B. P. Lathi, Zhi Ding - Modern Digital and Analog Communication Systems-Oxford University Press (2009)

12.12 MATLAB Exercises 715
is known as fading. Channels that exhibit only a time-varying gain that is dependent on the
environment are known as flat fading channels. Flat fading channels do not introduce any ISI
and therefore do not require equalization. Instead, since flat fading channels generate output
signals that have time-varying strength, periods of error-free detections tend to be followed
by periods of error bursts. To overcome burst errors due to flat fading channels, interleaving
forward error correction codewords is an effective tool.
Converting Frequency-Selective Fading Channels
into Flat Fading Channels
Fast fading frequency-selective channels pose serious challenges to mobile wireless communications.
On one hand, the channels introduce ISL On the other hand, the channel characteristics
are also time varying. Although the time domain equalization techniques described in Secs. 12.3
to 12.6 can effectively mitigate the effect of ISi, they require training data to either identify
the channel parameters or estimate equalizer parameters. Generally, parameter estimation of
channels or equalizers cannot work well unless the parameters stay nearly unchanged between
successive training periods. As a result, such time domain channel equalizers are not well
equipped to confront fast changing channels.
Fortunately, we do have an alternative. We have shown (in Sec. 12.7) that OFDM can convert
a frequency-selective channel into a parallel group of flat channels. When the underlying
channel is fast fading and frequency selective, OFDM can effectively converts it into a bank
of fast flat-fading channels. As a result, means to combat fast flat-fading channels such as code
interleaving can now be successfully applied to fast frequency-selective fading channels.
We should note that for fast fading channels, another very effective means to combat the
fading effect is to introduce channel diversity. Channel diversity allows the same transmitted
data to be sent over a plurality of channels. Channel diversity can be achieved in the time
domain by repetition, in the frequency domain by using multiple bands, or in space by applying
multiple transmitting and receiving antennas. Because both time diversity and frequency diversity
occupy more bandwidth, spatial diversity in the form of multiple-input-multiple-output
(MIMO) systems has been particularly attractive recently. Among recent wireless standards,
Wi-Fi (IEEE 802. lln), WiMAX (IEEE 802.16e), and cellular LTE (long-term evolution) have
all adopted OFDM and MIMO technologies to achieve much higher data rate and better
coverage. We shall present some fundamental discussions on MIMO in Chapter 13.
12. 12 MATLAB EXERCISES
We provide three different computer exercises in this section; all model a QAM communication
system that modulates data using 16-QAM constellation. The 16-QAM signals then pass
through linear channels with ISl and encounter additive white Gaussian noise (AWGN) at the
channel output.
COMPUTER EXERCIESE 12.1 : 16-QAM LINEAR EQUALIZATION
The first MATLAB program, Exl 2_1 . m, generates 1,000,000 points of 16-QAM data for transmission.
Each QAM requires T as the symbol period. The transmitted pulse shape is a root-raised cosine with a
roll-off factor of 0.5 [Eq. (12.23)]. Thus the bandwidth at the baseband is 0.75/T Hz.
% Matlab Program <Ex12_1 .m>
% This Matlab exercise <Ex12_1 .m> performs simulation of
% linear equalization under QAM-16 baseband transmission
% a multipath channel with AWGN .