UC San Diego, USC and UCLA Awarded NSF Grant to Study Fundamentals of Cyber-Physical Systems
|Computer Science and Engineering professor Rajesh Gupta is leading the UCSD end of the joint effort with UCLA and USC.|
San Diego, CA, August 29, 2008 -- A coalition of computer scientists from UC San Diego, UCLA and the University of Southern California (USC) have been awarded a grant from the National Science Foundation (NSF) to study software design and run-time techniques for "cyber-physical systems," computer systems that are tightly integrated with the physical world.
"This is one of the first grants from NSF in this domain," says Rajesh Gupta, a professor of computer science and engineering at UCSD's Jacobs School of Engineering and a researcher affiliated with the California Institute of Telecommunications and Information Technology (Calit2). Gupta is one of the principal investigators on the $900,000 project (UCSD will receive $225,000 of the total funds). Other PIs include USC computer science Professor Ramesh Govindan and electrical engineering Professors Mani Srivastava and Paulo Tabuada of UCLA.
In addition to the NSF's research interest, the Networking and Information Technology Research and Development (NITRD) — the nation's primary source of federally funded research in computing, networking and software — has identified cyber-physical systems (CPS) as its number one research priority. Over the next three years, Gupta and his fellow researchers hope to establish the scientific principles and engineering practices governing the design of CPS, or "physically-coupled systems."
One example of a CPS is a sensor-based network that measures strain at various points of a bridge and sends information to a central station for determination of deterioration and damage prognosis. Other examples include smart transportation networks that optimize traffic movement in real time, based on road conditions; automatic pilot avionics; or in-home medical monitoring, such as the device developed by Calit2 at UC Irvine and Project Walk, a Southern California-based recovery program for those affected by spinal cord injuries (SCI) (read the full article about SCI and in-home health care).
The researchers will focus on four fundamental challenges inherent to the field as an intellectual discipline:
- Programming CPS software to capture data from the physical world and translate it into observables for embedded computation;
- Developing formal methods for software modules that indirectly interact with each other through the physical world;
- Programming CPS applications to allow for resource management that adapts to changes in the physical world; and
- Developing system software that shares physically-coupled sensor and actuator resources in distributed settings.
Explained Gupta: "In computer science, we typically have a limited set of observables. In physics, the observables could be very diverse. A cyber-physical system goes beyond the role of logic in computers to also include the physics in real-life systems. We know how to prove or disprove propositions in computer science or the physical laws in physical sciences through entirely different intellectual traditions. With CPS, we have to build both the mathematical rigor and physical empirical settings suited for CPS and construct theories based on observational data."
"There are great examples of why you need to construct CPS differently from mere modifications of existing computing machines," added the professor, who holds the Qualcomm Endowed Chair in Embedded Microsystems at UCSD. "The algorithms, the software, the logic, the theories -- they all need to be re-termed with physics in view."
As a means of illustration, Gupta cites the difference between the notion of time in a logical system and that in a physical system. The corresponding notions of time - logical versus physical - have different underlying calculus and roles from constructing flight simulator software to real-time flight control systems: what works for the simulator can have disastrous consequences for the real-life avionics.
In addition to examining the fundamental principles behind CPS, the researchers also seek to develop educational modules that will make it easier for the CPS community to train engineering talent for building software for real-world systems.
"We thought hard about how we can construct knowledge pieces that will be valuable," said Gupta. "Today's computer science and engineering courses are monolithic constructions that don't compose well with others. We are looking at producing a number of smaller modules that will be self-contained, and can be integrated into courses across domains. For instance, a course on embedded software can use one of these topical modules to teach students to build different types of controllers."
In the future, Gupta says, once the fundamental theories behind CPS have been established, they will have a transformational effect on the way we live, improving everything from our health and living arrangements to transportation and energy resource allocation.
"Medical diagnostics and treatment will not be done in specific places such as clinics, since the data provided by CPS will be robust and accurate," he explained. "We will use transportation only when necessary and transportation will be easier to use and more effective. Energy usage will change at the societal scale because of the way we will be managing network power grids."
But before any of that can happen, Gupta says he and his cohorts must decide which computing fundamentals they will incorporate into their research approach.
"The problem with this new domain is what I would call the dangers of clean-slate thinking," he notes. "The temptation is to ignore everything we know already and build things from the ground up. But much of the recent progress in computing has been evolutionary in nature, meaning it's built upon conceptualization and capabilities achieved earlier. Some of it is valuable, but some could be distractions. Striking a balance between the two is something I expect will be fairly challenging."