Digital Design of Communication Systems - Digital Design of Signal Processing Systems: A Practical Approach - page 567

% OFDM modulation
ofdm_mod_out = [];
% Generate OFDM symbols for one burst of transmission
for index=0:ofdm_burst_size-1,
ifft_in = mod_out(index*N+1:(index+1)*N);
ifft_out = ifft(ifft_in, N);
ifft_out_cp = [ifft_out(end-K+1:end) ifft_out];
ofdm_mod_out=[ofdm_mod_out ifft_out_cp];
end
A burst is formed that appends a training sequence of Figure 13.41 to the OFDM data. A number of
OFDM symbols are sent in a burst. This number depends on the coherent bandwidth of the channel
that ensures the flatness of the channel on the entire bandwidth of transmission. The training is used
to estimate different parameters at the receiver:
% Generating training for start of burst, synchronization and estimation
% Generate the training offline and store in memory for
% appending it with input data
C=[11-11111-1];
A = fft(C);
Sign_Pattern = [-1 1 -1 -1];
G = [];
for i=1:L
G = [G Sign_Pattern(i)*A];
end
% Add cyclic prefix to G
train_sym = [G(end-K+1:end) G]/4;
% Appending training to start of burst
ofdm_mod_out = [train_sym ofdm_mod_out];
The signal is transmitted after it is up-converted and mixed with a carrier. In a non-line-of-sight
mobile environment the transmitted signal gets to the receiver through multiple paths. If there is a
relative velocity between transmitter and receiver, the received signal frequency also experiences
Doppler shift. Themulti-path effects cause frequency fading. The signal also suffers from timing and
frequency offsets. The channel impurities are modeled using a Rayleigh fading channel where
timing and frequency offsets are also added to test the effectiveness of estimation and recovery
techniques in the receiver. An additive white Gaussian noise (AWGN) is also added in the signal to
test the design for varying signal to noise ratios (SNRs):
% Multi-path fading frequency selective channel model
ts = 1/(186.176e3); % sample time of the input signal
fd = 5; % maximum Doppler shift in Hz
tau = [0 8e-6 15e-6]; % path delays
pdb = [0 -10 -20]; % avg path gain
theta = 0.156; % frequency offset
SNR = 5; & signal to noise ratio in dbs
timing_offset = 18;
chan = rayleighchan(ts,fd,tau,pdb);