Capstone Projects

Project: 1

Application Development in the “Gunrock” GPU Graph Analytics Framework

Sponsor: Prof. John Owens, and postdoc Serban Porumbescu

Description: John Owens’s research group focuses on GPU computing and has a large project on parallel graph analytics called Gunrock. We have a large need for application development on Gunrock, writing interesting graph applications that use our framework (we have a long list of these from our funding agency). We would hope to train you in GPU computing and in using our framework. This could potentially lead to MS thesis opportunities but also could be a shorter project with an option of switching to another group if interested. Funding may be available.

Requirements:
Strong C/C++ skills are a must; experience with parallel computing would be terrific but is not required. We need talented students who can learn quickly and work well both in a group and independently.


Project: 2

Optimizing Application Parameters in the “Gunrock” GPU Graph Analytics Framework

Sponsor: Prof. John Owens, and postdoc Serban Porumbescu

Description: John Owens’s research group focuses on GPU computing and has a large project on parallel graph analytics called Gunrock.  Gunrock applications have a significant number of user- specified parameters. We have historically set those parameters in an ad hoc way. It is challenging for Gunrock developers to find the right parameters. We wish to automate this process, and make it easy for a developer to find the best settings for dataset X run on application Y. At first we hope that a student would automate the search process (this would be more of a project-sized piece of work), but what we really hope is that a student will be able to gain some insight into *why* we want to use particular settings and by so doing, be able to set better defaults (this would be more of a thesis-sized piece of work). We would hope to train you in GPU computing and in using our framework. Funding may be available.

Requirements:
Python is likely the language of choice; C/C++ skills are also desirable. Strong (text) writing skills. Experience with parallel computing would be terrific but is not required. We need talented students who can learn quickly and work well both in a group and independently.


Project: 3

Design and Implementation of an FMCW Radar System

Sponsor: Prof. Xiaoguang Leo Liu

Description: The project focuses on the implementation of a Frequency Modulated Continuous Wave (FMCW) radar system that can perform range, Doppler, and Synthetic Aperture Radar (SAR) measurements. In the first quarter, the students build an FMCW radar system using off-the-shelf components. In the second and the third quarter, the students focus on either improving the system by implementing their own design or utilizing the system for a specific application, such as speed limit enforcement or remote vital sign detection.

Requirements:
Register for EEC 289K in Winter or Spring 2019.


Project: 4

Non-intrusive Detection of Human Motions or Vital Data using WiFi/Bluetooth

SponsorProf. Zhi Ding (ECE)

Description: Wi-Fi networks are widely available for communication purposes. Interestingly, the ability to detect motion and extract vital animal information has also been recently demonstrated. In this project, we shall build a testbed and experiment with the use of WiFI/Bluetooth for motion detection and subject monitoring with the intrusive video camera. It is important to see whether such technologies can detect simple gestures and detect potentially risky behavior or symptoms of illness.

This project contains both software and hardware aspects.

Requirements:
Familiarity with common programming languages such as C, C++, Python.
Courses in Signals, Systems, Communications, and Networking.
Candidates will be interviewed by the Faculty mentor. Ideally, we are looking for a group of 2-3 students.


Project: 5

Implementing Domain Specific Accelerators for RISC-V

SponsorProf. Venkatesh Akella (ECE)

Description: The discipline of computer architecture is witnessing two dramatic, potentially game changing trends, just as the demand for computation is growing exponentially with machine learning and data-driven-everything. First is the end of Moore’s law and the second is the emergence of an open source computer architecture movement (analogous to Linux in the OS community) based on RISC-V. The future of computer architecture will involve creating domain specific architectures centered around a processing core to improve the performance of a set of closely related applications. In this project we will explore designing and programming domain specific accelerators for emerging applications in machine learning and computer security based on RISC-V and prototyping them on a FPGA or the Amazon Cloud.

Requirements:
You should have taken EEC 270 or equivalent (graduate course in computer architecture) and also EEC 298 in Spring 2019.
Ideally, we are looking for a group of 2 or 3 students.


Project: 6

Semi-Autonomous Wall-Climbing Robot for Marking on Vertical Walls

Sponsor:Prof. Saif Islam

External Sponsor (potential): Dr. Aykutlu Dana

Description:  The objective of this project is to create a vehicle that is capable of following a trajectory on a flat surface programmed by the students.  The system would be equipped with robust and reliable navigation system and will host a device to draw preprogrammed linear and curvilinear lines using a marker on the surface at precise positions. Additional device will be used to sense and apply pressure on the drawn lines at varying speed with a goal of erasing the lines. The vehicle is expected to operate both on horizontal as well as vertical flat surfaces. The positioning of the system will be based on remote sensing of the position and orientation of the vehicle using real-time image processing and 3D scene generation. A stereoscopic 3D capture camera (Kinect) along with OpenCV or Matlab code will be used to extract scene and vehicle information. Luminescent markers on the vehicle will be utilized using mathematical framework borrowed from super-resolution localization microscopy to accurately estimate position and trajectory of the vehicle.

The project will involve robot design, sensing, motion planning and control, mapping and localization functionality.  The students will build model using a 3D printer, servos and microcontrollers with wireless connectivity. The students will familiarize with Python libraries, microcontroller programming and sensor interfacing, image processing and feature extraction, real-time feedback and feedforward control, 3D scene generation.

Requirements:
Programming in Python, Matlab, OpenCV.
Register for EEC 289K in Winter or Spring 2019.


Project: 7

Target brain stimulation using surface electrodes

Sponsor: Weijian Yang (ECE)

Description: Delivering electrical field into the brain for stimulation has been shown to be effective to treat depression, stroke, dementia and several other medical conditions. The existing brain electrical stimulation paradigms either rely on electrodes implanted deep into the brain or surface electrodes on the skull. The former approach is highly invasive whereas the latter one lacks a spatial specificity. Recently, a new technology utilizes temporal interference of fields from multiple surface electrode pairs to noninvasively stimulate specific brain regions. In this project, we will optimize the design parameters of such temporal interference system to further increase the spatial specificity of the stimulation region, through finite element method simulation. We will also build a prototype of this electrical stimulation system and test it on rodents.

Requirements:
Electronic circuits, Electromagnetic waves, Matlab. It is a project for 1~2 students.


Project: 8

Time-resolved near-infrared spectroscopy for blood oxygenation measurement

Sponsor: Weijian Yang and Soheil Ghiasi (ECE)

Description: Blood oxygenation is the fraction of oxygen-saturated hemoglobin relative to total hemoglobin (unsaturated + saturated) in the blood. Healthy individual regulates a very precise and specific balance of oxygen in the blood. There is medical significance to monitor oxygen saturation in patients. Near-infrared (NIR) spectroscopy provides a noninvasive approach to conveniently measure the blood oxygenation. In this project, we will study the various approaches of NIR spectroscopy for such measurements. In particularly, we will investigate and develop a time-resolved NIR spectroscopy system, which could not only provide the measurement results from the typical continuous-wave (CW) systems, but also rich information of the tissues under the measurement probe. We will develop the model, perform simulation, explore the components, build and characterize the prototype, and perform in-vitro (and in-vivo) measurements.

Requirements:
Electronic circuits, Basic optics, Matlab. It is a project for 2 students.


Project: 9

Implementation of tunable coherent sources for fast multi-spectral time resolved fluorescence spectroscopy and imaging

SponsorDiego R Yankelevich (ECE)

Description:
            Fluorescence spectroscopy is a powerful technique for in vivo tissue diagnosis. Of particular interest for clinical applications, time and wavelength-resolved fluorescence spectroscopy has proven to be a reliable technique for tissue characterization since it is capable of determining multiple parameters including fluorescence intensity, spectrum, and lifetime.
We propose the development of compact sources of tunable radiation between 400 and 450 nm (violet-blue), to be used in Fluorescent spectroscopy, that are based on cascaded nonlinear wavelength conversion using aperiodically poled lithium tantalite pumped by microchip Q-switch lasers.  The microchip laser monolithic construction, where the laser crystal is directly contacted with the end mirrors, makes them very compact and alignment-free. Such lasers will be pumped with laser diodes either directly or via an optical fiber and will emit pulses of between 5 and 1000 mJ pulse energy, pulsewidths in the nanosecond to sub-nanosecond range and repetition rates from single-shot to hundreds of kHz. Their simple and compact construction, as well the use of widely available inexpensive pump laser diodes, makes them reliable at a fraction of the cost of amplified fiber lasers. The infra-red pulses generated by the micro-chip lasers will be converted to the blue spectral range by using cascaded nonlinear interactions using a single aperiodic nonlinear crystal to obtain optical parametric oscillation and sum-frequency generation.

Requirements:
Undergraduate courses which examine plane wave electromagnetic wave propagation, electrical and electronic circuits.  Interfacing with scientific instruments using LabView or similar platform.  ECE236 and ECE237A concurrently.


Project: 10

Numerical Data Protection Using Numerical-based Codes

Sponsor: Robert Redinbo(ECE)

Description:

Numerical data are represented by digits (usually bits) in a computer word which is stored or transmitted by using a finite-field (e.g., binary ) error- correcting code. The digits are encoded by expanding to a larger group of digits, transmitted or stored individually. However, there exist numerical-based error- correcting codes that accept the numerical data as a group of numerical symbols, expanding them to a larger group of numerical symbols; numbers into numbers. These codeword symbols are ultimately described by a digital format where the defining digits are stored or transmitted in the usual way, but with no additional coding introduced.
When using numerical codes the numbers processed can encounter roundoff noise. Since the data are normally processed after retrieval from coding, there is roundoff noise introduced anyway.
Most previous work on numerical-based coding has concerned rows of a discrete Fourier transform (DFT) matrix. A large class of DFT codes is defined through a matrix structure that dictates the parity-check equations (over numbers). These checks are actually DFT coefficients. These equations provide codewords with separation properties permitting error correction methods to be effective.
Errors are considered as large numerical excursions added to the codewords. The large errors occur only on a few codeword symbols. Error correction requires the erroneous codeword symbols be located followed by determining the actual additive error values (possibly complex numbers).
The only known technique for locating and evaluating error excursions uses a Berlekamp-Massey Algorithm {first discovered for binary error-correcting codes, also applicable to numerical fields}. The algorithm examines the parity-checking symbols associated with a codeword and locates errors by iteratively determining the shortest recursive formula for modeling these observed parity values. At the end of the locating phase, the known parity coefficients are extended followed by an inverse DFT that produces the errors’ numerical values at the proper locations in a codeword. These aspects of numerical-based coding can be a project in itself.
The design of a numerical data protection system and evaluation simulations of its performance can be a complete project. MatLab tools are appropriate for such a project. This would be a “software” project.
An alternate project in a similar vein could be the implementation of a data protection scheme on a stand-alone processor. The numerical data, e.g., floating- point format, is processed within a hardware system (simulation tools are available in the department). This could involve a hardware description language, e.g., Verilog, with the necessary “test bench” verifying overall operations.

Requirements:
Proficiency in Matlab OR Verilog.  EEC 269A/B is desirable.


Project: 11

High Impact Materials and Devices

Sponsor: Prof. Jerry Woodall

1) Photonic and Electronic Devices based on ZnSe Heterojunctions.

Qualifications: An intense interest in theory, analysis and experimental work on new or undeveloped compound semiconductor materials and devices.  Work will require being trained to work in the clean room of the Center for Nano and Micro-scale Manufacturing (CNM2).
Contact:
Prof. Jerry Woodall, Kemper 2001 (jwoodall@ucdavis.edu).
Group Leader: Zongjian Fan

2)Phase change materials systems for storing heat generated by solar and wind power captured by selective absorbing materials.

Qualifications: An intense interest in theory, analysis and experiential work on new or undeveloped materials for capturing solar and wind power, storing it as heat and conversion to electrical power.
Contact:
Prof. Jerry Woodall, Kemper 2001 (jwoodall@ucdavis.edu).
Group Leader: Richard Dering.

3)Photonic and PV Devices based on Liquid Phase Epitaxy.

Qualifications: An intense interest in PV devices made of  AlGaAs and GaP. Work will require being trained to work in the clean room of the Center for Nano and Micro-scale Manufacturing (CNM2).
Contact:

Prof. Jerry Woodall, Kemper 2001 (jwoodall@ucdavis.edu).
Group Leader: Hui-Ying Siao, and Zongjian Fan

4) Splitting water with Al-Ga alloys

Qualifications: An intense interest in theory, analysis and experiential work on new Al-Ga materials for splitting water into hydrogen gas, heat, and alumina.
Contact:
Prof. Jerry Woodall, Kemper 2001 (jwoodall@ucdavis.edu).
Group Leader for Material R&D: Joel Schmierer and Miheer Shah.
Group Leader for Demo Engineering:  Ricky Obregon

5) Ga2O3 based Exploratory Materials and Device Engineering. 

Qualifications: An intense interest in theory, analysis and experiential investigation of a relatively new and unexplored wide gap semiconductor, Ga2O3 for UV-Solar Blind Detector and Thermo-electric Generator (TEG) applications.
Contact:
Prof. Jerry Woodall, Kemper 2001 (jwoodall@ucdavis.edu).
Group Leader: Ryan Bunk