Ionut Constandache

Email: ionut [at] cs [dot] duke [dot] edu Office: D307 LSRC, 214 Hudson Hall Office Phone: (919) 684-1749 Computer Science Department Duke University

Research         Experience         Graduate Work         CV

---

I have graduated in Spring 2011 with a PhD from the Department of Computer Science at Duke University. I moved to California and joined Sun/Oracle in the Hypervisor/Virtual Machine Group.

I am a PhD candidate in the Department of Computer Science at Duke University. I am working under the supervision of Professor Romit Roy Choudhury. I earned my Bachelor of Science degree in Computer Science from University Politehnica of Bucharest, Romania, in September 2005. I joined Duke University in Fall 2006.

Graduate Work

Projects

• Micro-Blog: a virtual information telescope using mobile phones and social participation
  • Human Localization using Mobile Phones: We localize users indoors based on their phones' compass and accelerelerometer sensors. These sensor readings are consolidated into a user trail, which represents the user's path through the indoor environment. Multiple users' trails are intersected and organized in a connectivity graph. Based on this graph a user can be escorted to the location of any other user. Location drift (due to noisy compass and accelerometer readings) is corrected through opportunistic encounters with other users (detected using sound signaling) or with a fixed beacon installed at an arbitrary location in the environment.
  
  
  • Mobile Phone Localization without War-Driving: We propose a mobile phone localization scheme which uses compass and accelerometer sensors and an initial GPS fix to estimate the user subsequent locations. Our solution records a person's walking pattern (displacement and direction) and matches it against possible path signatures generated from a local electronic map (e.g. Google Maps). Electronic maps enable wide coverage and eliminate the need for WiFi infrastructure and expensive war-driving. Our results show localization accuracies comparable to GPS and better energy-efficiency than GPS and WiFi localization.
  
  
  • Logical Localization through Context Sensing: We propose indoor logical localization (e.g. inside Starbucks, Target, in the office etc.) as an alternative to physical localization (GPS, WiFi, or GSM) known to be inaccurate in indoor environments. By sensing sound, light and color, we create a photo-acoustic signature of a location. Further, we use accelerometers to infer broad classes of user-motion dictated by the nature of the user's current location. These attributes are collected through mobile phone sensors and combined in an identifiable fingerprint. Using this ambient fingerprint, the user's logical location can be accurately identified.
  
  
  • Using Mobile Phones to Write in the Air: We propose a new user-mobile phone interface based on hand-waving in the air. Users are allowed to write characters by holding the phone like a pen and gesturing in the air. The user hand movements are classified into circles and lines (based on the phone accelerometer readings), and then organized into characters and words. An air-writing-aware spell checker is used to correct likely errors (e.g. identifying a D instead of a P). If users conform to a few constraints, the phone can reliably decode their air-writing.
  
  
  • Energy-Efficient Localization for Mobile Phones: Localization solution that multiplexes between different localization sensors (GPS, WiFi and GSM) to maximize location accuracy for a given energy budget. The project shows how mobility profiles and population statistics can be used to improve location accuracy.
• ORCA: a system for dynamic on-demand provisioning of shared networked resources.
  • Secure Control of Portable Images: protocols for mutual authentication and secure key exchange between a virtual machine instantiated from a portable image (e.g. a generic virtual appliances with no keys wired in) and the machine's authorized controller (either a virtual computing utility or the VM owner).
  
  
  • Node Agent: web service alowing remote programmatic control and post-install of drivers and software packages in a virtual machine(VM).
• COPSE: a mobile computing testbed
  • VirtualWifi: Xen DOM U device driver allowing wireless network card configuration commands to be pushed and executed in Dom0.

Coursework

CPS210 Operating Systems
CPS212 Distributed Information Systems
CPS214 Computer Networks
CPS220 Advanced Computer Architecture
CPS230 Advanced Algorithms
CPS271 Machine Learning
ECE255 Performance, Reliability and Availability Modeling of Computer Systems
CPS215 Wireless Networking and Mobile Computing
CPS296.2 Emerging Mobile and Distributed Systems
ECE299.5 Discrete Optimization