Presentation made in the 2017 IEEE System of Systems conference. Co-authored by Stephen Clement, David McKee, Richard Romano and Jie Xu (University of Leeds), Jose-Maria Lopez (Simware Solutions) and David Battersby (Jaguar Land Rover)

1. The Internet of Simulation:
Enabling Agile Model Based
Systems Engineering for Cyber-
Physical Systems
Stephen Clement, David McKee, Richard Romano, Jie Xu University of Leeds
Jose-Maria Lopez-Rodriguez Simware Solutions
David Battersby Jaguar Land Rover

2. Motivation
• Large-scale complex cyber-physical systems
• 20+ billion IoT devices by 2020
• 300 ExaBytes of data
• Increased usage of simulation and virtual engineering
• Geographically distributed working with multiple
stakeholders
• Increasing requirements for rapid prototyping
• Increased computing requirement for individual
simulations
• Growing need for integration and co-simulation to explore
system behaviours
• Digital twins
Raghu Ramakrishnan, CTO for Data
“…if we can concentrate and build platforms that will lift the
level of abstraction …you could instead focus on
understanding whatever your domain is, …[if] the tools for
making sense of your data …were that much more accessible
…the total cost of ownership would plunge…that's [a] form of
[Cloud] amortization. …the tools need to come closer and
closer to how you think about your world...”
“The way to build a complex system that
works is to build it from very simple systems
that work.”
Kevin Kelly, Founding Executive Editor
Wired Magazine

3. Internet of Simulation
• Amortize large-scale simulation and improve co-
simulation
• Net-Centric simulation platform
• Powered by the cloud based SOA
• Open APIs to connect simulations, people, processes
and things together
• Adherence to standards to guarantee connectivity with
third-party systems
• Simulation as a Service (SIMaaS)
• Provides self-contained readily integrated simulations
• Hosted as cloud services or on connected devices
• Workflow as a Service (WFaaS)
• Connect simulations together to define complete
systems
M&S

4. SIMaaS: Simulation as a Service
• Simulation tools and models are wrapped
as services and hosted in the Cloud
• Execution, timing and QoS management
provided by interfaces and underlying
infrastructure
• Provides interface for any type of
simulation on demand
• “Virtual Things”
• Uses :
• Virtual Engineering
• V&V mixing real and virtual
• Model of reality or decision support

5. WFaaS: Workflow as a Service
• Combine simulations together by defining interactions and shared data
• Hierarchical viewpoint allowing reuse of existing workflows
• Provide automated simulation integration via analysis of supplied simulation
interfaces and interactions
• Combine simulations with data stores and real-world devices
• Managed execution of co-simulation:
• Scheduling & distributing simulations
• Migration of live simulations
• Management of QoS
A B
EA1 D
Sub-Workflow

6. Extending IoS to IoT
• Cyber-physical system of systems with virtual components
• Virtual design and integration testing
• Iterative hardware and software tests
• Digital twin, through life-cycle development of cyber-physical systems

7. Standards: Communication
• Data Distribution Service (DDS)
• OMG standard for communication
• Publish-Subscribe model
• Inherent QoS
• Web-Services (WS-*)
• Collection of supporting standards already widely used in Cloud
• Asynchronous Message Queues
• Kafka, Rabbit MQ, ØMQ
• Common in Software as a Service (SaaS) applications
• Supports web standards and protocols

8. Standards: Composition and Orchestration
• Process Workflows
• Multiple languages for specification: YAWL, BPMN, BPEL
• Defines the flow of execution between services
• Many patterns defined by Russell et al. (www.workflowpatterns.com)
• Simulation Workflows
• Data-flow common in languages such as Simulink and Labview
• Inherently parallel, co-simulation is required across workflow
• Control time steps and synchronisation of workflow
• Mapping between system design and simulations
• Simulation interactions imply system interactions

9. Standards: Integration
• Functional Mockup Interface (FMI)
• Functional Mockup Units (FMU) as C code packaged with required libraries
• Not distributed, single shared memory environment
• tool support does not imply full co-simulation. Standards testing only between 2
simulation.
• High Level Architecture (HLA)
• IEEE 1516-2010 (HLA Evolved) distributed M&S standard
• Limited tool support & large barrier to entry
• Critical runtime infrastructure only specified as abstract interface
Standards define communication, data transfer and synchronisation methods.
No modelling framework for widespread simulation integration

10. Simulation Integration
• Complex systems require multi-disciplinary
simulation
• Heterogeneous development and execution
environments
• Integration environments
• Simulink, SimWorkbench, Modellica, Simware
• Social Problems:
• Lack of documentation
• Lack of centralised repositories
• Is my model up to date?
• Tool availability?
1. Collect suitable models and documentation
2. Check compatibility of models
• Signals
• Representations, Internal constants
• Time steps
3. Connect models
• Connect signals
• Transformation functions
4. Validation and Verification
• Existing test data
• Engineer intuition
• Correctness of representation
Integration Process

11. Technical Integration Problems
• Differing time steps: requires extrapolation, can result in instantaneous changes
• Incompatible constants: causes undefined behaviours, not always immediately recognised
• Incompatible signals or representations: Require transformation functions
• Lack of API: not easily integrated
• Validation and Verification: does the integrated model make sense?
• Co-simulation boundaries: especially a problem for physical simulations.
Model fidelity
Data sharing
infrastructure
Incompatible
interfaces
Data types
Timing
requirements
Model
complexity
Tool
incompatibility
Engineering Time
CoSimulation
Infrastructure
Model A
Model B
Independent
Development
Integration
Incompatible Requirements
Emergent Properties
Negotiation

12. Agile Systems Engineering using IoS
• Given:
• Rapid and reliable integration of simulations
• Mapping between system design (architecture and interactions) and simulations
• Variable fidelity of simulation
• Ready connection to physical and virtual components
• Allows:
• Rapid design iteration
• Easy sharing of design state among engineers
• Prototype at all stages of development mixing between maturity levels
• 3rd party suppliers providing prospective models.
• Traverse abstraction layers

13. Case Studies: Automotive Industry 4.0
Rapid
Prototyping
3rd Party Sims
Integrate external
Systems
Integrate with
local government
models and
systems
Digital twin
factory and
supply lines
Design offices readily
share models and iterate
on current designs
Multiple update
rates and time
steps
OEM models
Data from
other vehicles

14. Case Studies: Driving Sim
Throttle
Brake
Engine
Vehicle
Dynamics
TCU
Transmission
Micro-Service
Service
Leeds Driving Simulator

15. Case Studies: Server Heat
• Simulated data-center workload
• CPU execution and power draw
• Thermodynamic model of heatsink
• Fan modelling power and air flow

16. Case Studies: CITIUS
Research new technologies to command, control and interoperate autonomous vehicles in
military operations

17. Conclusion
• Growing need for integration of simulation
• We already do this, lets do it better!
• Internet of Simulation, a possible solution but:
• Need increased support from standards and vendors
• Design simulations with reuse in mind
• Agile Systems Engineering
• Example case studies

18. To know further…
University of Leeds - http://institutes.engineering.leeds.ac.uk/computing/research/distsys/
Simware Solutions - http://www.simware.es/internet-of-simulations.html
Jaguar Land Rover - http://www.jaguarlandrover.com/2016/technology-lab