Introduction/Overview

Introduction

Hello! My name is Vishal Kotcherlakota. I am a second-year Electrical Engineering major in Sixth College here at UCSD. My advisor is Professor Massimo Franceschetti. I work with his graduate student, Paolo Minero. Both are affiliated with the Electrical and Computer Engineering Department at UCSD, as well as the California Institute for Telecommunications and Information Technology (Cal-(IT)²).

Project Summary

My project proposal is available for download at http://ieng9.ucsd.edu/~vkotcher/PDFs/proposal.pdf.

My project is about acoustic localization. I’ll briefly describe the project here, using the diagram below.

In our test room, I have placed microphone clusters (shown in blue). This diagram shows three, but in my experiments that number may vary. The clusters pick up sound from a source (shown in red). Each cluster analyzes the signal by using the time difference of arrival (TDOA) between the various microphones in the cluster. The cluster will then create a vector (shown in green) that points towards the location of the source. In a perfect world, all the vectors will intersect, and the point of that intersection will be the coordinates of the source (shown in black). In reality, this will probably be the point closest to all vectors, and it will have some sort of tolerance measure.

My network of clusters will be distributive. This means that each cluster will communicate with all other clusters and collaboratively find the location of the source. I will set up a system where each cluster is connected to a windows-based laptop computer with a wireless card, running MATLAB with networking packages. On each computer will be a MATLAB m-File which will get input from the microphone cluster, create the TDOA vector, and provide a means for communicating with other clusters in the experiment. We will then compare our data to data for an existing method where all microphone clusters communicate with a single computer (a centralized method).

The applications of acoustic localization are boundless, but will be discussed in a later post in more detail. For more details about the project itself, refer to the project proposal. I will be publishing more about the project in the coming weeks.

Goals

My goals relating to this project are as follows.

• Learn the TDOA algorithm.
• Generate MATLAB m-Files for the TDOA algorithm and the system as a whole. Create test benches to simulate each step in our process.
• Perform a simulation of the experiment locally on a single machine, then across a wireless network.
• Build a “test room”. “Room” should be flexible in size and configuration.
• Experiment with multiple configurations of the “room” and vary the number of clusters. Determine a standard of performance and measure experiments with this standard.
• Compare all distributive network findings to centralized network findings.
• Document all findings and publish a paper and poster with the research findings.

  My personal goals are as follows:

• Learn how to work without external initiative. All my work up to this point has been assigned to me, with defined steps I must complete. This time, I need to frame the problem appropriately before I can attempt to solve it.
• Get to know my graduate student and faculty advisor well. I want to continue working with these individuals into next year, and would like to establish a good working relationship with both of them.
• Learn more about research processes and protocols, and academia as a whole. It is my intention to go into industry after I graduate, but I am very young in my career, and want to take every opportunity to define my direction and gain experience.
• Have fun and enjoy a rewarding summer experience.

  Challenges
  

  I see the following potential challenge.Using the microphones to measure TDOA. Measuring with two microphones is easy enough, but what about three or four?

• Learning the in’s and out’s of MATLAB. Specifically;
  • Calling m-Files from m-Files
  • Using network packages to communicate with other machines.
  • Collecting and saving data in files on disk and processing results automatically.
  • Debugging.
• Studying any signal processing procedures which relate to this topic.
• Learning how to document my work correctly.
• Writing the paper/giving the presentation.

  Conclusion
  

  This seems like enough for a first post. I look forward to completing this project and enjoying a great summer in San Diego. See you all later!