Adaptive
Equalizer
In modern digital communication systems
data is transmitted at high speed through some band-limited channels (e.g.
dial-up telephone network, high-frequency (HF) wireless radio channel etc.).
Transmission over such channels is distorted by bandwidth restrictions (e.g. 3
kHz for most telephone line), multipath dispersion, and in HF radio channel
Doppler fading may cause a problem as well. All these problems result in
inter-symbol interference (ISI) in digital communication, which causes high data
error rate.For this reason an
equalizer is generally required in the demodulation to compensate such
interference before the data can be further processed. Because the channel or
the distortion may change with time (time-variant), such equalizer must be
adaptive. There are many kinds of equalizers: optimum equalizer estimates the
maximal likelihood data sequence according to the channel knowledge. Such
equalizer has usually very high computational complexity and is impractical or
infeasible for time-variant channel.
Many kinds of low complexity equalizer are
proposed. They include linear equalizer, which operates with different criteria
e.g. peak distortion, minimum mean square error (MMSE), etc. Decision feedback
equalizer (DFE) is a very popular nonlinear equalization scheme. In most case it
is a combination of a linear equalizer and a nonlinear decision part, both parts
can be implemented with different algorithms. Most recently, iterative
equalizers (Turbo equalizers) are researched where channel estimation and
channel decoding are considered as well.
All these techniques may be combined to reach
a lower data error rate at low computational complexity.
This is a very old but still very active research area.
There are several kinds of adaptive
equalizers in WAVECOM decoders: decision feedback equalizer based on Kalman
algorithm, equalization based on estimated channel matrix and Turbo equalizer.
These equalizers are used in most HF-ACARS, MIL
and STANAG modes, because most of them employ a bandwidth efficient single
carrier 8-PSK modulation scheme for high speed data transmission (up to ca. 10k
bps).
The improvement of bit error rate with
equalizer is usually big, this depends on the signal quality and quality of the
equalizer. The following two phase constellation diagrams show the demodulation
with and without the decision feedback equalizer used
in WAVECOM decoder. A real-time recorded signal from an HF-antenna is used. The
phase constellation of the demodulation without the equalizer is almost equally
distributed on the phase plane, whereas the one with the equalizer shows a quite
concentrated distribution around the 8 possible phase points. The error rate of
demodulation with an equalizer is usually small, in this example it is about 2%,
the one without using an equalizer is about 60%, this makes the demodulated bits
totally useless.
No Adaptive Equalizer |
WAVECOM Adaptive Equalizer |
