Home       Academic Experience       


BSEE and  MSEE Degrees

I entered the Electrical Engineering program at the prestigious Shiraz University, Shiraz, Iran, in 1990 following my ranking
among the Top 10 Most Outstanding High School Graduates in the Nation. I received this most distinguished title based on the results of the Iranian National University Entrance Competition. This is a very competitive exam attended by several hundred thousand college applicants each year. In 1990, the number was about 120,000.


I received my engineering degree in 1994 with the citation Highest Honors for an outstanding grade-point average and completion of an intellectually-challenging curriculum. I joined the graduate program in the same institution and received my Master's degree in the summer of 1997. My thesis was written in English and was titled ``An Adaptive Approach to Digital Filter Bank Design''. For this thesis, I received the maximum numerical mark possible (20 out of 20) and the citation of Highest Distinction for completion of original research of exceptional quality.


PhD Degree

Upon finishing my Master's degree, I applied to and received admission offers (with full financial support) from several
top-ranking universities in the United States and Canada including Caltech, Georgia Tech, University of Waterloo and University of Toronto. For personal reasons, I decided to do my PhD in Canada so I accepted the offer from University of Toronto which is the largest and the most reputable university in Canada. I started my PhD under the supervision of world-renowned systems and control theorist Professor Bruce A. Francis in January 1998.

During my PhD research, I developed a fundamental statistical theory for information processing in multirate linear systems. The main theme of the theory to infer sample values or spectral properties of a high-resolution signal low-resolution observations. It also provides a means for quantifying the amount of statistical information gained through each observation, as well as a precise mathematical definition for optimal multi-resolution signal decomposition. The main theory consists of four interconnected sub-theories as described below.

  • Multirate Statistical Inference: uses the Maximum Entropy principle to estimate the spectrum of a high-resolution random signal given statistical properties of low-resolution observations.

  • Information Theory of Multirate Systems: provides a measure for the quantity of information gained about a non-observable high-resolution signal through low-resolution statistical measurements.

  • Multirate Signal Estimation: takes sample values of given low-resolution observations and estimates sample values of the non-observable high-resolution signal under measurement.

  • Algebraic Theory of Optimal Multirate Systems: provides a rigorous framework for comparing different multi-resolution measurement systems in terms of scalability. Furthermore, this theory clarifies and generalizes several existing notions of optimality for multirate filter banks.


Post Doctoral Fellowship

Upon completing my PhD degree, I joined the very young and talented University of Toronto Professor Parham Aarabi in his then newly established Artificial Perception Laboratory (APL).

In my capacity as a Post-Doctoral Fellow in APL, I applied the theory I had developed to several novel signal processing applications. These applications included fusing the data recorded by a low-resolution microphone array, time delay of arrival (TDOA) estimation between multirate signals, phase-based speech enhancement in microphone arrays, and sensor validation.

I also became interested in sensor networks and the problem of merging statistical information in a distributed network of sensors. While working on this problem, I developed a very general class of algorithms for distributed signal processing in sensor networks. The results have been published in several papers and in Chapter 38 of the Embedded Systems Handbook published by CRC Press in 2005.

A complete account of my research contributions to the fields of multirate signal processing and sensor networks will be presented in more detailed form in my book ``Multirate Statistical Signal Processing'' which is due to be published by Springer-Verlag in 2006.



1. Teaching philosophy:

I believe that the purpose of teaching is more than just communicating facts and developing basic skills. The goal of teaching is to develop students' ability to make critical judgments of ideas and approaches. In serving this goal, I have found that one must constantly challenge the students and force them to take a firm stance for their beliefs.

I like to introduce the core of course material and then let arguments flow from students' interaction. This gives the students
the opportunity to explore different perspectives which, in turn, helps them develop the problem solving skills they need to participate in engineering practice or academic research.

During my two-year tenure as a Post Doctoral Fellow at the Artificial Perception Laboratory, University of Toronto, I had the
opportunity to directly advise several undergraduate and graduate students' research. I discovered that students need assistance not only with technical direction, but also in identifying and prioritizing goals, developing and executing research plans, and
effectively communicating results. Students must be allowed to grow into the research, and hence one must constantly recalibrate the amount and substance of direction. I see a doctoral program as being successful when the student no longer needs direction.

2. Teaching experience: 

I taught two undergraduate courses at the University of Toronto:

  • MIE346: "Analog and Digital Electronics for Mechatronics'' in 2001

  • ECE231: "Introductory Electronics'' in 2003 and 2004.

I was the sole instructor for the first course. I coordinated the lectures, labs and tutorial sessions with the help of four teaching assistants. The second course was a multi-section one with 3 instructors, 21 teaching/lab assistants and more than 400 students. I am very pleased to say that at the end, I received one of the highest teaching evaluations of any instructor/professor in the entire Department. Thanks to this success, the Department offered me a lectureship position for the 2004 Spring term as well.


Research Interests

I am interested in developing distributed information processing algorithms for sensor network, RFID, and biometric identification applications. This is an ambitious goal demanding a multidisciplinary research program and a merger among several established disciplines such as digital communications, signal processing and network computing.

1. Signal processing for sensor networks:

A network of sensors can yield a rich multi-modal stream of sensory data. Furthermore, it can be highly scalable, cost effective and robust with respect to failure of individual nodes. However, realizing the full potential of large distributed sensor systems requires fundamental advances in the theory of distributed data fusion. For instance, the network must support both high- and low-end data transmission and processing, while allowing for interpretation, modelling, and fusion of sensed information.

A primary goal will be the development of signal processing algorithms designed expressly for sensor network applications. More specifically, I will focus on the following three fundamental problems: scalable fusion algorithms for sensor networks, multi-resolution transmission of sensory information in large networks, and localization and self-calibration of sensor nodes.

2. Extending RFID to sensing applications:

Radio Frequency Identification (RFID) is an emerging identification method that relies on storing and remotely retrieving data using devices called RFID tags. An RFID tag is a small object that can be attached to or incorporated into a product, animal, or person. RFID tags are passive devices containing a silicon chip and antennas which enable them to receive and respond to radio-frequency queries from an RFID reader.

RFID tags require no battery, and thus no maintenance. In view of this, they have the potential to be used as environmental sensors on an unprecedented scale. However, there are many engineering challenges that must be addressed before RFID sensing becomes economically viable. I am very interested in signal processing aspects of RFID sensing. Sample research topics include accurate tag localization, simultaneous reading of multiple tags, and data fusion in distributed reader/tag networks.

3. Scalable biometric identification systems :

A biometric is a physiological or behavioral characteristic of a human being that can distinguish one person from another. Examples include fingerprint, iris scan, face geometry, etc. A biometric verification system is a system that compares  "matches'' a previously stored template with the biometric provided. The result of this comparison is either an "accept'' or "reject'' decision. A biometric identification system, on the other hand, tries to find a match between the provided biometric and a set of templates stored in its database.

As biometric identification system begin to be deployed in large scale applications, the number of enrolled templates grow as well. This, unfortunately, leads to increased risks of false acceptance and false rejection. Furthermore, the search time becomes prohibitively long too. It is very important, therefore, to design scalable identification systems where a smart search instead of the usual exhaustive database search is used. This is a very challenging research area with deep connections to information theory, pattern recognition and distributed algorithms. An algorithmic solution to this fundamental issue could result in exciting new possibilities for biometric technology.


- Copyright Omid Jahromi 2004 -