- Battery-based detection gives warning of attacks on mobile devices
- Exploring the hardware/software boundary
- Speeding up supercomputers to study human viruses
- Reconfiguring for bioinformatics
Virginia Tech computer engineers have devised an early-warning system to protect mobile electronic devices and their networks from malicious attacks. The system, which works on equipment such as cell phones, PDAs and handPCs, is based on the behavior of a battery when a mobile device is under attack.
Grant jacoby tests a battery-based intrusion detector for mobile devices
“A new generation of hackers is emerging who specialize in disrupting and hijacking wireless communications of PDAs and smart phones, which, by default, are not configured to be secure,” said Grant Jacoby, who developed the system with his Ph.D. advisor Nathaniel Davis.
He described how some worms deplete batteries by constantly scanning for Bluetooth-enabled devices, and how other attacks can be targeted specifically to consume resources which, in effect, drain batteries and compromise performance. “In the commercial and military sectors, these accelerated battery depletion activities can trigger mission failure and loss of revenue” he said.
The Tech system operates through a host intrusion detection engine (HIDE) that monitors power behavior and notes consumption irregularities. A scan port intrusion engine (SPIE) determines the IP source address and port, as well as the energy signatures of the attack. The system then correlates and compares the data to a variety of the most common attacks. This process could also alert security administrators earlier of an ongoing attack than current means by using mobile devices in this fashion as sensors.
The goal of any detection algorithm for mobile computing is to identify attempts before they are successful and not tap too much performance or memory in the process, according to Jacoby. “The question becomes how much power to expend in the effort to protect the device. This is why we developed a software-based solution, since embedded systems are not only expensive, but also make significant energy constraints on these small devices.”
Computer engineers work at the interface of hardware and software — and Bradley Fellow Neil Steiner wants to define that interface; its boundaries, properties and behavior; and ultimately identify a unified model. Just as electromagnetics relies on boundary conditions, Steiner questions whether boundary conditions influence hardware/software interactions.
He defines hardware as tangible and able to function on its own — whether it is digital, quantum, biological, or optical. Soft-ware is a form of information: intangible, structured, but unable to function on its own. He is intrigued by the energy/information relationships described by IBM researcher Rolf Landauer, and the possibility that a larger structure awaits below the surface.
Steiner proposes that hardware and software may be interchangeable, paralleling each other and sharing similar topologies. “Is an FPGA configuration hardware or software? Or perhaps, potential hardware?” Software emulation of hardware also blurs the boundaries, he acknowledged. “What do we really know about this interface and its boundaries? Do we know what ‘forces’ come into play? Are the boundaries fixed? Smooth? Malleable?” The questions are important because “engineers are great at finding ways to exploit new properties and behaviors,” he says.
Mark Jones and Paul Plassmann have developed a toolbox of scalable algorithms and software for complex simulations that reduce CPU time on massively parallel computers, such as Virginia Tech’s System X. The toolbox, which has been successfully used in large-scale combustion calculations, has obtained computational speed ups of 1,000 times. Working with Karen Duca of the Virginia Bioinformatics Institute, the team is applying the toolbox to understand the Epstein-Barr Virus (EBV) in humans. The team is linking simulations representing the interaction of EBV and lymphocytes (B and T cells) with the underlying biochemical reactions networks regarding the body’s tonsil system.
Public production of genomic sequence data has been doubling annually for the past decade, presenting a challenging puzzle to unravel. Peter Athanas and researchers at the Virginia Bioinformatics Institute are investigating relationships that exist between multiple sequences of amino acids to detect variability in a family of proteins and as a step in understanding relationships between organisms. The computing time for multiple sequence alignment can be lengthy, yet can be reduced by orders of magnitude using specialized tools created in the Configurable Computing Lab. By providing dramatic speedup, tools like this may enable researchers to better understand protein construction and evolutionary factors. Shown above is the construction of a Ras protein — a molecular switch — built from amino acid components.