OFDM is
conceptually simple, but the devil is in the details! The implementation
relies on very high speed digital signal processing and this has only
recently become available at a price that makes OFDM a competitive technology
in the marketplace.
OK, so what is the simple concept behind OFDM? Take one carrier
and modulate it using Quadrature Phase Shift Keying (QPSK) where each
symbol encodes 2 bits. This modulation is at a certain symbol rate. For
the purposes of this discussion let's say 1000 symbols per second.
Modulation theory tells us that the spectrum of such a modulated
signal will have a sin(x)/x shape with the first null at 1000 Hz. Now
if we have a second carrier that has a frequency exactly 1 KHz higher
than the first, and modulate it with the same symbol rate, it turns out
that both signals can be recovered without mutual interference.
To make the whole exercise worth while, take the preceeding paragraph
and multiply it by a factor of, maybe, 256 or even more. And while you
are at it, instead of using a 2-bit symbol (QPSK), use a 6-bit symbol
(64-QAM). This can cram an amazing amount of data into a relatively small
bandwidth.
The problem with this simple-minded approach is that it takes lots
of local oscillators each locked to the others so that the frequencies
are the exact multiples that they should be. This is difficult and expensive.
DSP to the rescue! Each of the oscillators can be a digital representation
of the sine carrier wave that can be modulated in the numerical domain.
This can happen simultaneously for all of the carriers. The resulting
output of each channel is added and then blocked. Since we have a representation
of the signal in the frequency domain but need to modulate an actual carrier
in the time domain, we just perform an Inverse Fast Fourier Transform
(IFFT) to convert the block of frequency data to a block of time data
that modulates the carrier.The receiver acquires the signal, digitizes
it, and performs an FFT on it to get back to the frequency domain. From
there, it is relatively easy to recover the modulation on each of the
carriers.
In practice, some of the carriers are used for channel estimation
and there are extra bits added for error detection and correction. Doing
this is called Coded Orthogonal Frequency Division Multiplexing (COFDM).
Coding is now so common that many people drop the "C", as unnecessary,
assuming that coding is used.
In
FDMA each user is typically allocated a single channel, which is used
to transmit all the user information. The bandwidth of each channel is
typically 10kHz-30kHz for voice communications. However, the minimum required
bandwidth for speech is only 3kHz. The allocated bandwidth is made wider
then the minimum amount required to prevent channels from interfering
with one another. This extra bandwidth is to allow for signals from neighboring
channels to be filtered out, and to allow for any drift in the center
frequency of the transmitter or receiver. In a typical system up to 50%
of the total spectrum is wasted due to the extra spacing between channels.
This problem becomes worse as the channel bandwidth becomes narrower,
and the frequency band increases.
Most
digital phone systems use vocoders to compress the digitized speech. This
allows for an increased system capacity due to a reduction in the bandwidth
required for each user. Current vocoders require a data rate somewhere
between 4- 13kbps, with depending on the quality of the sound and the
type used. Thus each user only requires a minimum bandwidth of somewhere
between 2-7kHz, using QPSK modulation. However, simple FDMA does not handle
such narrow bandwidths very efficiently.
TDMA partly overcomes this problem by using wider bandwidth channels, which are used by several users. Multiple users access the same channel by transmitting in their data in time slots. Thus, many low data rate users can be combined together to transmit in a single channel which has a bandwidth sufficient so that the spectrum can be used efficiently.
There
are however, two main problems with TDMA. There is an overhead associated
with the change over between users due to time slotting on the channel.
A change over time must be allocated to allow for any tolerance in the
start time of each user, due to propagation delay variations and synchronization
errors. This limits the number of users that can be sent efficiently in
each channel. In addition, the symbol rate of each channel is high (as
the channel handles the information from multiple users) resulting in
problems with multipath delay spread.
OFDM overcomes most of the problems with both FDMA and TDMA. OFDM splits the available bandwidth into many narrow band channels (typically 100-8000). The carriers for each channel are made orthogonal to one another, allowing them to be spaced very close together, with no overhead as in the FDMA example. Because of this there is no great need for users to be time multiplex as in TDMA, thus there is no over head associated with switching between users. The orthogonality of the carriers means that each carrier has an integer number of cycles over a symbol period. Due to this, the spectrum of each carrier has a null at the centre frequency of each of the other carriers in the system. This results in no interference between the carriers, allowing then to be spaced as close as theoretically possible. This overcomes the problem of overhead carrier spacing required in FDMA. Each carrier in an OFDM signal has a very narrow bandwidth (i.e. 1kHz), thus the resulting symbol rate is low. This results in the signal having a high tolerance to multipath delay spread, as the delay spread must be very long to cause significant inter-symbol interference (e.g > 500usec).
Another advantage of OFDM is its ability to handle the effects of multipath delay spread. In any radio transmission, the channel spectral response is not flat. It has fades or nulls in the response due to reflections causing cancellations of certain frequencies at the receiver for narrowband transmissions. If the null in the frequency occurs at the transmission frequency then the entire signal can be lost. Multipath delay spread can also lead to intersymbol interference. This is due to a delayed multipath signal overlapping with the following symbol. This problem is solved by adding a time domain guard interval to each OFDM symbol. Intercarrier interference (ICI) can be avoided by making the guard interval a cyclic extension of the OFDM symbol.
There are, however, certain negatives associated with this technique. It is more sensitive to carrier frequency offset and sampling clock mismatch than single carrier systems. Also the nature of the orthogonal encoding leads to high peak-to-average ratio signals: or in other words, signals with a large dynamic range. This means that only highly linear, low efficiency RF amplifiers can be used.
