Introduction
5
on detection performance. Various receiver configurations and modulation
formats are considered and, in each case, theoretical or simulation results are
illustrated. The purpose is to establish ball-park limits on allowable synchro
nization errors. The second part concentrates on estimation criteria and gives a
self-contained account of ML estimation methods. In particular, likelihood
functions for continuous-time and discrete-time observations are derived for the
additive white Gaussian noise (AWGN) channel and the concept of wanted and
unwanted parameters is discussed. These parameters play a fundamental role in
the computation of the MCRB. The third part gives closed-form expressions of
the MCRBs for timing, phase and frequency under various modulation condi
tions.
Chapter 3 investigates carrier frequency estimation with passband PAM
modulation. A distinction is made between two rather different situations, de
pending on whether the carrier frequency offset is expected to be small or com
parable with the symbol rate. Different estimation methods apply in the two
cases. In particular, data-aided or decision-directed schemes can be used with
small offsets whereas non-data-aided schemes are inevitable otherwise. As is
intuitively clear, data-aided and decision-directed methods are much more ac
curate than non-data-aided ones. In fact, circuits in the first category perform
close to the MCRB while the others are far from it.
Chapter 4 concentrates on frequency estimation with CPM modulation.
The same distinction between “small” and “large” frequency offsets is made as
in Chapter 3. As opposed to PAM modulation, however, few methods are
available for small frequency offsets and, what is worse, they are limited to bi
nary symbols and a modulation index equal to 1/2. On the contrary, a variety of
estimation schemes can be used with large frequency offsets. Their perfor
mance is far from the MCRB, however, especially with long frequency pulses.
In consequence, narrow-band tracking loops are needed to achieve small esti
mation variances. Of course, this translates into rather long acquisition times.
Chapter 5 is the longest and is concerned with phase estimation in PAM
modulations. Its first part focuses on phase recovery for transmissions over
AWGN channels. Costas loops are popular synchronization schemes for con
tinuous transmissions over these channels. They are easily designed to com
pensate for (small) frequency offsets and have excellent tracking performance
in the absence of phase noise. In any practical situation, however, some degree
of phase noise is inevitable due to oscillator imperfections. The resulting track
ing degradations can be limited by proper loop design. This subject is ade
quately addressed and criteria are provided to minimize the phase errors. The
central part of the chapter considers frequency-flat fading channels. Here, the
signal is affected by a multiplicative distortion (MD) which is modeled as a
slowly varying Gaussian random process. As samples of the MD (taken at the
symbol rate) are needed for coherent detection, the problem arises of estimating