Overview
"Digital Mockups" is a joint UC Riverside/Irvine research project developing
real-time observable/controllable executable models of physical systems.
Such real-time models can be used during development/test of the cyber
device in cyber-physical systems, such as a medical ventilator, as well as
for device-user training. As depicted below, a cyber device can be
connected to a mathematical model of the physical environment implemented
on a computing platform. That model's implementation represents a
"digital mockup" of the physical environment. For example, a medical
ventilator can connect to a human respiratory system model, a cardiac
pacemaker to a model of human heart electrocardiography, or a satellite
might think that it is in orbit when really it is being fed sensor data
by an environmental model of space. Digital mockups enable easier testing
of dangerous or expensive scenarios, more thorough testing of border
cases, fully automated testing, faster testing, reproducible testing, and
even a view of what's happening inside the physical system. They are useful
for test/development of new devices, for training/education (e.g., training
of respiratory therapists on ventilator use), and for reproducing field
errors.
Digital mockups can be used directly (typically requiring bypassing a device's
transducers and instead interfacing directly to internal device processors)
or can be incorporated into hybrid physical/digital mockups.
Synthesis of Models
Complex models commonly run much slower than real-time on normal computers.
A key aspect of the project is to compile mathematical models into circuits
for real-time implementation on Field-Programmable Gate Arrays (FPGAs). Circuits
are a good match for physical models, due to both items involving fine-grained
parallel computation and local communication. However, tools for automatically
compiling a model's differential equations into circuits do not exist today. We
have invented an automated tool flow for compiling high level model descriptions
to networks of Processing Elements for real-time execution on an FPGA.
See this link for more information.
Observe/debug
Digital mockups enable the possibility of observing the internal behavior
of a physical system, such as internal pressures in different regions of
a human lung during ventilation. Insights from such observation can lead
to better designs and to better user training. Furthermore, digital mockups
enable the possibility of controlling the physical model: stop, play,
fast forward, step, rewind, etc. Another key aspect of the project is
to create circuits such that internal values can be observed in real-time,
and debug control is incorporated too.
Methodology
Current digital mockup practice is scattered and ad hoc. Another project
aspect involves defining a discipline for digital mockup creation and use.
B. Miller, F. Vahid, T. Givargis. MEDS: Mockup Electronic Data Sheets for Automated Testing
of Cyber-Physical Systems Using Digital Mockups. Design Automation and Test in Europe (DATE), March 2012, pp. 1417-1420.
B. Miller, F. Vahid, and T. Givargis. Digital Mockups for the Testing of a
Medical Ventilator. ACM SIGHIT Symposium on International Health Informatics (IHI),
2012, pp. 859-862.
C. Huang, F. Vahid, and T. Givargis. A custom FPGA processor
for physical model differential equation solving. IEEE Embedded Systems Letters,
Sept. 2011, pp. 113-116.
B. Miller, T. Givargis, F. Vahid. Application-Specific Codesign
Platform Generation for Digital Mockups in Cyber-Physical Systems
IEEE Electronic System Level Synthesis Conf. (ESLsyn), June 2011, pp. 1-6.
S. Sirowy, T. Givargis, F. Vahid. Digitally-Bypassed Transducers:
Interfacing Digital Mockups to Real-Time Medical Equipment.
Int. Conf. of the IEEE Engineering in Medicine and Biology
Society (EMBC), Sept 2009.
Videos
The below video shows an implementation of a digital mockup for a commercial ventilator device. An FPGA hosts a lung model and
interfaces to internal ventilator processor busses to intercept actuator and sensor commands.
This work is supported in part by the National Science
Foundation (CNS1016792) and by the Semiconductor
Research Corporation (GRC 2143.001). Any opinions, findings, and
conclusions or recommendations expressed in this material are those of the
authors and do not necessarily reflect the views of the National Science
Foundation or other supporting agencies.