Erik Lindskog

Licentiate Thesis, Report UPTEC 95069R, 82pp, April 1995.

Outline:

When digital data are transmitted over dispersive radio channels, a Decision Feedback Equalizer (DFE) can be utilized to estimate transmitted symbols. This detector subtracts an estimate of the intersymbol interference caused by previous symbols, before making symbol-by symbol decisions. The present thesis focuses on two questions related to the use of DFE's:

• Robust design, to reduce the effect of model uncertainties and decision errors.
• The use of DFE's in receivers with multiple elements. Many coefficients may then have to be adjusted based on short training sequences.

Abstract:

The use of Decision Feedback Equalization for mobile radio applications is considered. Algorithms for indirect model-based MMSE design, which are robust against imperfect channel and noise modelling as well as error propagation, are presented. These algorithms can also be designed to be robust against a slow time-varying dynamics of the transmission channel. An example for a GSM-type mobile radio channel is included.

Algorithms utilizing a circular antenna array are also considered. The challenge here is that detectors with a rather large number of filter parameters are to be adjusted based on short training sequences. Several algorithms based on combined spatial and temporal equalization are suggested. Both direct and indirect schemes of low complexity are investigated. Separate tuning of the beamformer coefficients and subsequent equalization yields better performance for short training data records than a simultaneous adjustment of all coefficients. All the investigated algorithms perform well in spite of co-channel interference. To avoid ill-conditioned matrices in the solution for very short data records, regularization is introduced in some algorithms.

An adaptive inteference cancellation scheme for an indirect spatio-temporal equalizer is also addressed. Finally, beamforming with a modified version of the sample inversion method is considered. It is applied to a GSM signal with sampling offset.

1. Summary. (In Postscript : compressed(gz) 37K , uncompressed 89K.)
   (In Pdf 290K.)
2. Robust Decision Feedback Equalizers
3. Designing Decision Feedback Equalizers to be Robust with Respect to Channel Time Variations
4. Combined Spatial and Temporal Equalization Using an Adaptive Antenna Array and a Decision Feedback Equalization Scheme
5. Spatio-Temporal Equalization for Multipath Environments in Mobile Radio Applications
6. Indirect Spatio-Temporal Equalization and Adaptive Interference Cancellation for Multipath Environments in Mobile Radio Applications
7. Beamforming with the Sample Matrix Inversion Method for a GSM Signal with Sampling Offset.

An Appendix with a derivation of the robust DFE is also included.

Related publications:

PhD Thesis by Erik Lindskog.

Later Conference paper (IEEE VTC'97), on the implementation of Viterbi algorithms in antenna arrays.
Later Conference paper (IEEE ICUPC'96), on direction-of-arrival parametrization.
Later Conference paper (Asilomar'95), on utilizing adjacent TDMA frames.
Multivariable DFE's utilized as multi-user detectors, in IEEE PIMRC'95.
Robust filter design and uncertainty modelling, as described in Automatica 1993.
DFE design based on exactly known IIR model (IEEE Trans. IT 1990).