LL Logo



Recent Results in Blind and DF-Based Beamforming
and Equalization at
Brigham Young University

A. Lee Swindlehurst
Brigham Young University
Electrical and Computer Engineering
Provo, UT 84602
email: swindle@ee.byu.edu

Abstract Signal recovery in the presence of co-channel interference is accomplished by means of both spatial equalization (i.e., beamforming), and also temporal equalization in situations involving non-ideal (e.g., multipath) channels. Nearly all of the techniques (blind or otherwise) have been developed to solve either the spatial or the temporal equalization problem, but not both simultaneously. In addition, it appears that little effort has been devoted to studying relative algorithm performance, particularly in the area of beamforming.

In this presentation, some recent research results at Brigham Young University in these latter two areas are presented, as outlined below:

(1) We have undertaken a comprehensive study of DF-based beamformers in an attempt to classify them in terms of both mean squared error (MSE) and signal-to-interference-plus-noise ratio (SINR) performance. Algorithms studied include classical and linearly constrained beamformers, least squares, total least squares, principal components methods, and techniques that attempt to approximate the Wiener solution.

(2) Two new techniques for spatial equalization have been developed, the first designed to maximize SINR when highly correlated multipath is present, and the second based on spatial decision direction for digital signals. We have also analyzed the bit error rate performance of the latter technique for M-ary PSK signals.

(3) A method for partially blind, joint spatio-temporal equalization has been developed for time invariant multipath channels. The algorithm involves a least-squares fit to the data model in the frequency-domain, and provides an identifiable parameterization even in situations where the number of multipath rays far exceeds the number of sensors in the array.

Not surprisingly, our work has led us to conclude that the most promising approaches for equalization are those that attempt to incorporate both spatial and temporal information into the process, rather than ignoring one or the other. Simulation results are presented to validate our conclusions.



LL Logo Disclaimer

Direct comments and questions to: webmaster@ll.mit.edu

MIT Lincoln Laboratory. All rights reserved.