| | |
Summary: IEEE COMMUNICATIONS LETTERS, ACCEPTED FOR PUBLICATION 1
A New Fast Householder-Based Fractionally-Spaced FIR MMSE-DFE
Computation Algorithm and its Real-Time Implementation
K. M. Zahidul Islam, Naofal Al-Dhahir, Russell McKown, Robert Dawes, and Christopher Stillo
Abstract--We present a new parallel algorithm for computing
the optimum coefficients of the fractionally-spaced FIR MMSE-
DFE based on the Householder orthogonal transformation which
avoids the back-substitution computational bottleneck in [1]. The
resulting significant reductions in the algorithm's execution time
are demonstrated through a real-time FPGA implementation.
Index Terms--MMSE-DFE, fractionally-spaced, QR decompo-
sition, householder transformation.
I. INTRODUCTION
THE finite-impulse-response (FIR) minimum mean-
squared error (MMSE) decision feedback equalizer
(DFE) is a widely-used single-carrier equalization structure as
it outperforms linear equalization while significantly reducing
the detection complexity compared to maximum-likelihood se-
quence detection. Its applications include single-carrier digital
TV receivers, high-speed backplane serial links, IEEE 802.11b
|
