Research


Overview

My research interests are in computer communication networks in both wired and wireless
domains.  Specific areas of interest include:

            1. Analysis and design of MAC protocols for wireless networks
            2. Performance evaluation and analytic modeling of computer networks
            3. Protocols for sensor networks
            4. Anomaly detection in computer networks
            5. TCP performance issues

A brief description of my research is given below.

Biological Models for Computer Networks: With their inherent complexity, efficiency,
and large scale, insights from biological systems can prove to be very helpful in the design
and evaluation of computer networks. I am currently working on applying biological
models for evaluating the performance and dynamics of computer networks. This approach
will be used for modeling the spatio-temporal dynamics of the networks in terms of their
performance, growth and pathological conditions. In our recent publication in IEEE
INFOCOM, we have shown that the dynamics of the propagation of polymorphic computer
worms have a remarkable similarity to quasi-species co-evolution in biological systems. We
also derived the maximum allowable response time of the intrusion detection system in
order to contain such worms and the optimal worm mutation rates. Our current work in this
direction focuses on automatic signature generation techniques for polymorphic worms. In
another result, we have demonstrated how population based models can be used to model
the lifetime of battery constrained wireless networks and used it to develop node recharging
and redeployment schemes. Currently we are working on biologically motivated protocols
for information dissemination and retrieval in large scale networks. I am funded by a NSF
CAREER award for this project.

2. Medium Access Control in Wireless Networks: Ad hoc and wireless networks have
been increasing in their popularity in the recent years. However, before they achieve their
full potential and support emerging applications, a number of research issues need to be
solved. My research interest is in developing medium access control protocols for wireless
networks in both IEEE 802.11 or WiFi based local area networks as well as IEEE 802.16 or
WiMax based metropolitan area networks. My research group is currently working on
mechanisms to increase the spatial reuse in wireless networks while maintaining acceptable
quality of service. Our previous work in this area has developed a distance-aware virtual
carrier sensing mechanism which can significantly improve the MAC layer spatial reuse as
compared to 802.11 DCF. We have also developed detailed analytic models to evaluate the
scaling properties of the MAC layer spatial reuse, traffic characteristics of 802.11 DCF,
queueing and channel access delays associated with both 802.11 DCF and PCF, and 802.16
based networks. We have quantified the contribution of the contention and backoff
mechanism on the delays and shown the tradeoff between larger transmission opportunities
and reduction in the MAC layer delays. In the area of WiMax, we have also developed
efficient scheduling strategies to simultaneously support subscriber station delay and
bandwidth constraints. I am currently funded by Intel corporation, WiMax forum and an
NSF ITR grant for this project.

3. Protocols for Sensor Networks: With recent advances in fabrication technology, large
scale sensor networks, capable of pervasive monitoring and controlling physical systems
has become possible. My group is currently working on developing protocols for the
coordination, information dissemination and information collection in such large scale
networks. To addresses the problem of energy-efficient data acquisition in sensor networks,
we have developed a swarm intelligence based distributed protocol that uses a packet pair
technique to set up the lowest energy gradient pointing to the sink at each sensor node. As
a complement to this work, we also addressed the problem of energy efficient information
broadcasting in sensor networks. We have developed a constellation based technique that
forms a broadcast tree on the fly, without the need for neighborhood information and
accommodates high levels of node mobility. We have also worked extensively on
multicasting in ad hoc networks, and more specifically, developed protocols with localized
control for both loss recovery and congestion control. Finally, my group has also worked
on and developed distributed protocols for tracking mobile targets using sensors with limited
communication and processing capabilities. I am currently collaborating with researchers
from Intelligent Automation Inc. to develop sensor nodes with our proposed protocols as
part of a DoD STTR initiative.

4. Anomaly and Change Detection in Computer Networks: In collaboration with
researchers from Bell Laboratories, we are working on techniques to detect routing
anomalies in both wired and wireless networks. In our work on wired networks, we have
focused on BGP routing anomalies, instabilities and cascading failures that may result
from misconfigurations, malicious attacks and equipment failure. To detect such
instabilities, we have developed an online mechanism based on statistical pattern
recognition techniques. In wireless networks, we are currently working on predicting fading
and shadowing effects and developing rate control mechanisms based on these prediction
schemes. For this work, we are collaborating with researchers from Phillips Research
Laboratories.

5. Network Modeling and Analysis: I am particularly interested in various aspects of
analytic models for evaluating the performance of networking protocols and hardware.
Analytical modeling using stochastic and probabilistic approaches has been a central
feature of my research work. In the past I have developed detailed analytic models for
evaluating the latency and steady state throughput of TCP connections. My doctoral
dissertation work has also developed models which show how TCP can lead to scaling
effects in network traffic. I have also worked extensively on the performance evaluation of
switching architectures for both unicast and multicast switches, working in both
continuous as well as discrete time. I also have a number of publications related to
queueing analysis of MAC layer protocols. One of the main research topics I am working
on right now is an NSF funded project on applying biological models to computer networks.

     

Currently funded projects:

1. NSF CAREER Award, Biologically Motivated Models for Spatio-Temporal Behavior of
Computer Networks: Performance, Growth and Pathological Behavior, 2004-2009.

2. NSF ITR: Community Wireless Networks for Last-Mile Broadband Interconnectivity: An
Experimental Research Program, 2003-2007.

3. Intel corporation: Infrastructure Mesh Wireless Networks, 2004-2006.