Innovative Projects Realized

The objective of this project is to design and implement digitally controlled high efficiency DC-DC switching converters suitable for VLSI power management applications. The main challenge in this work is to develop a robust power management scheme to dynamically vary the supply voltage as a function of varying clock frequency for the VLSI chip. The purpose of this scheme is to provide supply voltages to the VLSI chip that is just sufficient to meet data processing speed requirement while minimizing the DC and MOS gate leakage (in deep sub-micron devices) power dissipation, thereby maintain optimized power consumption for all processing requirements. In addition, this project will also require the design and implementation of energy and area efficient on-chip voltage and temperature monitoring circuits to provide important information on the operating conditions to the power management system.

Our laboratory is interested in developing imaging software and instrumentation that will allow us to measure the spatial dynamics of growing neuronal structures in a small, transparent, model organism called C. elegans. C. elegans is a small nematode (round worm) with a uniquely simple nervous system—it contains only 302 neurons and the connectivity between these neurons has been fully mapped out. Since the worm is a popular model organism, there are a vast arrays of molecular tools to fluorescently label individual neurons so one can map out is structural design and even read out and modulate its activity. We have several experiments in progress to measure the dynamical movement of neurons and also their activity in C. elegans at the network level. The eventual goal is to be able to reconstruct the neuronal dynamics that underlie universal behaviors like sensory measurement and response in holistic fashion. One example is as follows:

Depending on the student, the project can lean towards computation, experiment, or a combination of both. Students interested in learning microscopy, some neurobiology, some microfluidics, will be trained so they can work independently on developing a protocol to get time-lapse images of neuronal development. Students interested in learning computational data analysis techniques will be trained to perform image processing programming using MATLAB or LabVIEW. It is also possible for students to work on both the computational and experimental side of this project.  

Intensity-modulated radiation therapy (IMRT) is an advanced cancer treatment technology that uses beams of high energy x-rays to deliver radiation to a tumour.  In IMRT, radiation beams are divided into many small beamlets, whose intensities are individually adjusted in order to deliver a dose of radiation that conforms to the shape of the tumour.  The intensities of each radiation beamlet are computed using specialized software.  In this software, optimization is the primary technique driving the computations.  That is, the treatment planning problem is modeled as a mathematical optimization problem, and then solved using mathematical algorithms. In this project, we will design and test an optimization methodology aimed at determining weights for the IMRT treatment planning problem in an automated and objective manner.  The specific optimization methodology we will use is called “inverse optimization”.  The standard or “forward” problem involves choosing weights, and solving the optimization problem in order to come up with a treatment.  In inverse optimization, we input the treatment, and the optimization gives us the weights.  We will use treatments of prostate cancer patients from Princess Margaret Hospital (PMH), one of the largest cancer hospitals in the world (and based right next to the University of Toronto campus), as input to our optimization models.

We expect this research to lead to an improved understanding of the “best” weights to use in treatment planning for prostate cancer.  We intend to show that by using optimized weights in treatment planning, we can design treatments that are almost identical to clinically designed treatments, without following a time-consuming trial and error process.  Ultimately, we hope to improve the treatment planning process by providing increased automation where possible.  We will also study the effects of different anatomical geometries on the determination of optimized weights for different patient classes.

Our project consists of a combination of magnetic resonance imaging (MRI) and numerical simulations to model the blood flow dynamics in major cerebral arteries. In doing so, we hope to determine the shear stresses that are inflicted on the vessel walls of the brain, which should correlate with the vessel’s ability to dilate in response to a vasoactive stimulus. This autoregulatory effect can be measured in terms of cerebrovascular reactivity and is shown to be a useful biomarker for cerebral disease.

The experiment protocol consists of a series of MRI scans to determine the flow velocity and spatial-temporal profile at target locations within the circle of Willis, which connects all the major supply arteries of the brain. Data from these scans will be imported into a simulation program that utilizes computational fluid dynamics (CFD) to predict blood flow properties for the entire circulatory system in the brain.

The selected student will primarily be responsible for developing a protocol for translating the MR data acquired from healthy volunteers into input values that can be accepted into the CFD simulation. Specific duties include, but are not limited to, vessel segmentation and edge detection, noise error simulation, source code debugging, and anatomical and flow data modeling. Literature research will also be an important component of the project as the student will need to be knowledgeable of past and current publications on this subject.

Graphical Processing Units (GPU)’s have matured, and now provide the highest computational performance per dollar.  To be scalable, a tighly coupled cluster of GPU’s must support efficient implementations of real world algorithms.

This project will implement a software signal processing pipeline using GPU architectures.  CITA is currently acquiring a major GPU cluster, which will be ready for scientific use this winter.  This technology, if successful, has the potential of bringing the information revolution to astronomical signal processing, specifically interferometric aperture synthesis imaging.  At the heart of an array of radio telescopes is the correlator, which combines the signals from all pairs of antennas.  Currently, they are primarily implemented in hard coded, application-specific hardware, and the cost and rigidity of such systems is a major limitation of radio astronomy.

Software implementations of DSP systems have the benefits of very high flexibility, and are able to benefit from the latest developments in Moore’s law.  Astronomical hardware development efforts often require 10 years from design to deployment, with a corresponding 10 year lag in performance.   An efficient GPU-based code could bring the cost of correlators down by more than a factor of 10.

The student will be implementing a parallel cross-correlation pipeline for use at various radio telescopes, including the Canadian Hydrogen Intensity Mapping Experiment (CHIME), and the Indian Giant Metrewave Radio Telesope (GMRT).

This is a major long-term and on-going project being undertaken by CanAssist. An organization dedicated to developing assistive technologies and services that will improve the quality of life and independence of those with disabilities (please see www.canassist.ca). One of our key research areas revolves around the development and provision of communication and control devices for people who have acute disabilities and for whom there are no “market solutions”. These range from the complex (for example, those based on eye-tracking or the detection of electromyographic (EMG) signals) to the relatively simple (laser-pointer detection systems).

Our goal is to make the communication program much more dynamic (i.e. predictive) than current programs. Additionally, it will be contextually responsive (i.e. there could be direct links to calendars/schedules/photo galleries etc.). The overall goal is to make AAC much more efficient and easier to use. The work will build on our considerable experience in this area and will be carried out in collaboration with a number of agencies that provide service to clients-children and adults. These agencies include, Special Education Technology BC (SETBC) as well as the South Island Distance School (SIDES). Furthermore, we will be working directly with industry (Ablenet) to bring this system to market

Students will be part of a highly interdisciplinary team and would be required to work with experts from a wide range of disciplines- including mathematics, linguistics, computer science, electrical engineering and psychology. They will also work directly with clients and their caregivers and health care professionals. Depending on the area of specialty, students would have to write code or develop hardware/firmware that would be used directly by clients.  They would take a lead role in testing and evaluating their product. They would also be required to provide excellent and comprehensive documentation and present their work at internal and external workshops/conferences. Since a major goal is to bring our devices to market, students would be expected to interact and collaborate closely with our industrial partners and any of the many agencies with whom we work.

Single chip multi-core processors have become the de-facto computational platform for application domains ranging from low-power mobile and embedded devices to high-end server products that form the backbone of the worldwide information technology (IT) infrastructure. Power management has always been an issue in the mobile computing space, and with the alarmingly increasing rates of energy consumption in data centers (an estimated 100 billion kWH by 2011), it has become a critical need in high-end server market as well. To provide fine-grained power management capabilities, multi-core systems are increasingly being instrumented with a number of on-chip control knobs that can be used to dynamically vary the power-performance characteristics of the chip.

Mathematical modelling is the process of creating mathematical models describing the behavior of physical systems. Mathematical models of many physical systems (e.g., engineering, biological, economical, and environmental) are governed by ordinary differential equations [1].  These differential equations are usually nonlinear, and parameters might appear both linearly and nonlinearly in these equations.  Delay differential equations are differential equations, where some delays exist in the forcing function; these delays are typical in engineering due to the inherent time lag between the sensing and control signals [2].

Recently we have proposed a homotopy optimization methodology [3] based on gradient algorithms to find the global minimum in certain problems of parameter identification for ordinary differential equations.  We exploit symbolic computations by using MAPLE for efficient generation of sensitivity equations, which are required for optimization.   In homotopy methods [4], the objective function to be minimized is modified by adding another function whose optimum is known, and a morphing parameter is used to transform the modified function into the original objective function. A series of optimizations is performed while slowly varying the morphing parameter until the modified function is transformed back into the original objective function and during this process we obtain the global minimum.  This project involves developing morphing functions for delay differential equations (this work is closely related to observer design for delay differential equations [5]). The morphing function will be used in the homotopy optimization procedure for identifying the parameters for delay differential equations.

The student will be involved in programming and testing some of the algorithms that are being developed in our group.

In current NOX chemistry models developed for catalytic converters, the focus is on the precious metal (Pt, Rh and Pd) component activity and chemistry over the precious metal site. Any chemistry associated with the support material is ignored. However, in modeling both literature data and results obtained from preliminary work obtained at UW, there are indications that the support material plays a significant role in some steps of the chemistry. Since the literature does not contain results focused on the support role, we propose to investigate at a fundamental level the ongoing NOX chemistry on key support materials and then the effects of adding the active metal species. In approaching this in a step-wise manner we can isolate the impacts of each component and determine their relative significance.

We propose to fabricate six sets of catalyst samples. After this stage, platinum will be added to each and the same chemistry studied. Via comparing the NO oxidation and reduction data with and without Pt, the conditions where the support oxide chemistry will be relevant can be evaluated and the support chemistry isolated. Lastly, infra-red spectroscopy analysis (DRIFTS) will be used to analyze the chemical states of the NOX species interacting with the materials as a function of temperature. Therefore, for 6 catalysts, NO and NO2 adsorption, TPD after adsorption and NO oxidation and reduction with H2 as a function of temperature will be evaluated.

This work is being performed in collaboration with an industrial partner, a Tier I automobile manufacturer with a significant research presence in India. The data obtained will be directly provided to the vehicle manufacturer as input for their modeling efforts. The student will design and perform the experiments described above. In order to accomplish the objectives, the student will learn to operate the Catlab micro-reactor system, learn the fundamentals of mass spectrometry (MS), learn to analyze MS data, learn to operate a DRIFTS reactor and learn to analyze DRITS data. This will also require substantial literature review for the DRIFTS analysis, so that the data obtained can be readily analyzed.  Prior to the experiments being performed, the student is also required to author and submit a safety report focused on the reactors to be used, including the gases and their potential hazards.