SimWare-Kernel Real Time Communication System for Simulation (Paper by: Bruno Calvo, Ignacio Seisdedos)

Transcription

1 Real Time Communication System for Simulation (Paper by: Bruno Calvo, Ignacio Seisdedos) Introduction Concepts Standards Implementation Interoperability scenarios Applications

2 Introduction SimWare is a simulation framework based on: Distributed simulation architecture (esim) Mission support architecture (MSOE) Concepts: Publish / subscribe paradigm Use of a data model D.M. D.M. Component 1 Component 2 Component N... Kernel D.M. Data Model D.M. R e c Executive P a r E v t - 1 E v t - 2 E v t - 3 E v t - 4 D e s

3 Concepts Publish-Subscribe Client-Server vs Publish-Subscribe How is data identified? Type: data types defined in data model (temp, pressure, ) Topic: data sent by an entity (tempx, pressurex) Console 1 Console 2 Console 3 Pressure X Temp X Pressure Y Temp Y /Sensor/Temp* /Sensor/*X /Sensor/Pressure*

4 Concepts Data Model Objective: gather in a single document all data types used by the system Objects and interactions Atributes and parameters Transportation Frecuency Entities

5 Standards : Data Distribution Service for Real-Time Systems Adopted Finalized Revised DLRL DCPS

6 Standards RTESS IEEE (1, 2, 3) SISO DLC IEEE Evolved RTI Ambassador Federate Ambassador

7 Standards Comparison - Similarities Publish/subscribe data communication User specified data types described in external file (FOM - IDL) Multiple federations ( domains) can co-exist Standard defined API Content based filtering Suitable for simulation applications Tools for data specification, application development, debugging, monitoring

8 Standards Comparison - Differences Designed for simulation Hierarchical object models Persistent objects No implied QoS semantics Logical time management No standard wire protocol Designed for real-time supports hierarchical object graphs (DLRL) has durability on topics Rich QoS semantics No implied time management semantics Open standard RTPS wire protocol

9 Implementation Motivation for a -based The similarities between and make fit perfectly as the low-level layer of an enhanced version of /RTI. Benefits of this implementation Enhanced performance, which meets new interoperability requirements (optimized for RealTime, available on different RT OS,...) Direct connection / (simple data from sensors to federations) Present to the user new QoS parameters. Easy migration, since Runtime Infrastructure shall meet the latest Dynamic link compatible API. Standarize the wire-protocol (RTPS). is also available for real-time simulations, when there is no need of semantics.

10 Implementation Middleware based on publish-subscribe paradigm, offering in a single API, access to either domains or federations. Application Kernel () Kernel () Simulation Used inside simulator QoS parameters Simulation Used to join simulators Interoperability with current applications

11 Implementation Kernel vs Real Time entity entity Kernel Kernel Middleware /RTI Mäk /RTI DMSO Kernel_ Kernel_ Advantages Allows an easy distribution of components between and simulations Unifies applications of many Runtime Infrastructure vendors through the use of the same interface. Allows an uniform use of data. Allows dynamic (from XML file) generation of data types.

12 Implementation Kernel API Kernel has a user-level API based on: Simplicity: focusing on just sending/receiving data. Having an advanced API to offer services: Declaration management (data discovery) Instance based publications, instance discovery Sinchronization points Save / restore Attribute based ownership Durability

13 Implementation Services Basic services QoS Transportation types: defined in the data model QoS Frecuency: defined in the data model Persistence: volatile () + transient (Kernel) Ownership: property management Domains/Federations: many independent simulations at the same time Liveliness: declaration management, discovery, error management, etc. Dead Reckoning Future Content filters Variable frequency Simulation discovery

14 Implementation Data Model use options Source Code Generator Data model ModeloDatos.hpp Application ModeloDatos.hpp Lib. Kernel (generic) Aplication Lib. Kernel (generic) Specific application (simulation modules, etc) Generic application (gateway, recorder, etc.)

15 Implementation Gateway Application using Kernel middleware designed mainly to join and applications. Joins domains and federations: Manages mapping of data between data models Filters, in order to allow only relevant data to be sent from one simulation to the other Joins not only user data, but also services: Ownership. Durability Entity discovery

16 Interoperability Scenarios simulation Performance improved simulations QoS parameters K. K. K. Simulation simulation, offering filtered data to High performance in simulation, allows real-time Offer only relevant data to federates, at lower frequency Federate 1 Federate 2 Federate 3 Federate 4 K. K. Simulation Federation

17 Interoperability Scenarios Reusing applications into simulations K. K. K. My App Simulation Distribution of applications. Reuse of components.

18 Interoperability Scenarios Reduce integration time Join many federations with different /RTIs (DMSO, Mäk,...) (DMSO) (Mäk)

19 Examples Flight Simulator Cockpit I/O Executive ehost Par Evt-1 Rec Evt-2 Evt-3 Evt-4 Des Kernel Scenario IOS Recorder elogistic SimWare

20 Examples Missile Mistral Simulator Evt-1 Par Par Rec Evt-2 Evt-4 Evt-3 Rec Evt-1 Evt-2 Evt-3 Des Evt-4 Des 0 2 Par Evt-1 Rec Evt-2 Evt-3 Evt-4 Des 1 SimWare

21 Summary A -based could bring important benefits: Lower Latency Better communication determinism and RT behavior Enhanced QoS as defined by the standard Wire interoperability Communication with non-simulation platforms The SimWare Kernel is platform for distributed simulation that provides the best of both worlds: Offers a publish-subscribe that abstracts (joins) and Provides a powerful, unified single data model Enables many deployment scenarios which may combine existing implementation platforms Application Kernel () Kernel ()