Noncausal and bidirectional feedback equalization for high performance equalization and decoding

Yun-Ho Lee, Purdue University

Abstract

The goal of this research is to develop equalizers which can provide soft information to channel decoders for use in single stage, or the initial stage of iteration, of equalization and decoding architectures. Previous research by us and others has identified practical soft interference, ML and MAP type equalizers suitable for complex channels with large delay spread. These structures are appropriate for noninitial stages or iterations, typically requiring sufficiently accurate initial information to perform some type of soft cancellation to reduce complexity to practical levels. The single or noninitial stage or iteration problem is different, and must deal successfully with error correlation typical in coded systems and where soft cancellation is not yet possible. The simple decision feedback equalizer (DFE) provides a framework for such single or initial stage or iteration equalizers. The most promising approaches to achieving desired performance involve incorporation of soft processing, noncausality and bidirectionality in some form. Soft processing uses various models to make the regressor and/or error soft, and can be used in the symbol-wise decisions and/or filter coefficient computation. Noncausality refers to the use of future tentative decisions from another equalizer which is then used in a conventional interference canceller. Bidirectionality involves the combining of outputs from forward and backward DFE's. These various approaches have positive impact on error propagation and ISI reduction, but there is not a systematic approach which includes these different ideas. We have developed a basic approach which naturally combines noncausality and bidirectionality, while maintaining the simplicity of both. We formulate our approach for nonsymmetrical channels considering cross error correlation, and examine the effect of noncausality, bidirectionality and soft processing independently and together. Also, we show its theoretical bound and the relation with other bidirectional DFE. The method is employed as a first stage in a staged equalization and decoding architecture, and applied to North American HDTV terrestrial broadcast, and to the EDGE cellular standard to examine its performance.

Degree

Ph.D.

Advisors

Gelfand, Purdue University.

Subject Area

Electrical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS