Contributions to Discovery of new results in Computational Sensor Networks/Robotics/ Image Processing and Applications
- Dr. Iyengar's discovery in many problems of Computational Sensor Networks/Robotics/Oceanographic Image Processing is in developing a connecting link between the theory and applications in all of the above fields.
- Contrary to popular belief in 1988, Dr. Iyengar discovered "NC algorithms for Recognizing Chordal Graphs and K- trees "[IEEE transactions on computers 1988]. This break through result led to the extension of designing fast parallel algorithms by researchers from Stanford, UC Berkeley and AT & T Bell Labs. These are used in acyclic relational database schemes for image processing.
Dr. Iyengar's research in image processing algorithms was a key factor leading to the first fully automated interpretation of a satellite image of ocean in 1989 by US NAVY. - In 2008-10, Dr. Iyengar and his team discovered a new deterministic data structure for network design that immediately asymptotically enhaces the performances of several varied practical applications such as large-scale distributed publish-subscribe systems, transportation & logistics, VLSI circuitry etc.
A fundamental problem in computer science is to design oblivious data structures that solve buy-at-bulk problems in networks which enhance the performances of several varied practical applications such as large-scale publish-subscribe systems, transportation & logistics, VLSI circuitry etc. Particularly interesting is the problem of single-sink buy-at-bulk, where the task is to form low cost fused paths from any arbitrary set of sources to a single destination.
Recently, Srinivasagopalan, Dr. Busch, and Dr. Iyengar demonstrated that this problem can be solved efficiently in low doubling dimension graphs using a single oblivious spanning tree rooted at the sink. Spanning trees are important because they are the most compact data structures in terms of number of edges in graphs and useful for routing purposes. Further, this construction provides immediately a novel oblivious Steiner tree towards the sink, which was not known before how to compute in graphs. The authors are currently working towards constructing oblivious spanning trees in minor-free graphs and general graphs.
They have also considered the more general version of oblivious buy-at-bulk (with multiple sinks) and have shown that it can be solved optimally in planar graphs.
This resulted in the publication of- Srivathsan Srinivasagopalan, Costas Busch, S.S. Iyengar, "An Oblivious Spanning Tree for Single-Sink Buy-at-Bulk in Low Doubling-Dimension Graphs", IEEE Transactions on Computers, to appear, 2011
- Srivathsan Srinivasagopalan, Costas Busch, S.S. Iyengar, "Oblivious Buy-at- Bulk in Planar Graphs.", Accepted for publication in the proceedings of the Workshop on Algorithms and Computation (WALCOM 2011), in Springer-Verlag Lecture Notes Computer Science.