Engineers work unobtrusively across the street from the Rhinestone Wedding Chapel, Bobby’s Idle Hour bar and recording studios in Nashville, breaking out of the traditional boundaries of computer research at Vanderbilt’s Institute for Software Integrated Systems (ISIS) right in the heart of the city’s Music Row.
“In a way it’s synergistic,” says Janos Sztipanovits, E. Bronson Ingram Distinguished Professor of Engineering. “All the creative types come together in this area. It’s a good mingling place for both geeks and musicians.”
The founder and director of ISIS, Sztipanovits recently spearheaded the institute’s transition from smaller, less modern digs to new headquarters on 16th Avenue just blocks away from campus. It was a fitting upgrade for a team that won more than $17.5 million in research funding for 2011. Of that, $12.5 million represented new awards, all in major national research programs.
“We are a major source of design methods, and not only that, we create open source tools, which makes our new design technology widely accessible to the public.”
Fueling its pioneering research are rapid innovations in information technology that drive enormous changes in science and engineering. This information technology growth has an impact on virtually every system encountered by humans: health care, education, transportation, defense and even the environment.
Since its establishment in the School of Engineering in 1998, ISIS has become an internationally recognized science and technology center for both designing and creating physical and computational systems, from small, embedded devices like pacemakers to globally deployed complex systems such as networks of satellites.
“We are a major source of design methods, and not only that, we create open source tools, which makes our new design technology widely accessible to the public,” Sztipanovits says. “ISIS software tools get serious reviews every day from users worldwide, multiplying the impact of our academic publication tremendously.”
ISIS crosses boundaries without hesitation to find new ways to solve today’s intricate engineering problems, he says. “Our research portfolio reflects that agility completely. The technology core of what ISIS is building—model-integrated computing—is really at the epicenter of this transformation in engineering,” he says.
Patients and Defense
Recent ongoing research highlights the institute’s broad impact. Sztipanovits led an ISIS team, for example, in a collaborative project with the Vanderbilt University Medical Center to develop a patient management system for sepsis treatment. Triggered when bacteria invades through wounds or IV lines, sepsis causes the body to literally attack itself and leads to more than a quarter million deaths annually. Now in clinical trial in the hospital’s intensive care unit, Vanderbilt’s system for rapid sepsis detection integrates with an automated decision support system to help guide physicians through the involved treatment process.
The project is part of a larger collaborative effort with the Medical Center to create a new generation of health information systems that are privacy aware and secure. The effort is supported by the National Science Foundation as part of the Science and Technology Center TRUST (Team for Research in Ubiquitous Secure Technology), as well as the Department of Health and Human Services’ Strategic Health IT Advanced Research Project on Security (SHARPS), funded by a $1.6 million federal grant.
At the same time that ISIS piloted this innovative patient management system, a national project began that has the potential to transform the manufacturing processes in the defense industry. The Adaptive Vehicle Make research program, a flagship initiative of the Defense Advanced Research Project Agency (DARPA), represents a challenge to a large research team that includes representatives of prominent institutes, corporations and universities, of which Vanderbilt is a lead player.
The team must figure out, among other charges, how to build a complex vehicle like an amphibious combat vehicle in one-fifth of the usual time. ISIS’ Ted Bapty and Sandeep Neema spearhead the initial phase of the AVM project, focusing on design languages, automation and flow.
Computer software, hardware and myriads of physical components have to integrate seamlessly to meet DARPA’s challenge, the researchers say. The ultimate goal is democratization of design, where not only major manufacturers can come up with innovations, but small companies, individuals, even student groups have a chance to compete. To test the idea, DARPA will distribute the resulting tool suite to high schools and initiate national competitions where the best designs will be manufactured in automated AVM fabrication lines.
“There’s an incredible number of engineering domains or disciplines that have to be involved to make this happen,” says Bapty, research associate professor of electrical engineering.
The technology base, however, is common, says Neema, research associate professor of electrical engineering. “We should be able to apply these concepts to a variety of vehicles from a submarine to a flying Humvee.” (The flying Humvee only exists in the imagination—for now).
Another high-profile assignment has Sztipanovits and Xenofon Koutsoukos, associate professor of computer science and computer engineering, pairing with General Motors, the University of Maryland and the University of Notre Dame in a project called the Science of Integration for Cyber-Physical Systems.
The five-year, $5 million NSF-funded project tackles the precise and theoretically well-founded engineering of cyberphysical systems. CPS are the new generation of engineered systems built as networks of interacting computational and physical elements to deliver advanced capabilities in cars, aircrafts and spacecraft.
“We do not have a science to do this integration,” Koutsoukos explains. “The problem is extremely difficult and very costly. Companies design new models, and they have to do it fast while managing costs and making sure the product is safe.” At the same time, new design and technologies are rapidly changing.
Most cars and planes combine multiple components from multiple manufacturers and it is not always well understood how the components work together, Koutsoukos says. Further complicating matters is the issue of intellectual property—manufacturers don’t want to provide information that would inform competitors.
That is where ISIS engineers can make a real impact. Their computer modeling techniques help predict and evaluate how different parts—from software to hardware to motors, wires, various materials and moving parts—will interact.
The new integration science (supported by design tools) that the team is charged with creating would ease the integration of all components. In the final phase, the researchers will create virtual prototypes to simulate a vehicle so it can be tested before manufacturing—all while keeping down costs and avoiding errors.
Complex Software in the Air
Another area where ISIS researchers apply model-integrated computing is in creating models that can diagnose faults in systems before they happen. For one such project, Koutsoukos pairs with Gautam Biswas, professor of computer science and computer engineering, on a grant from NASA to improve software health management (the system dependability and prognostics) in modern aircraft.
Working closely with Honeywell and a regional airline, Biswas, Koutsoukos and others on the team are developing VIPR (Vehicle Integrated Prognostic Reasoner), a system which seeks to isolate, detect and prevent adverse events in commercial aircraft.
As part of the project, the researchers employed data mining algorithms to analyze years of flight data to uncover where irregularities occurred, find out why they happened and discover ways to detect problems earlier.
“In one adverse event we found, the engine shut down fairly soon after takeoff. The plane was forced to return to the tarmac,” Biswas explains. The FAA considers that a serious event, even though no one was injured.
The researchers went back at least 50 flights and analyzed data for that particular plane. They made an interesting discovery: A small leak had developed in a fuel meter near an engine. The engine, receiving erroneous information that it wasn’t getting enough fuel, began to overcompensate. The meter eventually ceased to function, which led to the engine overheating and shutting down. If the software system had communicated the fuel gauge malfunction earlier, the engine problem could have been avoided. “We are using data-mining algorithms to process data and derive the precise knowledge to catch faults earlier,” Biswas explains.
Challenging and High Stakes
Although ISIS has a partner list packed with household names ranging from aircraft manufacturers to the U.S. Department of Education, some of its most complex projects are part of the security and defense realm.
In one, Associate Professor Koutsoukos works with the Army Research Office in collaboration with MIT; the University of California, Berkeley; and the University of Memphis on a five-year DARPA-funded project to refine a sensor network for tracking and target recognition in urban terrain.
In a different collaborative effort, Gabor Karsai, professor of electrical engineering and computer science, leads a team that partners with George Mason University to create decision support tools to help the military determine the best course of action in complex situations. The work, sponsored by the Air Force Research Laboratory, has implications for disaster preparedness in emergencies like the aftermath of Hurricane Katrina. Evaluating potential problems in action plans means planners can make small fixes now to prevent big problems later.
Perhaps Karsai’s most exciting program is part of the creation of a network in the sky. It is called the F6 project and is funded by DARPA, with NASA acting as technical supervisor and Lockheed Martin and Kestrel Institute as subcontractors. The engineers are challenged to create an advanced space system of many smaller satellites that could communicate with each other while hurtling through orbit at 25,000 miles per hour.
“Conventional satellites are single, very expensive and very large, and if something goes wrong, very hard to repair,” Karsai explains. A networked system of smaller satellites creates redundancies that mean the failure or loss of one or two satellites wouldn’t be disruptive. The ISIS team will design and build the information architecture for the $5 million undertaking.
“This is a challenging and high-stakes project. In two years, we are going to do a flight test. Whatever we build will end up on the platform,” Karsai says.
Smartphone for the Defense
Akos Ledeczi, associate professor of computer engineering, and his team are working on a countersniper application for smartphones that will aid soldiers in battle. The app, called SOLOMON (Shooter Localization with Mobile Phones), is funded by a two-year, $500,000 grant from DARPA.
Here’s how it works: A custom headset worn by solders is programmed to collect the sound of gunfire and send the information to the soldier’s smartphone. Neighboring phones share the data, compute the location of the shooter and display it using Google Maps. Building on earlier prototypes built by Ledeczi’s team, this version runs off a single microphone per smartphone and does not require a central computer to work. Vanderbilt has applied for patents for the techniques used in this process.
ISIS has numerous ongoing projects related to applications that can make smartphones even smarter. Some involve creating and improving building blocks of software programs, called middleware. Other projects use the middleware as a jumping-off point. They rely on and share open source systems that make computer code more accessible and easier to use.
Cybersecurity and TRUST
The resilience of today’s software integrated systems depends on more than just combating the wear and tear caused by natural forces, Sztipanovits says. Today corporations, universities, government agencies and individuals have to prepare for cybersecurity issues.
“Now we’re dealing with an intelligent adversary,” he says. “We have to find ways for the system to protect itself.” Sztipanovits leads a variety of cybersecurity projects and is Vanderbilt’s principle investigator with NSF’s TRUST. TRUST partners—Carnegie Mellon, Cornell, Stanford, UC Berkeley and Vanderbilt universities—concentrate on the development of new cybersecurity science and technology.
The imagination and adaptability of ISIS engineers also flourishes in education. Biswas has worked for years with colleagues at Vanderbilt’s Peabody College for Education and Human Development to help students, especially middle schoolers, better learn and understand science.
A recent emphasis has been software-driven teaching aids. Biswas works with colleagues from Stanford University on an NSF-grant project called FACILE (Formal Analysis of Choice-Adaptive Intelligent Learning Environments) that helps students develop learning strategies.
Educators have documented that students learn better when they teach concepts to others, Biswas says. In this case, they will teach interactive computer agents and then use what they themselves have learned to solve challenges that relate to their own experiences, such as how to reduce carbon footprints in schools.
In a different project, Biswas and Peabody’s Associate Professor of Science Education Doug Clark and Assistant Professor of Education Pratim Sengupta are developing new projects where students learn by creating simulations and solving challenges in computer games.
In addition to research, many ISIS investigators are professors or instructors in the School of Engineering, and ISIS projects present opportunities for hands-on learning for engineering students. Currently, ISIS supports 38 graduate students as well as undergrads.
That’s part of the ISIS mission. “What we are doing is fascinating and intellectually challenging,” Sztipanovits says. “We feel all the time that we are at the heart of things. We are part of something big. And we want to attract the best minds to this area of study.”