Over the past
year I’ve been
writing about
potential transmission
standards for a nextgeneration
broadcast
platform (NGBP). The
most promising standards are all based on
orthogonal frequency division multiplexing
(OFDM), specifically “Coded” OFDM
(COFDM). This month I’ll take a closer look
at the basic technology behind COFDM.
OFDM uses multiple carriers—some of
the newer standards support up to 32,000
carriers per channel—unlike the current
VSB modulation. With VSB a single carrier
is modulated with data at a high symbol
rate. In an OFDM system, the data is distributed
across thousands of carriers, with
each carrier operating at a very low symbol
rate. Fortunately, it isn’t necessary to
use a separate modulator/demodulator for
each carrier.
Orthogonality is the key to easily generating
and demodulating a COFDM signal. This is accomplished
by generating a signal
with carriers that
are evenly spaced by
precisely the reciprocal
of the active
symbol period over
which the receiver
integrates the demodulated
OFDM signal,
making it easier to
demodulate the individual
carriers.
The best explanation
I found on how
this works in practice
(with the math to
support it) is in the
tutorial “The how and
why of COFDM” by J.H.
Stott at BBC Research and
Development. He writes, “More intuitively,
what this represents is a common procedure
of demodulating a carrier by means
of multiplying it by a carrier of the same
frequency (‘beating it down to zero frequency’)
then integrating the result. Any other carriers will give rise to ‘beat tones’
which are at integer multiples of ωu. All of
these unwanted ‘beat tones’ therefore have
an integer number of cycles during the integration
period Tu, and thus integrate to
zero.” In these examples, ωu = 2π ⁄ Tu.
Stott continues, “Hence, without
any ‘explicit’ filtering, we can
separately demodulate all the carriers
without any crosstalk, just
by our particular choice for the
carrier spacing. Furthermore, we
have not wasted any spectrum
either. The carriers are closely
packed so they occupy the same
spectrum in total as would a single
carrier—if modulated with
all the data and subject to ideal
sharpcut filtering.”
GENERATING/RECEIVING
OFDM
An inverse fast Fourier transform
(IFFT) is used to generate
the COFDM signal. The modulation
process involves some complex
math. Fortunately, digital signal
processing chips are available to handle
the computations. In an OFDM transmitter,
the digital data is sent to a serialtoparallel
converter. Each of the output signals corresponds
to one carrier and is a complex value representing the location (amplitude
and phase) of the bit in the QPSK or QAM
constellation.
This complex data is sent to the IFFT,
which is then sent to a paralleltoserial
converter connected to a digitaltoanalog
converter. After applying lowpass filtering,
the OFDM is ready to be converted to the
RF channel (if needed), amplified and sent
to the antenna.

Fig. 1: Baseband OFDM System, from a presentation by Dr. Jean Armstrong, La Trobe University, Australia 
The receiver takes the IF frequency
from the tuner and, after applying lowpass
filtering, sends it to an analogtodigital
converter. The output of the A/D converter
goes to a serialtoparallel converter and
then through a DSP performing a fast Fourier
transform (FFT) to convert it back to
complex digital data.
At this point, an equalizer can be added
to reduce multipath fading before the parallel data is converted to serial data and
sent to the digital demultiplexer/decoder.
Practical COFDM systems have to consider
the propagation path between the
transmitter and receiver. There is likely to
be multipath. A zero dB echo will cause a
ripple in the pattern where carriers will be
completely attenuated. In other cases, even
those without a zero dB echo, symbols
from one path can interfere with symbols
arriving at different times over longer or
shorter paths, introducing errors.
OPTIMIZING OFDM PERFORMANCE
Mobile reception creates another problem.
As the vehicle moves, the carriers will
be “smeared” due to Doppler shift from
rapidly changing path lengths in a Rayleigh
(nonlineofsight) environment.
COFDM handles these problems in different
ways. Coding helps with the zero dB
echo case. It spreads the data over multiple
carriers. Receivers use channel state information
(CSI) to give more weight to data
from carriers with high SNR, allowing reception
even in this difficult environment.
Multipath interference is handled by
adding a cyclic prefix, which takes data
from the end of the symbol period and
inserts it before the start of the data. Any
multipath signals offset up to the length
of the cyclic prefix will not cause interference.
Since the cyclic prefix data is using
space that otherwise could have been
used for unique data, there is a tradeoff between the amount of delay a COFDM
signal can handle and the amount of data
it can handle.
The problem of Doppler shift in mobile
reception is usually handled by reducing
the number of carriers, providing wider
spacing between carriers.
All of these “fixes” interact. Finding the
right mix of cyclic prefix, number of carriers and coding is one of the challenges
engineers face in designing a COFDM
transmission system. Engineers also have
to consider the modulation of the carriers
(QPSK, 16QAM, 64QAM, 256QAM, etc.),
and the number of “pilot” carriers used to
make it easier for the receiver to acquire
the signal and correct for the propagation
environment.
Next month, in Part 2 of this article,
I’ll provide some examples showing how
these parameters can be varied with a 6
MHz COFDM transmission system that
might form the basis for the U.S. nextgeneration
broadcast platform.
Comments are welcome! Email me at
dlung@transmitter.com.