CAREER: A Unified Approach to Iterative and Universal Methods for Signal Processing and Communications

University Of Illinois At Urbana-Champaign, Urbana IL

Investigators

Abstract

Singer ABSTRACT: One of the most difficult aspects of the application of theoretical methods to practical systems lies in the translation of a real-world problem into a manageable theoretical model. Often there are a number of aspects of the problem that are either difficult to model concisely, or that are completely unknown. For example, in many commercial and military applications, compact, efficient devices are desired that can operate in a wide range of possible environments. This research is developing a framework for explicitly dealing with such uncertainties in a number of important areas such as wireless and underwater acoustic communications, signal estimation and detection, and forecasting. In particular, this research leverages some of the new and exciting methods for dealing with uncertainty and complexity from the information theory literature. Problems of uncertainty and variability are being addressed through the development of robust signal processing methods motivated by contemporary applications in lossless source coding and competitive on-line algorithms. Such universal algorithms can provide a particularly effective means for handling uncertainty in a variety of adaptive estimation problems. These include universal estimation and equalization strategies which without prior knowledge of the signal or channel can asymptotically achieve the performance of the optimal filter tuned to the signal or channel in use. These algorithms can be practically implemented and integrated into larger systems with relatively low complexity through the use of graphical models and iterative processing techniques used in the study of turbo codes and low-density parity check codes. Iterative algorithms such as factor graphs provide a natural framework for the joint optimization of a variety of tasks typically treated separately. Motivated by the near-capacity achieving performance of iterative decoding and turbo-codes, this research is developing practical, efficient iterative algorithms for joint detection, estimation and decoding of digital communications over single and multi-user channels with inter-symbol interference. The educational component of this research program includes significant curriculum development at both the undergraduate and graduate levels, and a strong investment in the mentoring of graduate students. A key aspect of this research includes the active involvement of undergraduates in ongoing research. Singer LEVEL OF EFFORT STATEMENT: At the recommended level of support, Andrew Singer will make every attempt to meet the original scope and level of effort of this project.