Sept. 11, 2007 -- This summer, a small announcement from the Federal Reserve was duly noted in the business section of many daily newspapers. The Fed declared its intention to consolidate most check-processing operations into four regional centers and to cut 1,740 positions around the country. With more and more people making payments electronically and banks required only to include images of cancelled checks in their customers’ statements, the check processor is now going the way of the switchboard operator and others whose work in a large network can be accomplished much more efficiently by a computer.
This small change sheds light on a fundamental shift in the infrastructure that binds our society together. Today, as never before, these networks are almost exclusively electronic. The services we enjoy and have come to depend on — the power that lights our homes and drives our factories, the rapid transfer of funds that sustains the global economy, and the delivery of information by telephone, television and Internet — all are made possible by immensely complex, far-flung computer networks.
While these electronic networks can deliver services on a scale and with a level of sophistication never before possible, our increasing reliance on them requires us to take every precaution to ensure they function as intended and without interruption. “Because we depend on them so significantly, we must be convinced that they are sufficiently protected,” says John Knight, a University of Virginia professor of computer science.
To date, there have been two main approaches to protecting networks from being compromised. The first is to wall them off, a difficult proposition because networks are highly decentralized, making isolating them from attack virtually impossible. The second is to build networks that still function even if breached. “This situation has effectively created an arms race, where network designers are constantly working to stay ahead of network adversaries,” says Knight.
Knight has assembled a team of researchers from the University of Virginia, the University of California-Davis, the University of California-Santa Barbara, and the University of New Mexico to develop radically new ways to safeguard these networks.
The magnitude of the task they set themselves is captured by the title of the project — Helix: A Self-Regenerative Architecture for the Incorruptible Enterprise. The Department of Defense, in awarding the group a highly coveted Multidisciplinary University Research Initiative award, believes that they are equal to the task. The award will provide up to $4.6 million over five years to support their research. There were scores of preliminary proposals for each of the 36 MURI awards funded in 2007 — and Knight’s grant was one of three organized by researchers at the School of Engineering and Applied Science at U.Va.
“This is an important issue for the Department of Defense,” Knight says. “In recent years, the department has been moving to a concept of network-centric warfare in which real-time information from a wide variety of sources is available to soldiers in the field and to decision-makers.”
Knight and his colleagues have devised a combination of strategies to overcome the weaknesses of current approaches. They plan to create networks that dynamically and continuously alter their “attack surface,” the interface that adversaries encounter when trying to break in. This in itself will dramatically increase the difficulty of intrusion. Their second strategy is to make the network self-healing in the event that an attacker does get through. The challenge that Knight and his colleagues face is to create tools that will enable the network to sense that it has been breached, to identify the vulnerability that enabled the attacker to penetrate the network, to develop a response to eliminate the vulnerability, and to distribute and install this solution universally throughout the network without human intervention. As Anh Nguyen-Tuong, a co-principal investigator on the project and a senior scientist in the Department of Computer Science, points out, “You use the attack as a testing tool to pinpoint bugs.”
This is an enormously difficult technical challenge, but Knight stresses that his colleagues on the project are among the leaders in the field. “We are bringing together world-class expertise in biologically inspired security, program analysis, software engineering, and compiler and virtual machine functioning,” he says. “We have long-standing, productive working relationships with our colleagues at Santa Barbara, Davis, and New Mexico."
In addition to Knight and Nguyen-Tuong, U.Va. faculty members Jack Davidson, David Evans, and Westley Weimer are co-principal investigators on the initiative.