Académique Documents
Professionnel Documents
Culture Documents
M
ultiple Input beamforming techniques that may optimize
This article examines the Multiple Output system performance under various channel
basic theory of precoding (MIMO) technol- conditions. One technique, referred to as pre-
in a MIMO system, com- ogy has been shown to coding, is designed to increase and/or equalize
pares this with non-precod- provide higher data rates the received SINR across the multiple receive
ed MIMO, and describes with increased spectral antennas.
methods for measuring efficiency [1, 2]. The per-
system performance formance of a MIMO sys- MIMO and Precoding
tem is directly related to A standard 2 2 MIMO spatial multiplex-
the received signal-to-interference-and-noise ing scheme, shown in Figure 1a, assumes the
ratio (SINR) and the correlation properties wireless channel will provide four separate
that are characteristic of the multipath chan- connections between transmit and receive
nel and antenna configuration [3]. Although antennas. Each channel connection, shown as
the wireless channel can deliver low SINR at an arrow in the figure, represents a unique
some of the MIMO receive antennas, it is pos- combination of all transmission paths includ-
sible to improve system performance with the ing the direct Line of Sight (LOS) path, should
application of beamforming at the transmit- one exist, and the numerous multipaths creat-
ter. Though often used together, it is important ed by reflection, scattering and diffraction
to differentiate here that beamforming is a from the surrounding environment.
signal processing technique, which is very dif- Depending on the resulting channel condi-
ferent from beam steering where the direction tions, the MIMO system may not be able to
of the main lobe of radiation is changed. properly recover the transmitted data streams
Beamforming lends itself well to MIMO appli- (layers) if the SINR is too low at any of the
cations.The 3GPP Long Term Evolution (LTE) receive antennas. With the addition of precod-
specification [4] includes several transmit ing, as shown in Figure 1b, the transmitter,
Figure 1 Simplified block diagram showing the difference between (a) MIMO without pre-
coding and (b) MIMO with precoding.
having knowledge of the current sums and differences of the two input tions may also limit the number of
channel conditions, can effectively layers respectively. The weighting possible adjustments thus reducing
combine the layers before transmis- matrix for codebook 1 is the effectiveness of precoding.
sion with the goal of equalizing the An LTE system design requires a
signal reception across the multiple 1 1 1 thorough understanding of the trade-
W ( i) =
receive antennas. Precoding schemes 2 1 1 offs between performance, precoding
have been specified for spatially-mul- options and feedback constraints. A
tiplexed and transmit-diversity resulting in the following transmit- flexible measurement system can
applications [4]. This paper will ted data provide unique insight into the capa-
examine precoding in spatially-multi- bility of precoding under a variety of
1 (0) 1
plexed MIMO systems. y(0) ( i) =
x ( i) + x(1) ( i) simulated channel conditions includ-
Precoding is based on transmit 2 2 ing noise, interference and antenna/
beamforming concepts with the pro- 1 (0) 1 (1) channel correlation.
y ( i) = x ( i) x ( i)
(1)
vision of allowing multiple beams to 2 2
be simultaneously transmitted in the Precoding Measurement
MIMO system. The LTE specification This codebook selection allows a Examples
defines a set of complex weighting portion of each signal layer to be There are a variety of measure-
matrices for combining the layers transmitted through each antenna, ment tools that are necessary when
before transmission using up to 4 4 and depending on the channel condi- examining the performance of pre-
antenna configurations [4]. For a tions, providing some flexibility when coding and MIMO operating under
2 2 configuration, the weighting attempting to improve and equalize diverse channel conditions. Figure 2
matrix, W, is multiplied by the input the SINR at each MIMO receiver. shows a typical 2 2 MIMO mea-
layers to generate the precoded sig- The LTE specification for precod- surement setup consisting of a wire-
nals to be transmitted. ing spatially multiplexed transmis- less channel emulator, signal sources
sions includes a total of four codebook and signal analyzers. The channel
y(0) ( i) x(0) ( i) matrices for two transmit antenna emulator, such as the Agilent N5106
(1) = W ( i) (1) configurations and 16 codebook PXB Receiver Tester, is used to create
y ( i) x ( i) matrices for four transmit antenna realistic multi-channel, multipath
systems. Proper selection of the opti- environments including the effects of
Here, x(q)(i) are the input layers mal precoding matrix requires knowl- antenna and spatial correlations [3].
prior to precoding (q = 0, 1) and y(q)(i) edge of the current channel conditions Some commercial emulators have
are the precoded signals applied to at the transmitter. The channel condi- built-in baseband generators for gen-
each transmit antenna. The simplest tions are provided through feedback erating complex waveforms using
precoding matrix maps each layer to from the MIMO receiver creating a standards based or custom models
a single antenna dedicated to trans- closed-loop system. For an LTE pre- developed in software tools such as
mitting that layer, without any cou- coded downlink transmission, the Agilent Signal Studio, Agilent ADS
pling to other antennas. In this case, mobile terminal or user equipment and Agilent SystemVue. The outputs
the weighting matrix, defined with (UE) will measure the channel char- of the channel emulator are complex
codebook index 0, becomes acteristics and determine the precod- baseband waveforms representing
ing matrix index (PMI), channel qual- the precoded MIMO signals modified
1 1 0 ity indicator (CQI) and/or Rank Index by multipath, noise and/or interfer-
W ( i) =
2 0 1 (RI). This information will be sent to ence. These baseband waveforms are
the base station (eNB) which would then modulated onto RF carriers
resulting in the following transmit- modify the precoding codebook selec- using the analog in-phase (I) and
ted data as tion to improve overall system perfor- quadrature-phase (Q) inputs avail-
mance. As channel conditions may able on many RF vector signal gener-
1 (0)
y(0) ( i) = x ( i) change rapidly over time, it is impor- ators. The baseband data can also be
2 tant for the system to avoid excessive modulated onto RF carriers using
1 (1) delays when closing the feedback digital I and Q inputs on signal gen-
y(1) ( i) = x ( i)
2 loop. Reduction of the signaling over- erators such as the Agilent MXG
head and associated feedback delay is series. This is the preferred method
A second precoding matrix, accomplished by restricting the num- since it gives the best performance
defined with codebook index 1, pro- ber of codebook selections. Unfortu- and also enables automatic power
vides a linear combination of the nately, reducing the number of selec- calibration of the system to be per-
Figure 2 Measurement setup for characterizing the performance of LTE precoding under various multipath and
channel conditions as demonstrated using the N5106A PXB MIMO Receiver Tester, theN5182A MXG Vector Signal
Generator and the N9020A MXA Signal Analyzer.
formed In the measurement system RF signal sources, such as the Agilent measurement ports available on the
shown in Figure 2, the two RF signal MXG signal generators, phase coher- test equipment, there are several pos-
generators are the inputs to a two- ence is maintained by sharing the sible configurations for connecting a
channel MIMO receiver. It is impor- un-modulated local oscillator (LO) MIMO transmitter to the signal ana-
tant to note that when using multiple from one generator with the other as lyzer(s). For, example, limited MIMO
signal generators to simulate a shown in Figure 2. In some test sys- testing can be done with a single-
MIMO system, the generators tems with multiple RF signal genera- input analyzer by using, a power
require a stable phase relationship tors, such as in 4 4 and 2 4 config- combiner to add the multiple signals
during test though the equipment is urations, a separate RF signal gener- from the MIMO transmitters to the
not required to be phase-locked. The ator is recommended as a master common port on the analyzer. In this
term phase-locked is often referred to local oscillator in order to provide case, since the transmitted downlink
as "phase coherence" and describes a enough drive level to the local oscilla- reference signals are orthogonal in
fixed phase condition between the RF tor inputs of the signal generators. frequency and/or time, the individual
outputs of two or more signal genera- In the measurement example reference signals from each transmis-
tors operating at a specified carrier shown in Figure 2, the two-channel sion antenna port can be analyzed for
frequency. A proper phase relation- MIMO receiver is configured using EVM characteristics and timing
ship is important for the precoding two vector signal analyzers (VSA), errors using the single-input analyz-
operation as signals from separate such as Agilent MXA signal analyz- er. When two single-input analyzers
data layers are vectorially added ers configured with Agilent 89600 are available for test, a two-channel
before transmission based on known VSA software. A similar configura- MIMO transmitter can be directly
channel conditions. If the signal gen- tion can be used to test the perfor- cabled to the analyzers. In this case,
erators used to simulate multiple mance of an actual 2 2 MIMO the analyzers can recover the inde-
transmitters have an unknown receiver system by direct cable con- pendent data from each codeword
and/or time-varying phase relation- nection from the two signal genera- even for cases when precoding is
ship, the received signals may tors to the inputs of the MIMO applied such that each layer contains
include undesired phase offsets receiver. In this case, the channel some combination of each indepen-
resulting in reduced performance in emulator introduces multipath and dent codeword. This configuration is
one or more of the recovered data channel impairments that would be also useful for evaluating the effects
streams. An example will be provided found in a real-world environment. of the propagation channel where
later in this article showing the rela- When testing a MIMO transmitter or cross talk and cross coupling of the
tionship between signal generator eNB, the transmitter can be directly channels will occur.
phase offset and system performance. connected to the signal analyzer(s). A measurement example of the
In a test system using two modern Depending on the total number of potential system improvements that