Tsoulos George (ред.) MIMO System Technology for Wireless Communications

Подождите немного. Документ загружается.

214 MIMO System Technology for Wireless Communications

8.2.2 UTRA FDD Uplink

Figure 8.2 presents uplink frame and time slot structures for DPDCH and

DPCCH. In uplink, user and control data are I/Q code multiplexed and several

DPDCHs can be associated to a single DPCCH. In addition to pilot bits, the

TFCI and TPC uplink DPCCH slot contains feedback information bits (FBI)

used to convey partial channel state information when UTRA two-antenna

closed-loop transmit diversity modes are applied. Only user-speciﬁc dedi-

cated pilot channels are available in uplink.

Uplink users are not synchronized, and due to the non-orthogonality of

users’ channelization codes, multiuser interference cannot be avoided. Accu-

rate and fast TPC is indispensable to uplink performance, because otherwise

users in the vicinity of the base station would completely mask intracell

users on cell edges. Fast power control is an inherent characteristic of CDMA

systems, and it is applied in most of the downlink and uplink data channels

in UTRA FDD.

8.3 Current Multiantenna Methods in UTRA FDD

Conventionally, multiantenna techniques have mainly been considered

within base stations, because deploying multiple antennas in the user equip-

ment is not straightforward due to cost, complexity of signal processing, and

power consumption. This is still true in the present terminals that may

support several radios like WCDMA, GSM, GPS, Bluetooth, WLAN. Thus,

handsets that support only SISO processing may still require several anten-

nas for different frequency bands, although the current trend is to handle

the different frequencies with a single multifrequency antenna.

Multiantenna signal processing is useful in base station receivers, but the

use of multiple antennas for transmission requires careful study. All precod-

ing methods that modify the transmitted signal constellation must be stan-

dardized to ensure that all mobile receivers are able to estimate and detect

the signals from different antennas. An extensive standardization process

has been required even in the case of transmit beamforming.

FIGURE 8.2

Uplink frame structure for DPDCH/DPCCH.

DPDCH

adio frame = 15 time slots

10 ms

Data bits

Pilot bits TPC TFCI FBI

DPCCH

4190_book.fm Page 214 Tuesday, February 21, 2006 9:14 AM

Multiuser MIMO for UTRA FDD

215

Present UTRA FDD speciﬁcation supports two-antenna transmit diversity

and transmit beamforming. Such MISO systems for downlink have been

intensively studied, because at the time when the 3GPP standardization was

initiated, most scenarios predicted that the capacity of wireless networks

would be limited by the downlink connection. Although diversity MISO

does not promise such high peak data rates as information MIMO, transmit

diversity improves fading resistance and beamforming increases the system

capacity while its impact to individual user data rates is small. Information

MIMO techniques are not included in the UTRA FDD speciﬁcation, mostly

because the initial standardization work was carried out during the 1990s

when the development of practical information MIMO algorithms was still

only beginning.

8.3.1 Beamforming

Conventional transmit beamforming provides the simplest approach to

increase system capacity and coverage in downlink. The current 3GPP

speciﬁcation supports both ﬁxed beamforming (FBF) and user speciﬁc beam-

forming (USBF) modes in downlink. Fixed beamforming introduces addi-

tional sectors to the cell, because the short term variation of the beams is

small and a handover between the beams is required for mobile users. Each

beam is associated with a unique S-CPICH while P-CPICH is transmitted to

the whole cell.

Fixed beamforming is able to relax the code limitation in downlink, because

orthogonal spreading codes can be reused within the same cell due to spatial

ﬁltering. It is also possible to use different scrambling codes in different beams.

In the presence of ﬂat fading, the channelization codes of different users

within the same beam remain orthogonal because they are scrambled with

the same code. Different beams interfere with one another and increase

intracell interference, but when angular spread (AS) is reasonably small as

in macrocells, this interference remains acceptable. In addition to larger

interference levels, control signaling overhead increases due to inter-beam

handovers, because logically the ﬁxed beams behave in the same way as

different sectors in base stations. The overall effect on system capacity is

positive, though, and according to [14], a 2.4-fold system capacity gain can

be achieved with a four-element antenna array when compared to a single

antenna BS.

Instead of ﬁxed transmit beams, USBF generates individual beams to each

user. More sophisticated signal processing algorithms can be used, and con-

sequently, USBF is able to provide higher individual data rates than FBF. In

practice, USBF cannot be utilized in most cases within UTRA FDD downlink,

because it prevents the user from employing the P-CPICH as a phase refer-

ence, and according to the current speciﬁcation, S-CPICH cannot be

employed in individual beams. Thus, channel estimation must be based on

dedicated pilot channels whose transmit power is low, which may seriously

4190_book.fm Page 215 Tuesday, February 21, 2006 9:14 AM

216 MIMO System Technology for Wireless Communications

corrupt channel estimation in UE. Then only high data rate users near the

base station may take full advantage from USBF. User-speciﬁc beamforming

is optional for UE in HS-DSCH and, therefore, the network cannot assume

that all UE users are able to support it.

Downlink beamforming changes the statistics of the fading signal, and in

environments with small angular spread, array gain dominates diversity

gain. Therefore, the estimation of downlink beamforming gain over single

antenna transmission is rather straightforward: The gain is approximately

the same as the array gain. This is not as simple, though, in uplink.

In uplink, both UTRA FDD beamforming techniques can be implemented

in a straightforward manner in the receiver. The simplest approach is to com-

bine the selected signal paths using maximal ratio combining (MRC) in the

beam space. This leads to a standard Rake-receiver concept. The most chal-

lenging practical problems are related to beam selection (direction of arrival

(DoA) estimation in USBF) and cost-efﬁcient receiver structures. The former

problem is related to the large variety of physical environments with different

channel proﬁles. Furthermore, mobile’s transmit power is typically low for

low data rate connections, making the channel estimation difﬁcult. The latter

problem arises from the fact that baseband complexity increases rapidly when

spatio-temporal estimation processes are introduced. For example, preambles

of random access channel (RACH) need to be monitored simultaneously for

each beam in order to avoid additional delays in connection setup time.

Transmit beamforming improves the downlink capacity in UTRA FDD,

but it is important to note that receive beamforming is not necessarily the

best solution in uplink due to the lack of diversity gain [15]. This gives rise

to a tradeoff between uplink and downlink design targets. If downlink

capacity is the bottleneck, FBF provides a good solution in macrocells. On

the other hand, if good uplink coverage is the primary target, uncorrelated

antennas with suitable receiver algorithms provide a better solution.

8.3.2 Transmit Diversity

According to [15], basic receive diversity solution outperforms receive beam-

forming performance even when angular spread is small. Moreover, transmit

beamforming in downlink loses its good performance in terms of capacity

and coverage when AS becomes large, so that beams cannot be accurately

pointed to users anymore. Hence, there is a need for a multiantenna trans-

mission method in downlink that performs well when correlation between

the transmit antennas is low. In general, low correlation can be achieved

when the distance between antenna elements in the base station is several

wavelengths. Conventional beamforming algorithms do not apply anymore,

and to this end, several open-loop transmit diversity techniques have been

developed in recent years. The simplest space–time block code [16] has

been adopted into 3GPP speciﬁcation as a two-antenna open-loop transmit

diversity method. In UTRA FDD parlance, the scheme is referred to as

space–time transmit diversity (STTD).

4190_book.fm Page 216 Tuesday, February 21, 2006 9:14 AM

Multiuser MIMO for UTRA FDD

217

Space–time codes provide diversity reducing the variance of the received

signal, which further translates into reduced transmit power in downlink.

However, space–time codes do not increase the received SNR when com-

pared to single antenna transmission. With uncorrelated transmit antennas,

received SNR can be improved when short-term channel state information

(CSI) is made available in the transmitter. Such transmit diversity algorithms

are referred to as closed-loop transmit-diversity methods. The current UTRA

FDD speciﬁcation contains two closed-loop modes [4], where the relative

phase and power between the two transmit antennas is adjusted according

to feedback information bits signaled from mobile to a base station through

a dedicated control channel. (See Figure 8.2.)

UTRA FDD supports two P-CPICHs in order to enable efﬁcient channel

estimation of the two channels when transmit diversity is in use. In case of

closed-loop modes, feedback commands are determined from the P-CPICHs

and transmit weights are applied on dedicated trafﬁc channels. Common

pilots are transmitted with higher power than dedicated pilots, and there-

fore, common pilots should be used for channel estimation whenever pos-

sible. With closed-loop modes, dedicated pilots are used for the veriﬁcation

of the transmit weights (feedback commands may be erroneously decoded

in the base station, and the mobile cannot automatically assume that the

transmitter uses the weights that the mobile has instructed), and channel

estimates can still be calculated from P-CPICHs.

Closed-loop techniques outperform open-loop ones, particularly within

low-mobility environments, when the delay of the feedback signaling does

not exceed channel coherence time. Moreover, contrary to conventional

beamforming techniques, closed-loop algorithms do not require accurate

calibration of the transmit antennas due to feedback information. Only

coarse calibration of antenna elements is necessary in order to avoid spurious

antenna gain patterns and unexpected interference to the network.

The discussion on transmit diversity extensions to more than two transmit

antenna systems is ongoing within 3GPP [5,7]. The main concern is that the

beneﬁts from additional transmit antennas may not justify the additional

complexity of transceiver and system design. Firstly, transmit diversity

increases the robustness of the link, but it does not increase data rate. Secondly,

a difﬁcult backward compatibility with former releases is encountered if

transmit power of common pilot channels is divided between, say, four trans-

mit antennas to support channel estimation in UEs. Then legacy UEs that

are able to receive only two pilot channels lose performance, because they

lose half of the available power of the common pilot signals.

8.4 MIMO in UTRA FDD Uplink

MIMO discussion in 3GPP has focused downlink, but the introduction of

new services such as videophones will make it extremely important to reach

4190_book.fm Page 217 Tuesday, February 21, 2006 9:14 AM

218 MIMO System Technology for Wireless Communications

high spectral efﬁciency also in the uplink direction. Besides this service-based

demand, there are also two apparent reasons why MIMO in UTRA FDD

uplink is attractive and, thus, worthy of a closer study. First, base stations

will have two or more antennas in the near future, and at the same time, two-

antenna mobiles, which are at least capable of interference cancellation, will

become more common. Second, in contrast to UTRA FDD downlink, only

minor changes to the UTRA FDD speciﬁcations are needed to enable simple

uplink MIMO approaches.

8.4.1 MIMO Algorithms

We divide MIMO techniques into two main classes — diversity and infor-

mation MIMO. UTRA FDD uplink provides a good means to implement

diversity and information MIMO techniques, because uplink users are sep-

arated by different scrambling codes and, therefore, the whole channelization

code space is available for any one user.

8.4.1.1 Diversity MIMO

In diversity MIMO, M

replicas of the same data stream are transmitted by

using different orthogonal channelization codes but the same scrambling

code. This is depicted in Figure 8.3, which shows the system model of

diversity MIMO for two transmit antennas. The received signal in k

th base

station antenna becomes

(8.1)

FIGURE 8.3

System model for diversity MIMO.

yt h t

StjS

dm c

() () ()=  + 

=h

¨¨

{}

=   

xm x x xm x

Stbi ctiT s

() ,

() [] ( )GI

dpcch

(),tiT

I

Spreading

d,1

dpch

c,1

d,2

c,2

DPDCH

DPCCH

Scaling Scrambling

4190_book.fm Page 218 Tuesday, February 21, 2006 9:14 AM

Multiuser MIMO for UTRA FDD

219

where the subscript x



, c

}, refers to convolution with multipath channel

m,k

, and refers to chip-by-chip multiplication of user’s scrambling code

dpch

and channelization codes c

d,m

and c

c,m

of dedicated physical data channel

(DPDCH) and dedicated physical control channel (DPCCH), respectively.

Total transmit power is given by P

, and power ratio between DPDCH and

DPCCH is adjusted by scaling factors G

and G

, and user’s transmitted bits

in DPDCH and DPCCH are denoted by b

and b

. Users are not synchronized

in uplink, and the delay is denoted by I

. Finally T

, T

, and T

refer to the

lengths of the two channelization codes and the scrambling code.

Each transmit antenna uses different channelization codes c

d,m

and c

c,m

that the base stations can separate the signals from different antennas and

later combine them using maximal ratio combining. Thus, the scheme dou-

bles the usage of uplink channelization codes when compared to single

antenna transmission. The achievable bit rate with diversity MIMO is at

maximum 2 Mbps/M

. This reduction of maximum bit rate could be partly

avoided by puncturing the applied error correcting codes, but information

MIMO provides a more attractive choice since the reduced code rate quite

rapidly destroys system performance. However, for low and medium bit

rate services, this does not present a problem.

We note that in the case of ﬂat fading and perfect channel state information

in the receiver, the link performance of the proposed two-antenna diversity

MIMO algorithm is the same as with STTD, because the diversity order of

both schemes is two. However, the system utilizing orthogonal channeliza-

tion codes is more robust, because orthogonality of the received signals does

not depend on the channel estimation, as in the case of space–time coding.

8.4.1.2 Information MIMO

In the case of information MIMO of Figure 8.4, a composite data stream is

multiplexed into two or more independent substreams that are transmitted

from separate antennas by employing different scrambling codes. All streams

contain DPDCH and DPCCH so that in the base station they can be inter-

preted as signals from different independent users. The received wideband

signal in k

th antenna is given by

(8.2)

where again x



{d,c

}. We note that now information bits as well as scrambling

code chips vary between antennas. In contrast, the channelization code can

be the same in all antenna branches.

The present UTRA FDD speciﬁcation allows the mobile to use only a single

scrambling code as well as a single DPCCH and DPDCH. For the proposed





yt h t

StjS

dm c

() () ()=  + 

=h

¨¨

{}

=    

xm xm x x x

Stbi ctiT s

() ,

() [] ( )GI

dpcch,



tiT(),I

4190_book.fm Page 219 Tuesday, February 21, 2006 9:14 AM

220 MIMO System Technology for Wireless Communications

information MIMO, the speciﬁcation should be changed such that the use

of multiple scrambling codes is allowed in the mobile end. Then, it would

be possible to independently apply DPCCH and DPDCH to each scrambling

code. This does not represent a big change to radio interface, because it only

requires that a single user can set up M

different links. For the effective use

of several simultaneous links, some new code puncturing sets would also

be needed.

8.4.1.3 Advanced Receiver Techniques

As mentioned before, UTRA FDD uplink is interference limited. The inter-

ference-limited nature of CDMA systems results from the receiver design:

the reception is typically based on a matched ﬁlter that does not take into

account multiple access interference (MAI). However, MAI arises in multi-

path channels or with asynchronous communications when spreading codes

do not usually stay completely orthogonal anymore. This limitation can be

relaxed by employing advanced multiuser receiver algorithms, which aim

to reduce MAI. Information MIMO increases MAI in the base station, and

therefore advanced receivers become important to the system performance

when implementing information MIMO techniques.

In order to avoid the overwhelming complexity of multiuser detection,

various suboptimal multiuser receivers have been developed to cope with

MAI. The multiuser receivers can be categorized in several ways. One pos-

sibility is to classify the receivers to linear (decorrelating detector, minimum

mean square error (MMSE) detector) and non-linear receivers. Subtractive

interference cancellation (IC) is one example of the latter category, which

ﬁrst estimates the MAI component and then subtracts it from the received

signal to improve the reliability of the symbol decisions.

FIGURE 8.4

System model for information MIMO.

dpch,1

dpch,2

DPDCH

DPCCH

DPDCH

DPCCH

Spreading

Scaling Scrambling

4190_book.fm Page 220 Tuesday, February 21, 2006 9:14 AM

Multiuser MIMO for UTRA FDD

221

Cancellation of MAI can be carried out in parallel to all users, i.e., with

parallel interference cancellation (PIC) receivers, or in a serial fashion, i.e.,

using serial interference cancellation (SIC) receivers. In un-decoded (conven-

tional) PIC receivers, IC is based on tentative symbol decisions. The quality

of the tentative decisions is essential, because incorrect decisions increase

interference instead of decreasing it. In practice, the error rate of un-decoded

bits varies between 5 and 20% in UTRA FDD uplink. The error rate depends

mainly on the effective coding rate (i.e., rate of the channel encoder taking

into account possible repetition and puncturing), the radio environment, and

the quality of service requirements.

Unreliable tentative bit decisions compromise the performance of un-

decoded PIC. One possibility to improve the reliability of the decisions is to

ﬁrst decode the received symbols, re-encode the decoded symbols again,

and ﬁnally subtract the contribution of the re-encoded symbols from the

received signal. For more details we refer to [17], where this kind of decoded

PIC is applied to the convolutionally coded CDMA system. In [18], this idea

is extended to the Turbo-coded CDMA system.

8.4.2 Uplink Load Equations

Load equations provide a useful tool in both initial planning and perfor-

mance analysis of the network. Here we brieﬂy recall the load equations

[19,20] that will serve as a basis for further studies.

8.4.2.1 Basic Load Equations

In the following N

own

and N

other

refer to the number of own cell and other

cell users, respectively. The received wideband signal in baseband at time

instant t

can be expressed in the form

(8.3)

where n

) is complex zero-mean Gaussian, y

) is the interference from n

own cell user, and z

) is the interference from n

th other cell user. The system

load can be estimated from the sum of received wideband powers of different

users. Signals from different users are mutually independent and from Equa-

tion 8.3 the received wideband power is given by

where E refers to expectation, P

is the noise power, and

rt y t z t nt

() () () (),=++

¨¨

own other

total own other

rt I I I P

() ,

{}

==++

4190_book.fm Page 221 Tuesday, February 21, 2006 9:14 AM

222 MIMO System Technology for Wireless Communications

Received energy per data bit divided by the noise spectral density is an

important variable in the analysis of UTRA FDD uplink. For user n

it is given

(8.4)

where refers to the received power and is the ratio between bit rate

and chip rate of user n

. In UTRA FDD uplink, fast power control tries to

keep E

at a target level. Therefore, changes in the interference do not reﬂect

to E

, as long as power control is able to compensate the increased interfer-

ence through P

rx,n

. From Equation 8.4 we easily ﬁnd that

where is the load from the n

th user. For the received wideband power in

the own cell there holds

(8.5)

Furthermore, the total wideband power can be written in the form

(8.6)

where S

= I

other

own

is the intercell to intracell interference ratio seen by the

base station receiver. After combining Equation 8.5 and Equation 8.6 and

solving I

total

from the resulting formula, we obtain the following expression

for the noise rise

(8.7)

where the sum term deﬁnes the own cell load M

. In UTRA FDD uplink, load

control drives noise rise to a target level. The more loading is allowed in the

IytI z

nown other

own other

EE=

{}

¨¨

() , (()t

{}

= 



total rx

nrx,

nrx, total total

 =: 

IP I

own rx, total

own own

¨¨

M ·.

IIIP IP

NNtotal own other own

=++=+ +() ,1 S

µ SM== +

()



total

n=1

own

(),

4190_book.fm Page 222 Tuesday, February 21, 2006 9:14 AM

Multiuser MIMO for UTRA FDD 223

system, the larger is the required interference margin max{µ} and the smaller

is the coverage area. Interference margin deﬁnes the maximum allowed noise

rise, and typically values 1.0 to 3.0 dB are used for coverage-limited cases

with 20 to 50% load, and in capacity-limited cases, higher interference mar-

gins up to 6 dB can be used [19].

8.4.2.2 Load Equations for MIMO

In diversity MIMO with Rake receiver, the load from the nth user is of the

form

where the superscript D refers to the diversity MIMO. Diversity MIMO

reduces the required energy per data bit and there holds

(8.8)

where the received energy per data bit in single antenna system E

is divided

by the number of receiver antennas, because Rake collects energy from M

independent antenna branches. Factor depends on the channel diversity

and system-related issues such as the accuracy of channel estimation. Only

in simple channels, such as single path and ITU Pedestrian A channels, may

the effect of be noticeable, while in more realistic channels, such as

ITU Vehicular A and ITU Pedestrian B, is round 0 dB. The ITU channel

proﬁles [21,22] are listed in Table 8.1.

In information MIMO with Rake, the uplink load of a single user is of the

form

(8.9)

where is the ratio between bit rate and chip rate of the mth data stream

and is the received energy per data bit divided by the noise spectral

TABLE 8.1

ITU Multipath Channel Proﬁles

Model Delay Proﬁle [ns] Power Proﬁle [dB]

Pedestrian A 0 110 190 410 0,9.7, 19.2, 22.8

Pedestrian B 0 200 800 1200 2300 3700 0, 0.9, 4.9, 8.0, 7.8, 23.9

Vehicular A 0 310 710 1090 1730 2510 0, 1, 9, 10, 15, 20

EEM

MM n r

I /,

nm nm

nm,

4190_book.fm Page 223 Tuesday, February 21, 2006 9:14 AM