Research

---

I am a PhD candidate in the Computer Systems Lab in the department of Computer Science at Rice University. I am a member of the Monarch group at Rice University. I expect to complete my graduate studies by Fall 2006. These are my fields of immediate interest:

• Wireless Mesh networks
• Mobile Ad Hoc Networks (MANET)
• Mobile and Wireless Communications.

Presently, I am working with Dr. David B. Johnson on the Transit Access Points (TAPS) project. (Fall 2004 - current). The goal of this project is to provide a high-performance, scalable, and widely deployed wireless Internet. In particular, I am working on Coverage Improvement and Traffic Aware Routing in Wireless Mesh Networks.

Some similar projects are:

• Self-Organizing Neighborhood Wireless Mesh Networks at Microsoft Research
• Multi-Hop Mesh Networks at Intel Research.

---

Here is a chronological listing of some earlier projects that I have worked on:

• PhD. Dissertation: Throughput and Coverage Improvement in Wireless Mesh Networks (Fall 2004 - Fall 2006). A wireless mesh network includes a number of static, wireless base stations which are strategically deployed and controlled by the system. These base stations wirelessly forward user traffic over multiple hops to ``gateway'' base stations, each of which additionally has a connection to the internet. My research is part of the Transit Access Points (TAPs) project at Rice University and is also relevant to the emerging IEEE 802.16 standard being promoted by the WiMAX Forum. Both the IEEE 802.16 Working Group and the TAPs project envision providing wireless broadband access to residential and public areas, with end clients possibly being mobile. As part of my PhD dissertation, I have addressed the problem of routing among base stations for best-effort traffic in the presence of simultaneous traffic, a condition that earlier routing protocols have neglected. In my work, I have also taken into account the effects of channel loss and channel modulation rates.
  
  In collaboration with Byoung-Jo (``J'') Kim and N.K. Shankar from AT&T Labs, Middletown, NJ, USA, I have also developed a scheme to increase the coverage of base stations thus covering a given geographical area more economically than with existing schemes. This new scheme also improves the capacity of the covered area by allowing for spatial radio resource reuse. We have proposed this work in the IEEE 802.16 Task Group meetings.
  
  
• PRAAN (Fall 2003 - Fall 2004). I, together with Khoa To, Santashil PalChaudhuri, and Shu Du, fellow graduate students at Rice University, developed a new, portable system for physical implementation of ad hoc network routing protocols. This system, called PRAN, allows the use of existing simulations models of protocols, without modification, to create a physical implementation of the same protocol. This helps in easier debugging, validation, reuse, and maintenance, as a single code base can now be used for both simulation and physical implementation. Our system works in FreeBSD and Linux and currently has physical implementations of DSR and AODV, two common routing protocols for mobile ad hoc networks, created from the unmodified ns-2 simulation models of these two protocols. We presented a live demo of our work at at the Tenth Annual International Conference on Mobile Computing and Networking (MobiCom 2004).
  
  
• SAFARI (Fall 2003 - Fall 2004). A scalable architecture for ad hoc networking and services. This work involves an architecture which can scale an ad hoc network into a city wide scale. The unofficial website (maintained by santa at cs dot rice dot edu) is here.
  
  
• Modelling Mobility in Vehicular Ad Hoc Networks (Summer 2004). In this project we developed a tool to generate ns-2 compatible scenario files from real road maps, as available from the publicly accessible U.S. Census Bureau TIGER/Line database. Since they are dependant on real street maps, the scenarios produced by this tool are more realistic than most synthetic scenarios.
  
  
• TreeCast: A Stateless Addressing and Routing Architecture for Sensor Networks. (Spring 2004). Recent advances in technology have made low-cost, low-power wireless sensors a reality. A network of such nodes can coordinate among themselves for distributed sensing and processing of certain phenomena. In this paper, we propose an architecture to provide a stateless solution in sensor networks for efficient addressing and routing. We name our architecture TreeCast.
  
  We propose a unique method of address allocation, building up multiple disjoint trees which are geographically inter-twined and rooted at the data sink. Using these trees, routing messages to and from the sink node without maintaining any routing state in the sensor nodes is possible. Next, we use this address allocation method for scoped addressing, through which sensor nodes of a particular type or in a particular region can be targeted. Evaluation of our protocol using ns-2 simulations shows how well our addressing and routing schemes perform.
  
  
• Ad Hoc City (Fall 2002 - Spring 2003). In this project we designed a city wide ad hoc network composed of a wired backhaul of a few base stations and a large number of mobile wireless devices. Hence, this was a departure from mobile ad hoc networks towards wireless mesh networks. We augmented the cellular architecture by not requiring mobile wireless devices to communicate directly with the base stations, but allowed for these wireless devices to reach base stations through multihop wireless routes. We developed a new routing protocol called Cellular DSR (CDSR) for this multi-tier architecture. We evaluated the entire architecture by placing a handful of imaginary base stations in different parts of Seattle, and used publicly available mobility traces from a fleet of buses in Seattle.