The UMTS Network and Radio Access Technology: Air Interface Techniques for Future Mobile Systems
Jonathan P. Castro
Copyright © 2001 John Wiley & Sons Ltd
Print ISBN 0-471-81375-3 Online ISBN 0-470-84172-9



S
YSTEM
A
NALYSIS
F
UNDAMENTALS

2.1 F
UNDAMENTALS OF
S
YSTEM
A
NALYSIS

Third generation systems focus on providing a universal platform to afford multifarious
communications options at all levels, i.e. the radio as well as the core network sides.
This implies the application of optimum techniques in multiple access and inter-
working protocols for the physical and upper layers, respectively. This chapter dis-
cusses the background of the multiple access or radio part of the UMTS specification.
Several sources [5–9] have already covered all types of fundamentals related to the air-
interface. Thus, we focus only on the communications environment to access the radio
link performance for coverage analysis and network dimensioning in forthcoming chap-
ters.
2.1.1 Multiple Access Options

to convey the information. These systems include primarily Spread Spectrum (SS) sys-
tems, e.g. Direct Sequence Spread Spectrum (DSSS) and Frequency Hopping Spread
Spectrum (FHSS). In DSSS, emphasized in this book, the transmission bandwidth ex-
ceeds the coherent bandwidth, i.e. the received signal after de-spreading resolves into
multiple time-varying delay signals that a RAKE receiver can exploit to provide an in-
herent time diversity receiver in a fading environment. In addition, DSSS has greater
resistance to interference effects when compared to FDMA and TDMA. The latter
greatly simplifies frequency band assignment and adjacent cell interference. In addition,
capacity improvements with DSSS or more commonly referred to as DS-CDMA
2
, re-
sulting from the voice activity factor, which we cannot apply effectively to FDMA or
TDMA. With DS-CDMA, e.g. adjacent micro-cells share the same frequencies, whereas
interference in FDMA and TDMA does not allow this. Other benefits and features can
be found in [10–12]. Here we focus on the WCDMA or FDD mode and TDMA/
CDMA or TDD mode of the UTRA solution.
2.1.1.3 The UTRA FDD Mode: WCDMA
Figure 2.1 illustrates some of the UTRA Frequency Division Duplexing (FDD) charac-
teristics. This mode uses Wide-band Direct-Sequence Code Division Multiple Access
(DS-CDMA), denoted WCDMA. To support bit rates up to 2 Mbps, it utilizes a variable
spreading factor and multi-code links. It supports highly variable user data rates through
the allocation of 10 ms frames, during which the user data rate remains constant, al-
though the latter may change from frame to frame depending on the network control. It
realizes a chip rate of 3.84 Mcps within 5 MHz carrier bandwidth, although the actual
carrier spacing can be selected on a 200 kHz grid between approximately 4.4 and
5 MHz, depending on the interference situation between the carriers.
Ã
QrÃ
ÃArrpÃ
9vssrrÃrhqvtÃshpÃ

PVIUDPH
Uvr
WhvhiyrÃTrhqvt

Figure 2.2 UTRA TDD mode characteristics.
In this mode, the MSs can only access a Frequency Division Multiplexing (FDM) chan-
nel at specific times and only for a specific period of time. Thus, if a mobile gets one or
more Time Slots (TS) allocated, it can periodically access this set of TSs throughout the
duration of the frame. Spreading codes described in Chapter 4 separate user signals
within one or more slots. Hence, in the TDD mode we define a physical channel by a
code, one TS, and one frequency, where each TS can be assigned to either the uplink or
the downlink depending on the demand. Users may obtain flexible transmission rates by
occupying several TSs of a frame as illustrated in Figure 2.2, without additional proc-
essing resources from the transceiver hardware. On the other hand, when more than one
frequency channel gets occupied, utilization of transceiver resources will increase if the
wide-band transmission cannot prevent it. We achieve variable data rates through either
multi-code transmission with fixed spreading or through single code with variable
spreading. In the 1st case, a single user or users may get multiple spreading codes
within the same TS; while in the 2nd case, the physical channel spreading factor may
vary according to the data rate.
16 The UMTS Network and Radio Access Technology

2.1.2 Signal Processing Aspects
In the following, we review Signal Processing characteristics for the WCDMA as well
as TD/CDMA as a base to describe key functions of the UTRA FDD and TDD modes.
These include spreading aspects and modulation and coding.
2.1.2.1 The Spread Spectrum Concept
Digital designs of communications systems aim to maximise capacity utilization. We
can for example increase channel capacity by increasing channel bandwidth, and/or
transmitted power. In this context, CDMA operates at much lower S/N ratios as a result

. The latter quantifies the degree of interference rejection and can be de-
fined as the ratio of RF bandwidth to the information rate:
S
%
*
5
=
 
From Ref. [23] in a spread-spectrum system, thermal noise and interference determine
the noise level. Hence, for a given user, the interference is processed as noise. Then, the
input and output S/N ratios can relate as:
S
RL
66
*
1 1
ËÛ ËÛ
=
ÌÜ ÌÜ
ÍÝ ÍÝ
 
Relating the S/N ratio to the E
b
/N
o
ratio
4
, where E
b
is the energy per bit and N

System Analysis Fundamentals 17
E
S
LR
R
(
66
*
1 11
ËÛËÛ
= =
ÌÜÌÜ
ÍÝÍÝ
 
2.1.2.2 Modulation and Spreading Principles
In wide-band spread-spectrum systems like the FDD mode, the entire bandwidth of the
system remains available to each user. To such systems, the following principles apply:
first, the spreading signal has a bandwidth much larger than the minimum bandwidth
required to transfer desired information or base-band data. Second, data spreading oc-
curs by means of a code spreading signal, where the code signal is independent of the
data and is of a much higher rate than the data signal. Lastly, at the receiver, de-
spreading takes place by the cross-correlation of the received spread signal with a syn-
chronized replica of the same signal used to spread the data [23].
2.1.2.2.1 Modulation
If we view Quadrature Phase-Shift Keying (QPSK) as two independent Binary Phase-
Shift Keying (BPSK) modulations, then we can assume the net data rate doubles. We
now provide the background for QPSK to serve as background to the applications in
UTRA presented in Chapter 4.
For all practical purposes we start with M-PSK, where M = 2
b

c
is the unmodulated carrier frequency, A is the carrier amplitude, and T
s
is the
symbol interval. When T
b
is the input binary bit interval, T
s
may be expressed as
VE
ORJ77 0=
 
The power spectral density of an unfiltered M-PSK signal occupies a bandwidth which
is a function of the symbol rate r
s
= (1/T
s
). Thus, for a given transmitter symbol, the
power spectrum for any M-PSK signal remains the same regardless of the number M of
symbol levels used. This implies that BPSK, QPSK and 8-PSK signals each have the
same spectral shape if T
s
remains the same in each case.
Spectral Efficiency
For a M-ary PSK scheme each transmitted symbol represents log
2
M bits. Hence, at a
fixed input bit rate, as the value of M increases, the transmitter symbol rate decreases;
which means that there is in increase in spectral efficiency for larger M.
Thus, if for any digital modulation the spectral efficiency

would be B = r
s
. Typical values of
a
B
used in practice are 0.3–0.5 [3].
Thus, for M-PSK transmission using the Nyquist filtering with roll-off
a
B
the required
bandwidth will be given by
()
V%
%U=+a



Then the maximum bit rate in terms of the transmission bandwidth B, and the roll-off
factor
a
B
can be defined as

E
%
ORJ

% 0
U =
+a

here we review only the QPSK scheme. In QPSK each transmitted symbol (Figure 2.3)
represents two input bits as follows:
Input bits Transmitted symbols
00 A cos(w
c
t + 45°)
01 A cos(w
c
t + 135°)
11 A cos(w
c
t + 235°)
10 A cos(w
c
t + 315°)
System Analysis Fundamentals 19
The conversion from binary symbol to phase angles is done using Gray coding. This
coding permits only one binary number to change in the assignment of binary symbols
to adjacent phase angles, thereby minimizing the demodulation errors, which in a digital
receiver result from incorrectly selecting a symbol adjacent to a correct one.
Figure 2.3 illustrates a block diagram frequently used for any form of M-PSK modula-
tion. For QPSK, the multiplexer basically converts the binary input stream into two par-
allel, half rate signal v
I
(t) and v
Q
(t) (i.e. the in-phase and quadrature signals). These sig-
nals taking values +A/
¥
or –A/





Q

I

01

00

11

10

QPSK signal vector diagram

X

X

90

deg.

+

+


r

b

= bit/s

Output

QPSK

Signal

(a)


Figure 2.3 QPSK configuration, after [3].
Assuming a coherent demodulator, the latter includes a quadrature detector consisting
of two balanced multipliers with carrier inputs in phase quadrature, followed by root-
Nyquist filter in the output I and Q arms. Then, the resultant I and Q signals are sam-
pled at the centre of each symbol to produce the demodulator output I and Q signals,
which in turn are delivered to the decoder [3].
Generally, an M-PSK modulator produces symbols with one of M phase values spaced
 0
apart. Then each signal is demodulated correctly at the receiver when the phase is
within
0
radians of the correct phase at the demodulator sampling instant. If noise is
present, evaluation of the probability of error requires a calculation of the probability
20 The UMTS Network and Radio Access Technology


Q
= V volts (i.e. noise-free case). Thus, if we consider that the
noise phasors (n
1
and n
2
) are pointing in directions that are most likely to cause errors,
then a symbol error will occur if either n
1
or n
2
exceeds V.
Q axis
I Axis
noise
noise
n
2
n
1
Transmitted
signal
Received
signal

Figure 2.4 Transmitted and received signal vectors [3].
Now, if for simplicity we also assume that a QPSK signal is transmitted without
Nyquist filtering and demodulated with hard-decisions, the probability of a correctly
demodulate symbol value is equal to the product of the probabilities that each demodu-
lator low-pass filter output lies in the correct quadrant. Then the probability that the

System Analysis Fundamentals 21
where E
s
= A
2
T
s
/2 is the energy per symbol, N
o
/2 is the two-sided noise power noise
spectral density (in V
2
/Hz) at the demodulator input, and the function Q(x) is the com-
plementary integral Gaussian function. The error function erf(x) given by
()
()



HUI H[S G
[
[ \\=-
p
×
 
and complementary error function erfc(x) expressed as
( ) ( )
HUIF  HUI[ [=-



 
which at P
e1

EHFRPHV

VH

3 3 
 
Substituting P
e1
from equation (2.17) into equation (2.22), the QPSK symbol error
probability can be given by
V
V
R

(
34
1
ËÛ
 
ÌÜ
ÌÜ
ÍÝ
 
Now, for QPSK E
s
= 2E

are the same for both cases.
2.1.2.3 CDMA System Performance
As noted earlier, CDMA systems tolerate more interference than typical TDMA or
FDMA systems. This implies that each additional active radio user coming into the
22 The UMTS Network and Radio Access Technology

network increases the overall level of interference to the cell site receivers receiving
CDMA signals from mobile station transmitters. This depends on its received power
level at the cell site, its timing synchronization relative to other signals at the cell site,
and its specific cross-correlation with other CDMA signals. Consequently, the number
of CDMA channels in the network will depend on the level of total interference that the
system can tolerate. As a result, the FDD mode behaves as an interference limited sys-
tem, where technical design will play a key role in the overall quality and capacity per-
formance. Thus, despite advanced techniques such as multi-user detection and adaptive
antennas, a robust system will still need a good bit error probability with a higher level
of interference.
When we consider that at the cell site all users receive the same signal level assuming
Gaussian noise as interference, the modulation method has a relationship that defines
the bit error rate as a function of the E
b
/N
o
ratio. Therefore, if we know the performance
of the signal processing methods and tolerance of the digitized information to errors, we
can define the minimum E
b
/N
o
ratio for a balanced system operation. Then, if we main-
tain operation at this minimum E

ORJ 
&6
% 1
ËÛ
<
ÌÜ
ÍÝ
DQG
E
HR

ORJ 
&(&
% 1%
ËÛ
ËÛ
<
ÌÜ
ÌÜ
ÍÝ
ÍÝ
 
then
E
H
ORJ   
(
%
==-
G% 

IDFWRU7KLVHYHQWZLOO FDXVHUHGXFWLRQRIWKHQXPEHU
of users in a cell, because the interference from users in other cells has to be added to
WKHLQWHUIHUHQFHJHQHUDWHGE\WKHRWKHUPRELOHVLQWKHXVHU¶VFHOO PD\UDQJHIURP
to 0.55. In addition to the interference factor, we also introduce the imperfect power
con
WURORUSRZHU FRQWURODFFXUDF\IDFWRU ZKLFK UDQJHVIURP WR,QWHUIHUHQFH
FDQEHUHGXFHGE\WKHYRLFHDFWLYLW\IDFWRU
UDQJLQJIURPWR,IZHXVHGLUH
c-
tional antennas at the base station, the sectorized cell will have a sectors, the antennas
used at the cell each will radiate into a sector of 360/a degrees, resulting in an interfer-
ence im
SURYHPHQWIDFWRU $YHUDJHYDOXHVIRU   DQG VHFWRUFHOODUH

0.6 and 2.55, respectively [23]. Then incorporating all the preceding factors the user
capacity equation becomes:
S
ER


*
0
(1
 al
+bn
 
In the forthcoming section we also review pseudorandom sequences as part of the signal
processing aspects relevant for the operation of the UTRA modes.
2.1.2.4 Pseudorandom Sequences
Pseudorandom noise (PN), i.e. deterministic periodic sequences in WCDMA perform