Home Research Projects

Research Areas

• Bioimaging

   

  An image is worth a thousand words, especially in medical and biological applications. Nowadays, the imaging techniques and devices available are almost unlimited and the choices depend on the targeted application. Our group collaborates in a broad range of imaging projects: Positron Emission Tomography (PET) to identify molecular changes in malignant tissue at the millimeter scale; massively parallel brain imaging to perform in vivo two-photon imaging of fruit fly brains at the micrometer scale; and cryo-Electron Tomography (cryo-ET) to study structures in whole prokaryotic cells at the nanometer scale.  
  

• Biological modeling

   

  The goal of this project is to understand the dynamic behaviors of complicated biological organisms, which though constructed from fixed genomes, must have the flexibility to survive changing environments and various perturbations. We can exploit similarities between biological networks and engineering circuits to help understand complex biological systems with engineering concepts and tools. The initial work focuses on modeling and analyzing Caulobacter crescentus cell cycle regulation. Caulobacter has been a model system for studying asymmetrical bacterial division.

• Computational photography

   

  Digital photography up to now has mostly just replaced the traditional film with a silicon sensor, without substantial changes to the interface or the capabilities of a still camera.  However, as available computational power to cameras, cell phones, and other mobile systems continues to increase, computation can now be coupled much more tightly with the act of photography. Computational photography is a new area of computer graphics and vision, seeking to create new types of photographs and to allow photographers to capture images they never could before. This involves research both into new software algorithms for fusing data from multiple images, video streams, or other types sensors; into new hardware architectures to capture the data needed for the software; and into new types of user interfaces that allow simple control over the expanded capabilities we hope to discover.

• Microfluidics

   

  Microfluidic analysis systems are emerging as mainstream tools in biology, biochemistry and medicine.  Their sub-nanoliter volumes reduce required sample sizes, while their small feature sizes allow multiple analyses to be integrated into a single device, reducing analysis times. This pattern of increasing device densities and falling per-unit cost mirrors the historical trend in CMOS technology and offers the promise of full-function analysis systems, often termed lab-on-chip (LOC).  Just as solid-state electronics lead to the automation of computation, the goal has been to produce automated analysis systems and real-time analysis in the field.
  
  

  Realization of this promise, however, will require efficient design methods. Our group is working an applying routing and placement algorithms developed for CMOS chips to the design of microfluidic LOCs. We are also interested in applying software abstraction layers to the design problem to allow system designers to concentrate on their end application. In addition, we hope to ease the so-called “world-to-chip” interface problem by enabling better control of the LOC itself. We have developed microfluidic valves that allow us to implement fully static logic in the LOC thus distributing control and reducing the number of required inputs. We are currently investigating the application of this approach to large microfluidic LOC.    
  

• Rethinking Analog Design

   

  As technology scales, more components are being integrated on a chip making systems difficult to model and verify especially at mixed-signal boundaries. Component designs have also grown more complex as more non-idealities become significant rendering traditional design flows, using simplified equations of the transistors, largely broken. 

  This group is part of a larger Rethinking Analog Design effort, which is an industry supported research initiative.  The focus of our group is on "co-development of design tools and methodologies that will help shorten the design cycle while improving portability, documentation and the potential for re-use."  

• Rethinking Digital Design

   

  As the number of transistors on a chip increases exponentially, designs are growing increasingly complex.  Our ability to effectively design and verify extremely high gate count designs lags the rate at which we can add design complexity.    Power is now the limiting factor in chip design.  In effect, there is a need for efficient hardware, i.e., high performance low energy ASICs.  Such efficient designs are currently expensive.  The Rethinking Digital Design team is  pursuing new aproaches to digital design by considering ways to amortize non-recurring design and verification costs.
  

• Smart memories

   

  Smart Memories project is an effort to develop the computing infrastructure for next generation of applications. It is a single chip multi processor system with coarse grain reconfiguration capabilities, for supporting diverse computing models, like speculative multi-threading and streaming architectures. These features allow the system to run a broad range of applications efficiently. Research in this area involves VLSI circuits, Computer Architecture, Compiler, Operating systems and Computer Graphics.

  Web Links

  Smart Memories website
  The Smart Memories internal wiki
  The Smart Memories external wiki (Wikipedia)