Research Summary: Update 8/2007

The focus of our research since my transition to UCSD from UCI in 2002 has been on issues related to synthesis and operational efficiency in embedded systems. Continuing with that momentum we completed and released a set of tools within a parallelizing high-level synthesis framework, called SPARK. To date, SPARK has seen 4000+ downloads and sports an active user community, including its non-exclusive license by UCSD. We also made limited progress in formal capture and component compositions for system level models of embedded systems, their validation through formal verification and multi-processor simulations. Operational efficiency, measured in terms of energy efficiency across processing and communications was the centerpiece of work in our published literature during this period. This included our work on coordinated multi-radio interfaces, best demonstration award for a radio platform at the IPSN conference, optimized slowdown/shutdown of processors and radios on such platforms.

Beginning 2006, our attention has shifted towards exploration of spatial things in embedded systems: use of location in modeling, programming and in improving operational efficiency (for computation, energy). The focus of this effort is on the capture, modeling and treatment of spatial information in embedded systems. The overall vision is centered on enhancing the scope of embedded systems design from real-time to real-location. Currently, spatial reasoning in computing system is primarily treated as nothing more than an expression of concurrency, e.g., Pi calculus and ambient logic. We believe we can go much further than the confines of what logical inferencing can do in concurrent programming. Accordingly, we have structured our research in this area along three lines:

Improved real-time localization: this work currently led by Thomas Weng is focused on improving the accuracy of satellite-base localization to dm-scales in real-time through innovations in GPS augmentation by improve ephemeris, carrier-phase analyses. As a proof of concept Thomas is building an enhanced network-based localization server (ENLS) that leverages IPv6 for an improved infrastructure for infospatial addressing in embedded systems. This work straddles the boundary between GIS, radio communications and networking. We are working with researchers at MPL and SOPAC at SIO.

Improved use of real-space information in task scheduling: this work led by Ryo Sugihara is focused on modeling of spatial constraints, in the context of data collection by mobile nodes in sensor network applications, in optimizing operational efficiency in embedded systems.

Improved capture and use of spatial information in embedded programming: this is ongoing work in collaboration with R. K. Shyamsundar of IBM and TIFR and Amin Vahdat. As an exploration, Amin and I worked with Chalermak Intagonwiwat, to formulate the problem of spatially-aware resource allocation and programming. Since that collaboration, my work has evolved into programming support for spatial operations: We have worked on defining a  new set of observables related to spatial information (proximity, enclosure etc) that a programmer can use to model behaviors that evolve over space.


Research Summary: 2002

My research can be broadly classified into two categories: (a) those relating to our work on design automation, algorithms and techniques for efficient synthesis of digital circuits, system-on-a-chip and embedded systems; and (b) those relating to work on system architectures and design techniques. I have organized these endeavors into close nit projects: BALBOA, SPARK, AMRM, PADS and OSDPM. You can find a project to these project websites on my home page: http://www.ics.uci.edu/~rgupta    (the website is in transition to http://www-cse.ucsd.edu/~gupta but most links must work).

Recent Research Accomplishments:

Research Leadership:

Pointers