Conf. on Info. Sciences and Systems, Baltimore, MD, March 2001
Timing Recovery Based on Dispersion Minimization
W. Chung, J. Balakrishnan , W.A. Sethares and C.R. Johnson, Jr.
Abstract
This paper investigates the idea of applying
``dispersion minimization'' to the
problem of timing phase recovery, and analyzes the resulting cost
surface.
The minimization can be readily accomplished via a gradient-style
algorithm, and the behavior of the algorithm is analyzed in a variety
of situations. In spite of its self-learning adaptation,
expressions describing the error surface over which
the algorithm evolves show that the error surface is unimodal
for received signals which are band limited by the Nyquist frequency
for baud and $T/2$ sampling, even in the presence of multipath
interference.
For the baud sampling of raised cosine pulse shaped signals,
the algorithm tends to converge to a unique minimum, except in
certain extreme cases. Analysis shows that this algorithm performs as
desired in the absence of multipath with raised cosine pulse shaping.
