannot be modified, and communication start message detection is performed for other simulators using communication start notification messages. A method using both methods is also conceivable.

  The simulator according to Embodiment 3 of the present invention is characterized in that the communication start message of another simulator is detected by using the Management Object Model function of High Level Architecture. With such a configuration, it is possible to detect a communication start message of another simulator even in a simulator that cannot be modified with a commercial product.

  Any combination of Embodiments 1 to 3 of the present invention is included in the present invention.

101: SoS simulator, 300: Subsystem simulator, 301: Simulation processing unit, 302: Time progress control device, 303: Interface processing unit, 304: Communication start detection unit, 305: Message list, 306: Progression method selection unit, 308 : Message transmission unit, 901: Federation, 902: Federate, 903: Run-Time Infrastructure

Claims (7)

A time progression control device included in one simulator in a distributed simulation system in which a plurality of simulators cooperate based on logical time,
A communication start detection unit for detecting a communication start message of the own simulator and another simulator;
A message list for registering a communication start message detected by the communication start detection unit;
When the communication start message is registered in the message list, the Event Base method is selected as the time progression method, and when the communication start message is not registered in the message list, the Time Step method is selected as the time progression method. A progress method selection section to select as,
When a communication result message that is a response to the communication start message is received, a progress control unit that deletes the communication start message corresponding to the communication result message from the message list;
A time progress control device comprising: The time progress control device according to claim 1, further comprising a message transmission unit that transmits a communication start message of the simulator to another simulator in a delivery format that does not depend on a logical time. The communication start detection unit any one of claims 1 or claim 2, further comprising a message receiving unit that receives the communication start message transmitted by the delivery format that does not depend on the logical time from other simulators The time progress control device according to 1. The time progress control device according to claim 2 or 3 , wherein the delivery format not depending on the logical time is a time progress notification message by a Received Ordered message. A simulation processing unit for simulating a simulation target object or event;
An interface processing unit for performing interface processing with the distributed simulation platform;
The time progress control device according to any one of claims 1 to 4 ,
A simulator characterized by comprising: The simulator according to claim 5 , wherein the communication start message of the other simulator is detected by using a Management Object Model function of High Level Architecture. The simulator according to claim 5 or 6 , wherein the time progress is switched between the Time Step method or the Event Base method,
A simulator that uses the Event Base method for time progression,
A distributed simulation system comprising:

Images