Yogesh Piolet Thulasidharan

Projects at Michigan State University

Project 1:

Social Wireless Networks

I'm currently working towards a thesis in the area of Social Wireless Networks (also known as Mobile Social Networking). Social Wireless Network is formed by a group of users who have mobile devices such as Smartphones by taking advantage of the Bluetooth technology when it is not being used. There are a lot of potential applications that can be developed by using the Social Wireless Network. As an example, consider the scenario where the users are driving around in their vehicles looking for available parking spaces in a downtown area. A base station in each parking lot periodically sends out information about available parking space. As users move in and out of the parking lot, they receive this information and carry it along with them. As the users move away from the parking lot and come into contact with other users on the road, they spread this information. Thus, the recipient user can make a clever decision on choosing a parking lot.
Duration: Jan, 2011 to current

Project 2:

Conflict Detection and Management of Crew-Members using Wireless Body Area Networks
Sponsor: National Aeronautics and Space Administration (NASA)
Collaborator: Department of Psychology, Michigan State University

This project is the primary source of funding for my Graduate Research Assistantship (GRA). Body Area Network is a type of sensor network where the sensors are attached to the body (either externally or implanted). The sensorcard is mounted on top of a electronic device called Mote which has both processing and communication capabilities. A sociometric badge has been developed which contains the Cricket Mote with sensorcard mounted on top of it. The sociometric badge can measure heartrate of the individual wearing the badge, distance between individuals wearing the badge, acceleration of the individual's motion, sound intensity of the individual, light of the environment and temperature of the environment.
NASA, in association with Dept. of Psychology, intends to detect conflicts among their on-board crew members. This conflict information can be used in selecting the most co-operative and cohesive crew. The system that we have developed can be used to detect conflicts by using psychological clues such as "When two persons fight, their heartrate and sound intensities go high." Also, the distance between individuals and acceleration can be used to detect not only the movement of individuals, but also to determine who moved towards whom. All these information are plotted as time-varying graphs on a well-designed GUI.
Duration: August, 2010 to current

Project 3:

Posture Recognition using Wireless Body Area Networks
Sponsor: National Institute of Health(NIH)
Collaborator: Department of Kinesiology, Michigan State University

I'm actively involved in this project. Body Area Network is a type of sensor network where the sensors are attached to the body (either externally or implanted). The sensorcard is mounted on top of a electronic device called Mote which has both processing and communication capabilities. We have developed a wrist band, an ankle band and a thigh band using the Mica2 mote with a sensorcard mounted on top of it. Using the accelerometer readings from only three positions of the body, we determine the posture of the individual. The posture that we aim to recognize are sitting, standing, lying down, running at various speeds, bicycling, climbing up/down stairs, jogging and other exercise activities.
NIH, in association with Dept. of Kinesiology, intends to use our system to know how much is the energy expenditure for each of these activities. Also, they intend to know how much energy is spent by a common man (of all ages) on each of these activities on a per-day basis. While the Dept. of Kinesiology handles the calculation of energy spent for each posture, our system provides them with details as to how much time is spent by a common man on each of these activities on a per-day basis in a non-intrusive and convenient manner.
Duration: August, 2010 to current

List of Ph.D Course Projects (as part of course curriculum) at Michigan State University

  • Handwritten Digit Recognition Using MNIST Dataset
    [Course: Pattern Recognition]
  • Indoor Positioning System Using Wireless Sensor Network
    [Course: Wireless Networks]

Postgraduate projects at IIT Kharagpur

Current Project :

Identifying Contact Patterns In Intermittently Connected Networks
Project Advisor: Prof. Arobinda Gupta
Thesis: pdf
Half yearly Thesis Progress report: pdf
Half yearly Thesis seminar: pdf
Duration: July, 2009 to May, 2010

Abstract (in simple words): A few strategic Delay Tolerant Network (DTN) routing algorithms use the past contact history for deciding which mobile node must be chosen as an intermediate to transfer a packet to a destination. However, these algorithms have assumed the existence of many knowledge oracles, each of which is capable of answering the questions posed to them. Contact oracles can answer when two nodes are going to meet each other in the future. One of the challenges lying ahead is to determine the existence of patterns in the contact history among nodes and also the period of repetition of these patterns. Another important issue of this research project is that the contact history of all the nodes in the network may not be available to every other node in the case of DTN. Hence, a distributed approach for finding patterns in DTN would be a practical solution to the problem.

    Prior to this, I worked for a few months doing extensive literature survey and practising simulations on Distributed Constraint Satisfaction under the same advisor. We moved from understanding the basics of Distributed CSP to knocking the doors of Dynamic Distributed CSP. This problem still intrigues me.

    Puzzle lovers/game players may find this introductory paper on DCSP interesting :

Makoto Yokoo , Edmund H. Durfee , Toru Ishida , Kazuhiro Kuwabara, The Distributed Constraint Satisfaction Problem: Formalization and Algorithms , IEEE Transactions on Knowledge and Data Engineering, v.10 n.5, p.673-685, September 1998

List of Term Projects (as part of course curriculum)

Undergraduate Projects

  • A LAN monitoring tool (Language:Java,J2ME,JDBC) [Group of 4]
  • Graphics editor (Language:C++) [Teammate: R. Shashanka]
  • A Simple Address Book (Language:C++)
  • Lexical Analyzer (Language:C) [Teammate: R. Shashanka]