Beruflich Dokumente
Kultur Dokumente
A D VA N C E S I N S M A RT A N T E N N A S
OFDM MODULATION
MIMO technology will predominantly be used in
OMOD
broadband systems that exhibit frequency-selec-
tive fading and, therefore, intersymbol interfer-
OMOD
ODEMOD ence (ISI). OFDM modulation turns the
and
Separation frequency-selective channel into a set of parallel
flat fading channels and is, hence, an attractive
OMOD
way of coping with ISI. Figure 1 depicts the
schematic of a MIMO-OFDM system. The basic
principle that underlies OFDM is the insertion
(a) of a guard interval, called cyclic prefix (CP),
which is a copy of the last part of the OFDM
c0 r0 symbol (Fig. 1c), and has to be long enough to
c1 r1
accommodate the delay spread of the channel.
The use of the CP turns the action of the chan-
IFFT CP CP FFT
nel on the transmitted signal from a linear con-
volution into a cyclic convolution, so that the
cN-1 rN-1 resulting overall transfer function can be diago-
nalized through the use of an IFFT at the trans-
OFDM modulator OFDM demodulator mitter and an FFT at the receiver (Fig. 1b).
(b) Consequently, the overall frequency-selective
channel is converted into a set of parallel flat
fading channels, which drastically simplifies the
equalization task. However, as the CP carries
redundant information, it incurs a loss in spec-
tral efficiency, which is usually kept at a maxi-
(c)
mum of 25 percent.
In general, OFDM has tighter synchroniza-
■ Figure 1. (a) Schematic of a MIMO-OFDM system. OMOD and ODEMOD tion requirements than single-carrier (SC) mod-
denote an OFDM-modulator and demodulator, respectively; (b) single-anten- ulation and direct-sequence spread spectrum
na OFDM modulator and demodulator; (c) adding the cyclic prefix. (DSSS), is more susceptible to phase noise, and
suffers from a larger peak-to-average power
ratio. While general statements on overall imple-
mentation point-of-view comparisons of OFDM,
of the link, at no additional power or bandwidth SC, and DSSS are difficult to make, recent
expenditure [2–4]. The corresponding gain is industry trends show a clear preference for
available if the propagation channel exhibits rich OFDM-based solutions (e.g., IEEE 802.11n
scattering and can be realized by the simultane- WLANs, IEEE 802.16 WMANs, Flarion Tech-
ous transmission of independent data streams in nologies’ Flash-OFDM, and the system concept
the same frequency band. The receiver exploits developed in the context of the European FP6
differences in the spatial signatures induced by Integrated Project WINNER).
the MIMO channel onto the multiplexed data
streams to separate the different signals, thereby
realizing a capacity gain. SPACE-FREQUENCY SIGNALING IN
Diversity leads to improved link reliability MIMO-OFDM SYSTEMS
by rendering the channel “less fading” and by
increasing the robustness to co-channel inter- The signaling schemes used in MIMO systems
ference. Diversity gain is obtained by trans- can be roughly grouped into spatial multiplexing
mitting the data signal over multiple (ideally) [1], which realizes capacity gain, and space-time
independently fading dimensions in time, fre- coding [5], which improves link reliability through
quency, and space and by performing proper diversity gain. Most multi-antenna signaling
combining in the receiver. Spatial (i.e., anten- schemes, in fact, realize both spatial-multiplexing
na) diversity is particularly attractive when and diversity gain. A framework for characteriz-
compared to time or frequency diversity, as it ing the trade-off between spatial-multiplexing
does not incur an expenditure in transmission and diversity gains in flat-fading MIMO chan-
time or bandwidth, respectively. Space-time nels was proposed in [6]. In the following, we
coding [5] realizes spatial diversity gain in describe the basics of spatial multiplexing and
systems with multiple transmit antennas with- space-time coding with particular emphasis on
out requiring channel knowledge at the trans- the aspects arising from frequency-selective fad-
mitter. ing through multipath propagation and from the
Array gain can be realized both at the trans- use of OFDM.
mitter and the receiver. It requires channel
knowledge for coherent combining and results in SPATIAL MULTIPLEXING IN
an increase in average receive signal-to-noise
ratio (SNR) and hence improved coverage. MIMO-OFDM SYSTEMS
Multiple antennas at one or both sides of the The basic idea of spatial multiplexing is
wireless link can be used to cancel or reduce co- described above. It was shown in [3, 4] that the
channel interference, and hence improve cellular spatial-multiplexing gain or, equivalently, the
system capacity. number of spatial data pipes that can be
b/s
ried out for flat fading MIMO channels, it was 6
shown in [7, 8] that the corresponding results 4
are robust with respect to multipath-induced 2
frequency-selective fading. Moreover, in [8] it 0
was demonstrated that under real-world propa- 107 108 109 1010 1011 1012
gation conditions such as spatial fading corre- W (Hz)
lation (caused, e.g., by insufficient antenna
spacing), multipath propagation (leading to 8
frequency-selective fading) can be highly bene-
ficial in terms of spatial-multiplexing gain. 6
Multipath propagation tends to increase the
MT
angle spread perceived by the transmitter and 4
the receiver, which, in turn, increases the rank 2
of the channel matrix and hence the spatial-
multiplexing gain. This comes, however, at the 0
cost of increased receiver complexity due to 107 108 109 1010 1011 1012
the need to separate the multipath components W (Hz)
or, equivalently, to equalize the (ISI) MIMO
channel. ■ Figure 2. Top: capacity lower bound for 4 receive antennas and for varying
number of transmit antennas MT as a function of bandwidth W. Bottom: cor-
Spatial Multiplexing in MIMO-OFDM Systems — In an responding optimum (w.r.t. capacity) number of transmit antennas MT (with
OFDM-based MIMO system, spatial multiplex- a maximum of MT = 8). Figure taken from [20].
ing is performed by transmitting independent
data streams on a tone-by-tone basis with the
total transmit power split uniformly across the receive SNR. In the noncoherent case, where
antennas and tones. Although the use of neither the transmitter nor the receiver have
OFDM eliminates ISI, the computational com- CSI, the capacity behavior as a function of band-
plexity of MIMO-OFDM spatial-multiplexing width is markedly different: for full-band OFDM
receivers can still be high. This is because the systems (i.e., the transmit signal occupies all
number of data-carrying tones typically ranges time-frequency slots), beyond a certain critical
between 48 (as in the IEEE 802.11a/g stan- bandwidth, “overspreading” occurs, and the
dard) and 1728 (as in the IEEE 802.16e stan- capacity goes to zero. The “overspreading” phe-
dard) and spatial separation has to be nomenon was first described in [10] in the con-
performed for each tone. text of SISO systems and can be explained as
Recently, a new class of algorithms that alle- follows. Increasing the bandwidth results in a
viate this problem was proposed in [9]. The basic proportional increase in the number of indepen-
idea underlying these algorithms is to exploit the dent frequency-diversity branches (provided the
fact that the matrix-valued transfer function in a channel satisfies the uncorrelated scattering
MIMO-OFDM system is “smooth” across tones assumption). Since the receiver is not assumed
because the delay spread in the channel is limit- to have CSI, these diversity branches contribute
ed. Computational complexity reductions are to “channel uncertainty” which leads to a capaci-
obtained by performing channel inversion in the ty penalty. For large bandwidths (and hence
case of a minimum mean-squared error (MMSE) small SNR per degree of freedom) this penalty
receiver, or QR decomposition in a sphere eventually drives the capacity to zero. In the
decoder (or a successive cancellation receiver) MIMO case, increasing the number of transmit
on a subset of tones only and computing the and receive antennas, on the one hand, increases
remaining inverses or QR factors, respectively, the total number of degrees of freedom for com-
through interpolation. The resultant savings, munication and, on the other hand, results in an
compared to brute-force tone-by-tone channel increase in channel uncertainty. Since the total
inversion or QR decomposition, are proportion- available transmit power is split uniformly across
al to the number of tones divided by the product transmit antennas, increasing the number of
of the number of transmit antennas and the transmit antennas results in a smaller SNR per
channel order (upper-bounded by the length of degree of freedom which leads to the existence
the CP). In practice, a reduction in computation- of a finite optimum (in the sense of capacity
al complexity of up to 50 percent can be maximizing) number of transmit antennas.
obtained. The performance-complexity trade-off, Increasing the number of receive antennas, on
numerical properties, and memory requirements the other hand, yields an increase in the receive
of this new class of algorithms remain to be SNR and is hence always beneficial. In summary,
investigated in detail. for MIMO-OFDM systems operating at band-
widths of several GHz, such as MIMO-based
Noncoherent MIMO-OFDM Systems — With perfect ultra-wideband systems, it is generally not advis-
CSI at the receiver and no CSI at the transmit- able to use a large number of transmit antennas.
ter, and fixed transmit power, capacity increases Figure 2 provides a numerical result illustrating
with bandwidth until it saturates and is given by this phenomenon.