Signal Processing
Generalized Particle Filtering Theory and Its Applications
PIs: Petar M. Djuric, Mónica Fernández-Bugallo
In the past decade, particle filtering has generated tremendous interest among engineers and scientists with its capacity to process data that are modeled by dynamic systems. These methods belong to the family of procedures for sequential signal processing where the objectives are to filter, predict or smooth unknown and time-varying signals from available observations. The general area of work in this project is the building of a new class of particle filters, the development of their theory, and their application to a number of important tasks. Our main goal is to build a class of particle filters which do not use probabilistic model assumptions, and to develop various details of their theory including investigation of classes of functions that can be used in the construction of the new methods and developing guidelines for selecting parameters of the functions, choice of metrics, convergence issues, Rao-Blackwellization, robustness, and fusion of estimates from multiple particle filters. We also study the relationships of the proposed methods with the stochastic approximation and genetic algorithms. Finally, we apply these filters to problems primarily related to wireless communications and sensor networks. (NSF)
Advancements of Particle Filtering Theory and Its Application to Tracking
PIs: Petar M. Djuric, Mónica Fernández-Bugallo
The main objective of this work is to push ahead the theory of particle filtering and advance its use in tracking applications. To that end, the work is related to the improvement of existing particle filtering methods and the development of new ones. Besides theoretical development of the new and old particle filtering schemes, the objective is to apply them to various problems related to target tracking. Efforts include applications of the filter to tracking of single targets as well as to much more challenging tasks such as tracking of multiple targets where the number of targets may vary with time. Finally, scenarios that require multisensor tracking and data fusion are also be investigated. (ONR)
VLSI Design Methodology for Low Energy Arithmetic Units for Digital Signal Processing
PI: Sangjin Hong
We have developed several techniques for reducing power of arithmetic logic units for digital signal processing applications. This works will lay foundation for developing special-purpose architectures that can be used in many emerging applications (Symbol Technologies, Center for Electronics Imaging Systems (CEIS) - University of Rochester, Microelectronic Design Center).