RESEARCH
SUMMARY
My research interests are broadly in the areas of modeling and simulation of complex network systems with a focus on biological networks, distributed computing, wireless/mobile computing and optical networks.
My
PhD thesis research was concentrated on the stochastic modeling,
analysis and simulation of complex Biological Systems
where we are developing a discrete event simulation tool for a quantitative
understanding of complex processes in life sciences. Some details on this
software can be found at (http://crewman.uta.edu/dynamic/bone.php).
My research experience outside my PhD dissertation area span various network modeling and simulation techniques in distributed systems and wireless/optical networks that can be briefly summarized as follows:
1.1 Stochastic models for In-Silico Event-based Biological Network Simulation: PhD dissertation topic
I am currently working on modeling and simulation of complex biological networks. I work in the Biological Networking group, where our core goal is to apply system modeling techniques for holistic understanding of cellular processes.
My thesis work is centered around developing generic,
parametric and computationally fast models of basic biological events like
biochemical reactions (involves chemical kinetic theory concept from
biochemistry), molecule/ion diffusion (involves formulation and solution of
diffusion equations from biophysics), ligand-protein docking (borrowing concepts
from biochemistry), protein-DNA binding (borrowing concepts from
biochemistry/biophysics). My research also focused on developing a new paradigm
to build complex bioevent models using these basic models like models for
protein synthesis.
Our major research objective is to maintain the stochastic
nature of bioevents because many current models ignore this aspect of biological
events. The models are reusable for different cell conditions i.e., temperature,
cell volume, molecular velocity distribution are parameterized in the model.
These models are used to map the biological pathways as a thread of bioevents
and our discrete event based simulation framework, iSimBioSys and HimSim is used
to study the dynamics of the interaction of the gene regulatory, signal
transduction and metabolic networks in cells.
Our research team used this simulation to build an E. Coli
genome scale simulation and is currently working for Human heart muscle cell
model. Thus my PhD thesis area was essentially multi-disciplinary (three
committee members from the Computer Science department, one each from
Mathematics and Biology departments at UTA) with a strong emphasis on
computational sciences.
Future Research Plans:
The modeling requires concepts from biophysics and
biochemistry and mathematical biology. This involves queueing theory, collision
theory, chemical kinetics, differential equations (ODE/PDEs), control theory,
Monte carlo simulation and Markov chains. However, I have only developed the
basic models that can encompass a broad range of biological functions that are
essential for the simulation. There are a lot of research opportunities in
biological network modeling where we can improve on the models we have already
developed, and also come up with stochastic models of various other biological
functions that we have not considered e.g., ion transport using active/passive
pumps, protein folding, protein decay, chromatin remodelling and so on. This
will make the simulator both scalable and more accurate.
I also developed a Markov
chain based stochastic simulator that can analytically study a system of
biochemical reactions (without using
numerical solution techniques) and overcomes the stiffness problem of the
classical Gillespie simulator. These simulators are more flexible and
scalable than the standard ODE based system simulators and can still be improved
upon. Specifically, the Markov chain based stochastic simulator can be
substantially improved in terms of memory usage and computational speed by using
graph theoretic optimization techniques that reduce the number of state spaces.
The next step is to study more complex biological systems, which can lead to many interesting predictions of previously unknown biological phenomenon. In fact, during my internship at Pfizer Global Research (PGRD-Groton, USA), I worked on a system simulation of the RNA interference technology (RNAi). RNAi is expected to be the most promising direction in medicine research today, and a lot of work is required to be able to devise novel drugs using this newly discovered gene silencing mechanism.
2.1 Resource Management and Job Scheduling in Mobile Grids
1.
We proposed a generic mobile grid infrastructure to
harness the idle CPU cycles of mobile devices.
2.
We next formulated game
theoretic pricing strategies that can attract the mobile device owners to
contribute their devices for grid jobs.
3.
The pricing strategy is then used to devise cost-effective grid job
allocation schemes to the mobile nodes considering
i.
the processing cost (or delay)
at the mobile nodes
ii.
the internal jobs at the
mobile device (e.g. call processing activities)
iii.
the communication cost (or
delay) for transferring the jobs
iv.
the node mobility as the results of the jobs assigned by a particular
wireless access point (WAP) to the
nodes need to come back to the WAP after completion
4.
We also proposed a mobility
management algorithm to track the number of mobile devices present under a
particular (WAP) within a specific time period (in which the assigned jobs need
to complete).
2.2
Photonic Container Switching: MS thesis topic
My Master's thesis work was in the area of Optical Networks. The ever-growing demand of network capacity has resulted in the inception of Optical Burst Switching (OBS) offering all-optical transmission, high-speed data rates and format transparent switching. But the current OBS architecture is very complex requiring costly fiber delay lines and quality of service management techniques.
We proposed a new OBS architecture based on photonic container switching to be deployed in the core network. We showed that our architecture will solve most of the complexities of existing OBS mechanisms, and in fact will make the core an all-optical, zero packet loss network that will also guarantee equal QoS to all the users. The packets are actually packed in fixed size containers, which will be converted into an optical burst and transmitted through the network.
We also proposed some efficient algorithms to design a centralized scheduler that ensures zero packet loss and no optical-to-electrical switchings in the intermediate nodes. We analyzed the performance of our algorithms under varying traffic conditions and network topologies to ascertain their efficiency and robustness. Moreover, we identified the modules required for implementing such an architecture and conducted numerous experiments to study the feasibility and performance of such an architecture.
2.3 Optimization Problems in Wireless Network
During my internship at Nokia Research Center, I focused on designing a
hybrid architecture to support massively multiplayer games incorporating mobile
device users. Also, I worked on designing and implementing a dynamic
packetization algorithm and proposed a TCP-friendly transport layer protocol
that incorporates adaptive Forward error control and rate control techniques to
improve throughput in such wireless gaming infrastructures.
I have also worked in the
area of topology design and performance analysis of wireless mesh networks,
focusing on genetic algorithm based approaches for designing optimal
connectivity in an arbitrary mesh deployment scenario. The same concept was
extended to design a reliable Virtual Private Network (VPN) later on.
During my internship at
Nortel Research Labs, I worked on developing an analytical framework for
performance evaluation and capacity planning for Nortel's IP Multimedia Systems
that reduces the complexity of OPNET based modeling. I also worked on an
adaptive overload control algorithm for generic telecom switches that can
maximize the revenue of the wireless service providers. Some other research
works include combinatorial reverse-auction based slot scheduling algorithms and
modeling user-satisfaction factor to improve the quality of service in
multi-rate wireless systems.