My Education
I passed the Secondary Examination with 92.5% marks in the year 2003 from Debiswari Vidya Niketan, Hooghly, West Bengal. Then I was admitted to Belur High School, Howrah, West Bengal from where I passed my Higher Secondary Examination securing 73.4% marks in 2005. My major subjects in Higher Secondary were Physics, Chemistry, Mathematics and Statistics.
In 2005, through WBJEE examination I got an admission to the department of Computer Science and Engineering, Academy of Technology, Hooghly, West Bengal. I received my B.Tech degree in Computer Science and Engineering in the year 2009 with 8.65 DGPA
My next destination was Indian Institute of Technology, Guwahati. I secured AIR 430 in GATE 2009 and was admitted to the department of Computer Science and Engineering, IIT Guwahati for a two-year M.Tech course. From here I received my M.Tech degree in Computer Science and Engineering with 8.82 CPI in 2011. My Master's Thesis, entitled Dynamic Tree-Switching for Adaptation in Distributed Message-Passing System has been supervised by Dr. Sushanta Karmakar, Asst. Professor, Dept. of Computer Science and Engineering, IIT, Guwahati.
My Research
Presently I am a Ph.D research scholar in the department of Computer Science and Engineering, IIT Guwahati. I am doing research in the area of distributed system and networking under the guidance of Dr. Sushanta Karmakar, Asst. Professor, Dept. of Computer Science and Engineering, IIT, Guwahati. Designing and verification of algorithms for various distributed applications are my current field of interest.
Designing efficient algorithms for data forwarding in wireless sensor networks is a challenging problem due to limitation in sensor resources as well as inherent dynamics in the environment. Additionally, different message-passing based applications in sensor network that exploit broadcast, convergecast or multicast might impose various constraints to achieve the desired standard of performance metrics. Dealing with all these issues the main objective of this thesis is to design efficient algorithms for reliable data gathering in sensor network and further fine tuning the proposed schemes according to various application specific requirements. Data gathering or convergecast is one of the most popular applications in wireless sensor network where application data from all nodes are forwarded towards a base station or sink. Existing tree based convergecast schemes, although have been studied in literature for its efficiency, lacks in providing a complete solution considering all aspects of connectivity, coverage, fault tolerance, network lifetime and several QoS metrics. First contribution of this thesis proposes a distributed BFS tree construction rooted at the sink node as well as tree maintenance scheme for crash- tolerant data gathering in sensor network. Every node pre-computes an alternate parent during tree construction period such that on sudden failure of parent node, the tree can be repaired locally with minimum control overhead and repairing delay. Thus, application messages are delivered to the sink with minimum loss or redundancy even in presence of arbitrary node crash. Moreover, multiple simultaneous node failures have also been handled through reactive repairing technique. While maintenance of connectivity was one concern working with this problem, the effect of arbitrary node failure on sensing coverage and network lifetime were other major issues that have been addressed as subsequent contributions of the thesis. Considering irregular terrain property and optimal positioning of the base station, energy depletion rate gradually decreases from the base station towards the terrain periphery. Thus, both the connectivity and sensing coverage is affected as nodes near the sink die out of energy sooner than the leaves of the tree rooted at the sink creating holes. To enhance network lifetime, a gradient based node deployment strategy has been proposed that also maintains initial connectivity and coverage criteria. It has been observed that network lifetime will be improved if density of deployed nodes follow the gradient which is estimated to be the amount of energy dissipation at any intermediate node to that of the leaves in its rooted subtree. The proposed theory has been justified through worst case analysis of sensor network calculus. As unbalanced data gathering tree escalates the problem of uneven energy depletion in the network, a load-balanced distributed BFS tree construction scheme has been proposed. Further, to handle arbitrary node failure, an efficient tree maintenance scheme has been introduced which is local as well as cost-effective. Finally, for application specific customization, the proposed schemes have been augmented to serve in the field of road-side sensor network and border area sensor network. Data gathering for these applications demands reliable delivery of data within strict delay bound. Also nodes are highly failure prone in both road-side or border area environment. Thus designing efficient algorithms maintaining both connectivity and coverage as well as satisfying the application specific performance criteria is the principal goal of this part of the thesis. Exploiting the strip like structure of road network, a DFS tree based convergecast has been proposed that supports tree maintenance from arbitrary leaving and joining of nodes either due to execution of normal sleep wakeup based schedule or due to node crash. The proposed scheme assures delivery of application data to the sink within given time bound and without any loss or redundancy even in changes in underlying topology. Border area enforces the most strict bound on reliability and delay for data delivery. Thus, for this type of environment, the proposed solution constructs multiple BFS trees rooted at multiple sinks such that from every node there exists multiple node disjoint paths to different sinks. This assures delivery of data to the base station with zero loss and minimum delay. At node level, efficient parent node selection is dependent on three factors, traffic load, shortest path distance from the root of the tree and residual energy of the node. Proper implementation ensures efficient Quality of Service for the application as well as enhanced network lifetime with low overhead.