Digital Signal Processing Reference
In-Depth Information
implies that delayed versions of the same symbol start to interfere with adjacent
symbols, an effect which is known as
intersymbol interference
(isi). In order
to deal with the degrading effects of isi on the symbol error rate of a trans-
mission, a combination of channel equalization and
orthogonal frequency di-
vision multiplexing
(ofdm) is commonly being employed by wireless network
devices (e.g. 802
.
11a/g). In contrast to an equalization filter, which flattens
the frequency response of the channel, ofdm splits up the frequency band in
multiple parallel communication subchannels. Each of the subchannels in the
ofdm system handles a symbol rate which is considerably lower than the de-
lay spread of the channel, which wipes out the problem of isi. However, the
transfer of information remains impossible in those frequency bands which are
blanked out by deep fades. The result is an excessive rate of bit errors in cer-
tain subchannels of the ofdm system. In the ideal scenario, the controller of
the system would avoid transmission over subchannels with an excessive noise
level and redistribute the information over other, unaffected subbands. Obvi-
ously, this approach requires an active intervention from transmission side,
while the receiver must keep the transmitter informed on the current channel
conditions. The latter aspect turns out to be a big problem under varying chan-
nel conditions: by the time the transmitter gets informed about the transfer
characteristic measurement performed at receiver side, the information may
already be outdated.
This is one of the main reasons why ofdm-based systems employ bit-
interleaved coded modulation (bicm), which distributes information at the bit
level over multiple subchannels of the ofdm transmission. The general idea
behind this joint coding and modulation technique is to evenly spread the er-
rors over a longer sequence of symbols and as such avoid bursts of erroneous
bits being fed to the (Viterbi) decoder in the receiver. The net result of com-
bining ofdm with bicm is that the problem of isi is in fact transformed into a
problem of Gaussian noise, which turns out to be the particular specialty of a
traditional forward error coding (fec) mechanism. However, major opportuni-
ties are being missed here to make real advantage of the frequency selectivity
of a wideband transmission channel.
Modulation-aware decoding: signal reconstruction
With these thoughts in mind, and from the fact that ofdm is not exactly
the most energy efficient modulation technique due its large peak-to-average
power ratio, Chapter 3 introduced a new line of approach. Instead of involv-
ing the transmitter in the channel coding process, the core activities which deal
with the quirks of the channel are shifted from the transmission side to the re-
ceiver end. While the transmitter only employs a very basic modulation scheme
(wideband qpsk), it is the full responsibility of the receiver to deal with the
