Professional Documents
Culture Documents
The Principles of Ofdm: Signal Processing
The Principles of Ofdm: Signal Processing
The Principles of Ofdm: Signal Processing
The principles
digital signals, the information is in the form of bits,
or collections of bits called symbols, that are modu-
lated onto the carrier. As higher bandwidths (data
rates) are used, the duration of one bit or symbol of
value for N, the length of the OFDM are of infinite length or if at least one channel. Such an equalizer can be
symbol becomes longer than the time of the signals is periodic over the expensive to implement in hardware
span of the channel. Because of this range of the convolution. It is not prac- and often requires a large number of
configuration, the effect of intersymbol tical to have an infinite-length OFDM symbols in order to adapt its response
interference is the distortion of the first symbol, however, it is possible to make to a good setting.
LC samples of the received OFDM sym- the OFDM symbol appear periodic. In OFDM, the time-domain signal is
bol. An example of this effect is shown This periodic form is achieved by still convolved with the channel
in Figure 6. By noting that only the first replacing the guard interval with response. However, the data will ulti-
few samples of the symbol are distorted, something known as a cyclic prefix of mately be transformed back into the
one can consider the use of a guard length LP samples. The cyclic prefix is frequency-domain by the FFT in the
interval to remove the effect of inter- a replica of the last LP samples of the receiver. Because of the periodic nature
symbol interference. The guard interval OFDM symbol where LP > LC. Since it of the cyclically-extended OFDM sym-
could be a section of all zero samples contains redundant information, the bol, this time-domain convolution will
transmitted in front of each OFDM cyclic prefix is discarded at the receiv- result in the multiplication of the spec-
symbol. Since it does not contain any er. Like the case of the guard interval, trum of the OFDM signal (i.e., the fre-
useful information, the guard interval this step removes the effects of inter- quency-domain constellation points)
would be discarded at the receiver. If symbol interference. Because of the with the frequency response of the
the length of the guard interval is prop- way in which the cyclic prefix was channel. The result is that each subcar-
erly chosen such that it is longer than formed, the cyclically-extended OFDM rier’s symbol will be multiplied by a
the time span of the channel, the symbol now appears periodic when complex number equal to the channel’s
OFDM symbol itself will not be distort- convolved with the channel. An impor- frequency response at that subcarrier’s
ed. Thus, by discarding the guard inter- tant result is that the effect of the frequency. Each received subcarrier
val, the effects of intersymbol interfer- channel becomes multiplicative. experiences a complex gain (amplitude
ence are thrown away as well. In a digital communications system, and phase distortion) due to the chan-
the symbols that arrive at the receiver nel. In order to undo these effects, a fre-
Intrasymbol interference have been convolved with the time- quency-domain equalizer is employed.
The guard interval is not used in domain channel impulse response of Such an equalizer is much simpler than
practical systems because it does not length LC samples. Thus, the effect of a time-domain equalizer. The frequen-
prevent an OFDM symbol from inter- the channel is convolutional. In order cy-domain equalizer consists of a sin-
fering with itself. This type of interfer- to undo the effects of the channel, gle complex multiplication for each
ence is called intrasymbol interfer- another convolution must be per- subcarrier. For the simple case of no
ence. The solution to the problem of formed at the receiver using a time- noise, the ideal value of the equalizer’s
intrasymbol interference involves a domain filter known as an equalizer. response is the inverse of the chan-
discrete-time property. Recall that in The length of the equalizer needs to be nel’s frequency response. An example
continuous-time, a convolution in time on the order of the time span of the is shown in Figure 7. With such a set-
is equivalent to a multiplication in the channel. The equalizer processes sym- ting, the frequency-domain equalizer
frequency-domain. This property is bols in order to adapt its response in would cancel out the multiplicative
true in discrete-time only if the signals an attempt to remove the effects of the effect of the channel.
COFDM: Coded OFDM and those bits would not be weighted as quency shift of the received signal spec-
Coded OFDM, or COFDM, is a term heavily as bits from a strong subcarrier. trum. This shift causes a condition
used for a system in which the error COFDM systems are able to achieve called “loss of orthogonality” to occur.
control coding and OFDM modulation excellent performance on frequency- The frequency shift causes the OFDM
processes work closely together. An selective channels because of the com- subcarriers to no longer be orthogonal.
important step in a COFDM system is bined benefits of multicarrier modula- The orthogonality of the subcarriers is
to interleave and code the bits prior to tion and coding. lost because the bins of the FFT will no
the IFFT. This step serves the purpose longer line up with the peaks of the
of taking adjacent bits in the source Non-ideal effects in an received signal’s since pulses. The
data and spreading them out across OFDM system result is a distortion called inter-bin
multiple subcarriers. One or more sub- This section will examine the effects interference or IBI. IBI occurs when
carriers may be lost or impaired due to of non-idealities in an OFDM system. energy from one bin spills over into
a frequency null, and this loss would These effects will include impairments adjacent bins and this energy distorts
cause a contiguous stream of bit errors. and receiver offsets. Because the fourier the affected subcarriers. In Fourier
Such a burst of errors would typically transform is a fundamental operation in Transform theory this effect is called
be hard to correct. The interleaving at OFDM, the effects of several offsets can DFT leakage.
the transmitter spreads out the contigu- be intuitively understood by applying The left plot of Figure 8 shows the
ous bits such that the bit errors become fourier transform theory. spectrum of a received OFDM signal
spaced far apart in time. This spacing with no LO offset. For the purpose of
makes it easier for the decoder to cor- Local oscillator clarity, only one non-zero subcarrier
rect the errors. Another important step frequency offset was transmitted. Note that this subcar-
in a COFDM system is to use channel At start-up, the local oscillator (LO) rier is not interfering with its adjacent
information from the equalizer to deter- frequency at the receiver is typically dif- subcarriers. The spectrum of the non-
mine the reliability of the received bits. ferent from the LO frequency at the zero subcarrier actually extends over
The values of the equalizer response transmitter. A carrier tracking loop is the entire range of the FFT, however,
are used to infer the strength of the used to adjust the receiver’s LO fre- due to the orthogonal nature of the sig-
received subcarriers. For example, if the quency in order to match the transmit- nal, the zero-crossings of the spectrum
equalizer response had a large value at ter’s LO frequency as closely as possi- exactly line up with the other FFT bins.
a certain frequency, it would correspond ble. The effect of having an LO frequen- The right plot of Figure 8 shows the
to a frequency null at that point in the cy offset can be explained by Fourier received spectrum of the same signal
channel. The equalizer response would Transform theory. The LO offset can be with one non-zero subcarrier, however,
have a large value at that point because expressed mathematically by multiply- in this case there is an LO offset. This
it is trying to compensate for the weak ing the received time-domain signal by offset has resulted in a loss of orthogo-
received signal. This reliability informa- a complex exponential whose frequency nality, and the zero-crossings of the
tion is passed on to the decoding blocks is equal to the LO offset amount. Recall non-zero subcarrier’s spectrum no
so that they can properly weight the from Fourier Transform theory that longer line up with the FFT bins. The
bits when making decoding decisions. multiplication by a complex exponential result is that energy from the non-zero
In the case of a frequency null, the bits in time is equivalent to a shift in fre- subcarrier is spread out among all of
would be marked as “low confidence” quency. The LO offset results in a fre- the other subcarriers, with those sub-
Impulse noise
Impulse noise is a common impair-
ment in a communications system aris-
ing from motors or lightning. Impulse
noise is typically characterized as a
short time-domain burst of energy. The
burst may be repetitive or may be a sin-
gle event. In either case, the frequency
spectrum from this energy burst is
wideband, typically much wider than
the channel, but is present for only a
Figure 10. Illustration of the effect of a sampling frequency offset.
short time period.
One of the most important concepts
trum. If the spectrum expands too OFDM system is similar to its effect on to understand about OFDM and its
much, aliasing of the spectrum can a single carrier system. The signal-to- properties related to the FFT algorithm
occur. Either type of sampling frequen- noise ratio (SNR) is a function of the is how the algorithm changes the
cy offset results in IBI since the expan- total signal power over the total noise nature of the signal. In a single-carrier
sion or contraction of the spectrum pre- power across the received channel. The system, the symbol can be viewed as
vents the received subcarriers from lin- uniform noise contributes to the SNR of occupying all of the available frequency
ing up with the FFT bin locations. The each subcarrier in the OFDM system spectrum for the time duration of the
effect of sampling too fast is illustrated and the net result is equivalent to the symbol. A group of symbols then occu-
in Figure 10 and simulation results to effect on single channel systems. pies all of the spectrum for the duration
demonstrate this effect are shown in of the whole group, but in a time divi-
Figure 11. A sampling frequency offset Non-uniform noise sion arrangement.
can be corrected by generating an error Noise in a communications channel OFDM, using the FFT, takes symbols
term that is used to drive a sampling can often be shaped, or “colored”, by and creates these groups directly and
rate converter. various effects. These effects can then transforms them. They are no
include transmit signal imperfections, longer time-domain multiplexed, they
Uniform noise transmission channel characteristics, or are now frequency-domain multiplexed.
Additive white Gaussian noise receiver frequency shaping. The impli- The OFDM symbol is now a collection of
(AWGN) is the most common impair- cations of these effects for an OFDM these source symbols, and this OFDM
ment encountered in a communications system can be different compared to its symbol now has a much longer dura-
tion. Each original symbol occupies only
a small frequency region, but now occu-
pies that region for the entire OFDM
symbol duration. Figure 13 illustrates
this concept. For impulses that are
short in duration, the impulse energy
masks a smaller percentage of time of
each OFDM symbol compared to the
single carrier case. Impulse noise can
therefore have less of an effect on short
duration noise.
Carrier interference
Single-carrier interference arises
Figure 11. Simulation results showing the effect of a sampling frequency that is too high. Note that the
sample that was originally at bin 15 is now at bin 8.
from other sources that may co-exist in
the frequency range of interest. These
can be generated by nearby circuits or
system. In a wireless medium, the noise single-carrier counterpart. The modula- other transmission sources. The single
source is typically considered to be ther- tion of the OFDM system can be tai- carrier system must handle this inter-
mal noise that is Gaussian and uniform lored for the noise characteristics. One ference as a noise source for all informa-
across the frequency range. Additional method previously mentioned involves tion sent. The OFDM system can avoid
noise sources include atmospheric lowering the modulation (number of the frequency region of interference by
sources and solar radiation. In a con- bits/symbol) on subcarriers in a low disabling or turning off the affected sub-
tained media, such as a coaxial cable SNR environment as illustrated in carriers. Narrowband modulated
system, thermal noise will be present, Figure 12. Another method involves sources of interference can be consid-
Modern applications
OFDM has been chosen for several
current and future communications sys-
tems all over the world. It is well-suited
for systems in which the channel char-
acteristics make it difficult to maintain
adequate communications link perfor-
mance. Asynchronous digital subscriber
line (ADSL) provides a method of deliv-
ering high speed data over the phone
line. The system uses OFDM tech-
niques, calling their variation discrete
Figure 13. Comparison of single carrier versus OFDM spectrum. multi-tone (DMT). DMT includes fea-