Model-Based Integration for FMI Co-Simulation and Heterogeneous Simulations of Cyber-Physical Systems
2 likes33,804 views
This document discusses model-based integration of heterogeneous simulations for cyber-physical systems. It presents the Command and Control Wind Tunnel (C2WT) approach, which uses models of interactions and shared data to integrate different simulation tools. C2WT supports the Functional Mock-up Interface standard to integrate Functional Mock-up Units as high-level architecture federates. A case study demonstrates integrating a vehicle thermal management system simulation partitioned across two Functional Mock-up Units using C2WT.
Model-Based Integration for FMI Co-Simulation and Heterogeneous Simulations of Cyber-Physical Systems
- 1. 3/19/2014 1 Model-Based Integration for FMI Co- Simulation and Heterogeneous Simulations of Cyber-Physical Systems Himanshu Neema, Jesse Gohl, Zsolt Lattmann, Janos Sztipanovits, Gabor Karsai, Sandeep Neema, Ted Bapty, John Batteh, Hubertus Tummescheit, Chandrasekar Sureshkumar March 11, 2014 Outline 1. Cyber-Physical Systems (CPS-s) 2. Generic Command and Control (C2) System-of-Systems Architecture 3. Multi-Model C2 Simulation Integration Problem 4. Command and Control Wind Tunnel (C2WT): A Model-Based Multi-Model Integration Approach – Integration examples – Capabilities 5. FMI-CS Integration – FMI-Standard for Co-Simulation – Model-based integration of FMUs in C2WT Federate (modeling, generation, execution) 6. Case Study – Simulation architecture, data and integration models – Experimental Results 7. Conclusions and Future Work 8. Key References and Contact
- 2. 3/19/2014 2 1. Cyber-Physical Systems (CPS-s) • CPS are composed of physical and computing components that interact through embedded communication capabilities. • Simulation of CPS is highly challenging because they involve tight coupling of a large no. of domains such as physical, mechanical, electrical, thermal, cyber, biological, acoustics, hydraulics, … • No single tool is enough – needs a library of well-researched and matured tools and models (some over 10+ years!). • Huge tool integration challenges: (i) different models of computation, semantics (ii) requires significant time and analysis • For efficient integration we need a “model-based platform” that: – Enables modeling of interactions and shared data among simulations – Enables modeling and integration of systems with diff. MoCs – Supports automated simulation execution on clusters – Supports multirate modeling with dynamic time management – For FMU Co-Simulations provides an efficient and standard master algorithm – Provides set of tools for generating necessary artifacts for rapid synthesis of simulations 2. Generic Command and Control (C2) System-of-Systems Architecture: Can contain several individual Cyber-Physical Systems Adaptive Human Organization Mixed Initiative Controller Context Dep. Command Interpretation Adaptive Resource Allocation HCI Abstract Data Distribution Network Coordination Decision Support Commands Platform Commands Assigned Platform Commands Platform Status COP Elements COP Elements COP Elements Model-Based Experiment Integration Environment: C2 Wind Tunnel (C2WT) C2 issues to be studied experimentally: • Distributed Mission Operation – Synchronization and coordination – Distributed dynamic decision making – Network effects • Increased Information Sharing – Shared situation awareness – Common Operation Picture (COP) – Network effects • Seamless Integration of Manned/Unmanned Assets – Mixed-Initiative Teams • System Level Impact Analysis – Cyber attacks and Resilient solutions – Strategy/gaming
- 3. 3/19/2014 3 3. Multi-Model C2 Simulation Integration Problem Processing (Tracking) CPN Devs 3-D Environment (Sensors) Delta3D How can we integrate the simulated heterogeneous system components? How can we integrate the simulation engines? How can we rapidly synthesize and deploy integrated simulations? GME GME Simulation Interaction Simulation Architecture OMNeT++ Network Architecture Controller/Organization/Coordination Vehicle Dynamics SL/SF Key idea: Integration is about messages and shared data across system components. Why don’t we model these messages and shared data elements and use these models to facilitate model integration as well as system integration? 4. Command and Control Wind Tunnel (C2WT): A Model-Based Multi-Model Integration Approach CPN Models OMNeT++ federate Simulink Models CPN federate Delta3D Graphics Models Devs Java federate Simulink federate OMNeT++ Models Devs Java Models Physics federate Sensor simulation federate High-Level Architecture (HLA) Run-Time Infrastructure (RTI): Portico (open source) Simulation models •Data models -- interaction & data models •Integration models -- data-flow, timing, & parameters Configuration Domain specific federates Model transformation Domain specific models -- abstract simulation models
- 4. 3/19/2014 4 4.1 C2WT: Tool Integration Example: Simulink Quadrotor (X4) Simulink model (Courtesy: Berkeley Univ.) Modified model Add input-output bindings GME integration model Generated .m Receiver and Sender S-function code + .java code for representing Simulink federate HLA Run-Time Infrastructure (RTI) Code generation RTI runtime communication Output binding Input binding Signal flow Signal flow 4.2 C2WT: Tool Integration Example: Colored Petri-Nets (CPN) CPN Execution Engine (HLA federate, java) HLA Run-Time Infrastructure (RTI) CPN model put input tokens get output tokens simulation control • Loads external CPN model • Synchronizes CPN model execution with RTI • Adds Sim Control place and transition to CPN to ensure model time progress • One step rollback optimistic execution (state save/restore) • Converts HLA interactions and object attribute change events to CPN tokens and back. • Updates CPN input places with incoming messages • Reads and removes output tokens and send them as HLA interactions or object state change messages Modeler can define how the CPN model is connected to the federation. Maps between HLA messages (interactions and object attribute changes) and CPN input and output places. C2W Modeling Environment Import CPN model • CPN I/O • CPN types (color sets) Generate I/O binding monitoring
- 5. 3/19/2014 5 4.3 C2WT: Other tool integration examples and capabilities Huge library of supported tools and mechanisms: • Other simulation tools (Ns-2, Delta3D, Google Earth, Java/C/C++, FMU-CS, etc.) • Passive federates (e.g. Loggers, monitors, etc.) • Live components (e.g. Emergency response, Traffic conditions, Human-in-the-loop, etc.) • Advanced support (e.g. Legacy FOMs, COAs, Expt. Config., Remote deployment, Gaming, etc.) 5. FMI-Standard • Open standard to solve integration issues with different simulation tools, integrators, IP protection • Re-uses OEM models with FMI-standard conforming interfaces for model access, control, and manipulation FMI for Model Exchange: FMI for Co-Simulation Standalone: FMI for Co-Simulation Tool:
- 6. 3/19/2014 6 5.1 Model-Based FMI-CS Integration in C2WT FMI-CS in C2WT Simulation process Time synchronization & Data-exchange via HLA-Bus 5.2 FMU-CS in C2WT: Initialization procedure (simplified) Start Load FMU zip archive Parse model description Load FMU’s shared libraries Setup IO and HLA-intr. maps Setup logging, and load model parameters (e.g., federate and federation name, step-size, no. of micro-steps per step) End Error checking at each step and exit on error. FMUs are federated using JFMI APIs from Berkeley.
- 7. 3/19/2014 7 5.3 FMU-CS in C2WT: Execution procedure (simplified) Start Synchronize FMU federate start with C2WT federates Request RTI to proceed time by step-size & wait for grant Y Got SimEnd or Simulation finished? N End Update input variables with HLA updates Execute FMUCS doStep() in Y step-size/#micro-steps-per-step chunks Req. Granted? Update HLA with output variables Note: During micro-steps, FMUCS can update internal variables, but input variables aren’t updated. Also, rejection of time-steps is not supported. 6. Case Study • A high-fidelity model of a Vehicle Thermal Management (VTM) system built using Modelon’s Vehicle Dynamics Library (VDL). Key components include: – Driver, Vehicle (Engine, Transmission, Driveline, Chassis, Aerodynamics, External loads, and Brakes), Lumped engine thermal mass, Lumped transmission thermal mass, Engine coolant fluid circuit, Transmission oil cooling circuit, Heat exchanger stack, Low voltage battery, Alternator, Cooling fan and controller, and Grill shutters and controller. • Partitioned into two parts: – Driver-Vehicle: Vehicle mechanics, electrical, and driver – Thermal-Management: Fluid and thermal parts of model – Physical connections bisected by boundaries are converted to causal signals (e.g., heat in mechanical -> lumped thermal model -> determines thermal mass temperature)
- 8. 3/19/2014 8 6.1 Simulation Architecture, Data and Integration Models Simulation architecture: Data model: Integration model: Thermal- Management 6.2 Sample Experimental Results
- 9. 3/19/2014 9 7. Conclusions, and Future Work • C2WT: A model-based integration approach: – Rapid synthesis of multi-model distributed simulations – Significant less integration effort (mostly automatic) – Less error-prone, repeatable experiments – FMU Co-simulation as an HLA federate along with non-FMI simulations for System-of-Systems (SOS) simulations – Real-time simulations including live components and humans-in-the-loop – Fully configurable, data logging, deployment modeling and remote execution on cluster, Legacy FOMs and FOM-mapping, Course-of-action simulation, Blue/Red gaming experiments, Open-source • Future Work – Other simulation integration (e.g. SysML) – Hardware emulation for slower than real-time federates – Partition matrix (several approaches) for multirate simulations 8. Key References and Contact • J. Sztipanovits, “Composition of cyber-physical systems,” 14th Annual IEEE Int’l. Conference and Workshops on the Engineering of Computer-Based Systems (ECBS ’07). Washington, DC, USA: IEEE Computer Society, 2007, pp. 3–6. • Graham Hemingway, Himanshu Neema, Harmon Nine, Janos Sztipanovits, Gabor Karsai, “Rapid Synthesis of High-Level Architecture-Based Heterogeneous Simulation: A Model- Based Integration Approach”, Simulation 88(2), 217-232 (2012) • C2WT community wiki – https://wiki.isis.vanderbilt.edu/OpenC2WT • Functional Mock-up Interface – www.fmi-standard.org • HLA standard – IEEE standard for modeling and simulation (M&S) high-level architecture (HLA) – framework and rules http://ieeex-plore.ieee.org/servlet/opac?punumber=7179 • JFMI: A Java wrapper for the Functional Mockup Interface – www.ptolemy.eecs.berkeley.edu/java/jfmi • Contact: Himanshu Neema Senior Staff Engineer Institute for Software Integrated Systems, Vanderbilt University Email: himanshu@isis.vanderbilt.edu Ph: +1-615-497-8136, Fax: 343-7440 http://www.isis.vanderbilt.edu
- 10. 3/19/2014 10 Backup Slide 1: High-Level Architecture (HLA) • An IEEE standard for “interoperable” and “reusable” models and simulations. – Most used specification (also used in the demo) is IEEE HLA 1.3 (1998) – Most recent specification is IEEE HLA 1516 (2000+) • United States DoD-wide policy mandates ALL defense models and simulations to comply with the standard. • Primary goal is to provide a general purpose infrastructure for “distributed” simulation and analysis. • Software implementing the HLA specification is called Run-Time Infrastructure (RTI). – Several commercial and open-source RTIs are available. – In the demo we used an open-source RTI PORTICO v0.7.1 implemented in Java language (http://porticoproject.org/). Backup Slide 2: C2WT Capabilities Overview Passive Federates -Data loggers -Monitors -Analysis -Prognostics -Projections Live components - UAVs - Command & Control - Live deployment feedback - Live traffic conditions - Emergency Response - Human-in-the-loop Run-Time Infrastructure (RTI) Federates Simulation Tools -Simulink -Omnet++/Ns-2 -DEVSJAVA -Ogre3D/Delta3D -Google Earth -CPN Tools -Java/C/C++ -FMU-CS, etc. C2 Wind Tunnel Component Integration Framework Advanced support -Compute infra-structure models -Deployment config -Remote execution Expt. config -Legacy sys. & FOMs (FOM-mapping) -Course-Of-Action (COA) Simulation -Blue Vs Adversary game generations