Page 81 - Complete Wireless Design

Modulation

80 Chapter Two

will limit excessive intersymbol interference, since the demodulator would
have great difficulty in deciding whether an input signal was a 1 or a 0 if high
ISI were present. A raised cosine filter (a type of Nyquist filter) is commonly
employed for this purpose. Raised cosine filters are utilized to slow the transi-
tions of a digitally modulated signal from high to low, or from low to high, in
order to decrease the bandwidth needed to transmit the desired information,
without degrading the ISI and the BER at the symbol decision times, as dis-
cussed above. These filters are usually matched, with one placed between the
incoming data and the digital-to-analog converter (DAC) in the transmitter
and the other half placed in the demodulator of the receiver. This replicates
the response of a full Nyquist filter.
To compute the required bandwidth needed for a wide cosine filtered symbol
rate, the formula is BW symbol rate (1 ), with between 0 and 1. It would
be very bandwidth efficient if the BW could be equal to the symbol rate (this is
not quite practical), which is the same as 0 for a raised cosine filter. Anything
over this value of zero for is referred to as the excess bandwidth factor, because
it is this bandwidth that is necessary beyond the symbol rate BW value. We
will always require an excess bandwidth greater than the symbol rate; or an at
some value that is over zero. If equaled 1, the bandwidth necessary to transmit
a signal would be twice the symbol rate. In other words, twice the bandwidth is
required than the almost ideal situation of 0. A contemporary digitally mod-
ulated radio, however, will usually filter the baseband signal to a value of
between 0.2 and 0.5, with a corresponding decrease in bandwidth and increase
in the required output power headroom compared to an 1 device. Figure 2.35
demonstrates the effect on the digital input signal’s rise and fall times, the chan-
nel’s bandwidth, and the received constellations, as is varied.
Values of lower than 0.2 are very uncommon because of the increased cost
and complexity of building sustainably accurate filters (with high clock preci-
sion) in mass production environments. Any attempts at lower will also
increase ISI to unacceptable levels, along with the added expense of producing
amplifiers that must be capable of greater peak output powers without exces-
sive distortion products. Power back-off is required of these amplifiers because
of the elevated power overshoots (see Fig. 2.35) created by the increased fil-
tering of the digital signals by the Nyquist-type filtering, which limits the
transmitted bandwidth. For heavily filtered QPSK, the excess peak power
requires the solid-state power amplifier (SSPA) to have a P1dB that is at least
5 dB over what would normally be required for an unfiltered signal. This is to
allow the power overshoots of the signal enough headroom so as not to place
the SSPA into limiting, which would create spectral splatter into adjacent
channels. All signals that have a modulation envelope—even if not used to car-
ry information—will be affected by this Nyquist filtering, including QPSK,
DQPSK, and QAM signals.
Gaussian filters are another method of slowing the transitions of the signal
in order to decrease occupied bandwidth in the modulation scheme GMSK.
Unlike raised cosine filters, however, these filters create a certain amount of

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2004 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.