Showcase of ECE Undergraduate Research Projects

To help gauge the health of ecosystems on coral reefs, we developed a proof-of-concept prototype of an underwater camera module to automatically census coral reef fish populations. Recognition challenges include pose and lighting variations, complicated backgrounds, within-species color variations and small within-family differences between species (Fig. 1). During an early stage of the program, seniors Lev Barinov, Jigesh Baxi, and James Bibby carried out a project involving an epipolar analysis for the special case two cameras at a 45 degree angle looking through the ½” thick polycarbonate walls of an aquarium to determine the depth, and pose of free-swimming fish and disambiguate overlapping fish. At a later stage of the program, when a fully submersible prototype was tested at the New York Aquaium (Fig. 2) Edward Pavoni, Andrew Rapport, and Serena Tsang, carried out a Capstone project involving analysis of similar species that could be confused when overall shape and color features were insufficient for distinguishing them from each other. Using localalized pattern and color features within the similar species improved the recognition rate significantly. During the course of the project ECE graduate students Chetan Tonde and Ganesh Sundar helped in mentoring the undergraduate researchers. All of these students are listed as co-authors in a paper to be delivered in August at the SPIE "Applications of Digital Image Processing XXXV" conference in August, 2012.

The objective of this project was to create a mobile application hat will intelligently decide the most efficient order of tasks one should complete based on user requirements. Seetha first implemented a basic Android application to understand the accuracy of location services and how to best present information to the user. Server-side computations were made to easily group a list of tasks and provide the most efficient order and location for these services. Seetha used the Google Places API and Google Directions API to make the calculations.

Seetha Annamraju graduated with a Bachelor's degree in ECE with the class of 2012. Seetha worked with Prof. Janne Lindqvist since Fall 2011. During fall 2012, Seetha will start studies towards a Master's degree at Carnegie Mellon's Information Networking institute in Mobility.

Tele-rehabilitation is the provision of therapy at a distance, for example for patients living in rural areas. A young victim of severe trauma to the head trained on custom video games he payed on a PlayStation 3 while wearing a sensing glove. The technology developed at Rutgers allowed monitoring of progress (finger range of motion increase, speed of finger flexing, cumulative exercise time) from the Institute in New Jersey, while the patient was exercising at home in Indiana. Richard analyzed the game data so to extract objective values based on measures of glove sensitivity and repeatability.

Subsequently he graphed the data, which more clearly visualized progress over the 6 months of home therapy. Finally Richards findings were included in the paper "Long-Term Hand Tele-Rehabilitation on the PlayStation 3: Benefits and Challenges," Burdea G, Jain A, Rabin B, Pellosie R and Golomb G, 33rd Annual International Conference of the IEEE EMBS, pp. 1835-1838, 2011.

The design and development of fault-tolerant and resilient distributed applications and systems is a challenging undertaking. Multiple factors contribute to the challenges, including (i) heterogeneity of distributed systems, (ii) inconsistent middleware semantics and capabilities and (iii) large number of possible dynamical configurations. It is important that distributed applications have resilience against each of these attributes.

In this project, we will design and implement tests to guard against deployment and runtime errors of system middleware. Specifically we will enable the reliable deployment of SAGA-Python (Bliss) for a range of NSF funded infrastructure -- XSEDE and FutureGrid systems.

Over the past year, Aditya has been exploring various aspects of virtualization and cloud computing. He developed a layered cloud federation platform to provide BLAST-as-a-service, and is focused on developing delegation and federation policies for cloud platforms and deploying virtual machine images of CloudBlast, a Hadoop implementation of BLAST. He also explored the viability of scientific computing in an HPC-Cloud environment provided by FutureGrid and deployed the Weather Research Forecasting (WRF), IPARS Oil Reservoir Simulator and the Hadoop BLAST application on FutureGrid. He is currently working on developing an asynchronous replica exchange algorithm for protein docking and drug design on heterogeneous federated cloud infrastructure integrating public clouds with NSF XSEDE, OSG and DOE platforms, and is exploring methods for optimizing asynchronous communication in such an environment.

Ubiquitous computing is the notion of having computers blend into the background and being able to perform computations using information that is gathered from an environment without a lot of user interaction. Different sensor technologies such as WiFi and RFID are widely available that can help us gather information about an environment. This information can then be used in ubiquitous computing. Other systems have been developed to use these technologies but have not been successful in providing a platform for application development. GRAIL provides an infrastructure that converts raw sensor data such as signal strength and on/off states to real-world properties, such as temperature and location, which developers with little experience can use to build ubiquitous computing applications.

I created several applications using the GRAIL system in order to evaluate the usability of the system and its potential as a platform for developers to use to easily create ubiquitous applications.

Jonathan Chiu was in the Class of 2012 and pursued a major in ECE and a minor in Computer Science. He has presented at the Aresty Undergraduate Research Symposium in 2011 and 2012 and has been working under Dr. Martin since Fall 2010.

Blind signal separation is the process of separating signals from one another by observing only their mixtures. This procedure can be applied to image and sound processing, communications, and to biomedical signals such as EEG signals.

There are various methods of accomplishing blind signal separation. Independent component analysis (ICA) is one method that exploits the non-Gaussianity of independent signals. The JADE algorithm is an implementation of ICA based on the joint diagonalization of matrices.

I tested the effectiveness of the JADE algorithm on various types of signals and developed a method intended to correct the permutation and amplitude ambiguity present in ICA. I also attempted to window non-stationary signals into stationary parts so that the signals may be properly separated by the JADE algorithm.

Theresa Lye was in the Class of 2013 and pursued a major in ECE and a minor in Computer Science. She has been working under Dr. Petropulu since Fall 2010

We studied the efficient, simultaneous communication of many users/devices sharing a common frequency spectrum. Modern radio communication relies on assigning unique signals to users. Short signals allow efficient transmission. Large sets of signals allow many users at once. Sets of signals with low cross-correlation minimize interference among transmissions. Our work was to better describe the tradeoff among these goals. For a given signal length, we found a bound on the size of signal sets which are interference-optimal according to the Welch Bound, and for large sets we found a tighter bound on interference.

Marcus, J.; Budisin S.; Spasojevic, P.;
"On the size of binary MWBE sequence sets," Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks (WiOpt), 2011 International Symposium on , vol., no., pp.374, 9-13 May 2011
doi: 10.1109/WIOPT.2011.5930045

John Marcus was a Class of 2012 double major in ECE and Mathematics. He has studied coding theory and communications under Professor Spasojevic since Summer 2010, presented a paper at WiOpt in May 2011, and is continuing research in the field.