Norwegian University of Science and Technology

Department of Electronics and Telecommunications

Mai. 25, 2007

TTT4160 Mobile Communications

List of Formulas

1
1.1

Path Loss and Shadowing (Chapter 2)

Doppler frequency

The Doppler frequency is obtained by:

fD =

v
cos

(1)

where is the arrival angle of the received signal relative to the direction of motion. v is the
receiver velocity toward the transmitter in the direction of motion, and = c/fc .

1.2

Free Space Path loss:

The received signal:

Gl ej2d/
u(t)ej2fc t }
4d
The ratio of received to transmitted power is:
r(t) = Re{

Pr
=
Pt

Gl
4d

(2)

2
(3)

The received power in dBm as:

Pr dBm = Pt dBm + 10 log10 (Gl ) + 20 log10 () 20 log10 (4) 20 log10 (d)

(4)

Free-space path loss is defined as the path loss of the free-space model:
PL dB = 10 log10

1.3

Gl 2
Pt
= 10 log10
Pr
(4d)2

(5)

Two-Ray Model

The received signal of the two-ray model is:

"
#

Gl u(t)ej2l/ R Gr u(t )ej2(x+x )/ j2fc t

r2ray (t) = Re{
+
e
}
4
l
x + x0

(6)

where
= (x + x0 l)/c is the time delay of the ground reflection relative to the LOS ray,

Gl = Ga Gb is the product of the transmit and receive antenna field radiation patterns in
1

the LOS direction. R is the ground reflection coefficient, and Gr = Gc Gd is the product of
the transmit and receive antenna field radiation patterns corresponding to the rays of length
x and x0 , respectively.
The received power:

Pr = Pt

2 
 Gl R Gr ej 2


 l +

x + x0

(7)

where = 2(x + x0 l)/ is the phase difference between the two received signal components.
From geometry:
p
p
(8)
x + x0 l = (ht + hr )2 + d2 (ht hr )2 + d2
where ht is the height of transmitting antenna, and hr is the height of the receiving antenna.
The received power can also be approximated by (for large d):
 2 


2
Gl
4ht hr 2
G l ht hr
Pr
Pt =
Pt
4d
d
d2

(9)

Pr dBm = Pt dBm + 10 log10 (Gl ) + 20 log10 (ht hr ) 40 log10 (d)

(10)

or, in dB,

1.4

Simplified Path-Loss Model

Simplified path-loss model:


Pr = Pt K

d0
d


(11)

The dB attenuation is thus.


Pr dBm = Pt dBm + K dB 10 log10

d
d0


(12)

K is a unitless constant that depends on the antenna characteristics, and for omnidirectional
antennas:

(13)
K dB = 20 log10
4d0

1.5

Shadow Fading

The ratio of transmit-to-receive power = Pt /Pr is assumed to be random with a log-normal

distribution given by.
"
#
(10 log10 dB )2

exp
, > 0,
(14)
p() =
22 dB
2dB
"

2
dB
+ dB
= E[] = exp

2 2

#
(15)

10 log10 = dB +

2 dB
2

(16)

The distribution of the dB value of is Gaussian with mean dB and standard deviation
dB :
#
"
(dB dB )2
1
(17)
exp
p(dB ) =
22 dB
2dB

1.6

Combined Path Loss and Shadowing

The combination of the simplified path-loss model and the log-normal shadowing random
process. For such combined model, the ratio of received to transmitted power in dB is given
by:
Pr
d
dB = 10 log10 K 10 log10
dB
(18)
Pt
d0
where dB is a Gauss-distributed random variable with mean zero and variance 2 dB .

1.7

Outage Probability under Path Loss and Shadowing

For the combined path-loss and shadowing model, the outage probability becomes:
p(Pr (d) Pmin ) = 1 Q(

Pmin (Pt + 10 log10 K 10 log10 (d/d0 ))

)
dB

(19)

where Q denotes the Qfunction (See table on p.18) and Pmin is the minimum power requirement.

1.8

Cell Coverage Area

Given the log-normal distribution for the shadowing, the cell coverage area can be computed
by:
2
R2

r
rQ(a + b ln )dr,
R
0


2 2ab
2 ab
= Q(a) + exp
Q(
)
2
b
b

C =

where
a=

Pmin Pr (R)
,
dB

b=

10 log10 (e)
,
dB

(20)

(21)

and PR = Pt +10 log10 K10 log10 (R/d0 ) is the received power at the cell boundary (distance
R from the base station) due to path loss alone.

Statistical Multipath Channel Models (Chapter 3)

Let the transmitted signal be:

s(t) = Re{u(t)ej2fc t } = Re{u(t)} cos(2fc t) Im{u(t)} sin(2fc t)

(22)

where u(t) is the equivalent lowpass signal for s(t)with bandwidth Bu and where fc is the
carrier frequency.
The based-band representation of the received multipath signal is:
N (t)

r(t) = Re

X

n (t)u(t n (t))ej(2fc (tn (t))+Dn )

(23)

n=0

where n = 0 corresponds to the LOS path, N (t) are the number of resolvable multipath
components, n (t) is the delay in each path, n (t) is the amplitude and Dn (t) denotes the
Doppler shift.
We can simplify r(t) by letting
n (t) = 2fc n (t) Dn
Then the received signal in (23) can be rewritten as:

N (t)
X


r(t) = Re
n (t)ejn (t) u(t n (t)) ej2fc t

(24)

(25)

n=0

if assume u(t i ) u(t), we could get:

X



r(t) = Re u(t)ej2fc t
n (t)ejn (t)

(26)

In order to characterize the random scale factor caused by the multipath, we choose s(t) to
be an unmodulated carrier with random phase offset 0 :
s(t) = Re{ej2fc t+0 } = cos(2fc t + 0 )

(27)

With this assumption the received signal becomes:

N (t)

r(t) = Re

 X



n (t)ejn (t) ej2fc t = rI (t) cos 2fc t rQ sin 2fc t,

(28)

n=0

where the in-phase and quadrature components are given by

N (t)

rI (t) =

n (t) cos n (t),

(29)

n (t) sin n (t)

(30)

n=1
N (t)

rQ (t) =

X
n=1

2.1

Rayleigh Fading

For any two Gaussian randomvariables X and Y , both with mean zero and equal variance
2 , it can be shown that Z = X 2 + Y 2 is Rayleigh distributed and that Z 2 is exponentially
distributed.
If we assume a variance of 2 for both in-phase and quadrature components, then the
signal envelope
q
2 (t)
z(t) = |r(t)| = rI2 (t) + rQ
(31)
is Rayleigh distributed with distribution
 2


z
z
2z
z2
pZ (z) = exp = 2 exp 2

2
Pr
Pr

(32)

Thus the power distribution can be obtained by making the change of variables z 2 (t) = |r(t)|2
in (32) to obtain
1
1
2

PZ 2 (x) = ex/Pr = 2 ex/2 , x 0

(33)
2
Pr
If the sinal has a LOS component, the signal envelop can be shown to have a Rician distribution given by:
 (z 2 + s2 )  zs
z
pZ (z) = 2 exp
I0 ( 2 ), z 0
(34)

2 2

The average received power in the Rician fading is given by:

Z
Pr =
z 2 pZ (z)dz = s2 + 2 2
(35)
0

The Rician distribution if often described in terms of a fading parameter K, defined by:
K=

s2
2 2

(36)

Mading the substitutions s2 = KPr /(K + 1) and 2 2 = Pr /(K + 1), we can write the Rician
distribution in terms of K and Pr as:
s

2

2z(K + 1)
(K + 1)z 
K(K + 1)
I0 2z
PZ (z) =
exp K
, z0
(37)

Pr
Pr
Pr
The Nakagami fading distribution is given by:
pZ (z) =

 mz 2 
2mm z 2m1
exp
,
(m)Prm
Pr

m .5,

(38)

where Pr is the average received power and () is the Gamma function. The power distribution for Nakagami fading, obtained by a change of variables, is given by.
mx
m xm1
exp[ ].
pZ 2 (x) = ( )m
(m)
Pr
Pr

(39)

2.2

Level Crossing Rate and Average Fade Duration

For Rayleigh fading, the level crossing rate is:

2
LZ = 2fD e
p
where = Z/ Pr
The average fade duration is:
tZ =

LZ T
p(z(t) < Z)
1 X
ti
T LZ
LZ

(40)

(41)

i=1

Using the Rayleigh distribution for p(z(t) < Z) the yields:

2

e 1

tZ =
fD 2

(42)

p
with = Z/ Pr

2.3

Wideband Fading Models

The autocorrelation function for a WSS random channel is:

Ac (1 , 2 ; t) = E[c (1 ; t)c(2 ; t + t)].

(43)

The scattering function is defined as the Fourier transform of Ac ( ; t) with respect to the
t parameter:
Z
Sc (, ) =
Ac (, t)ej2t dt
(44)

2.3.1

Power Delay Profile

The power delay profile Ac ( ) , Ac (, 0).

The average and rms delay spread is:
Tm
and

sR
Tm =

2.3.2

R
Ac ( )d
= R0
0 Ac ( )d

2
0 (R Tm ) Ac ( )d

0 Ac ( )d

(45)

(46)

Coherence Bandwidth

Take the Fourier transform of the channel implies response:

Z
C(f ; t) =
c( ; t)ej2f d

(47)

The autocorrelation is thus given by:

AC (f1 , f2 ; t) = E[C (f1 ; t)C(f2 ; t + t)]
6

(48)

and
AC (f1 , f2 ; t) = AC (f ; t)
where f = f2 f1 . If we define AC (f ) , AC (f ; 0), then
Z
Ac ( )ej2f d
AC (f ) =

(49)

(50)

The coherence bandwidth is Bc 1/T , where T is typically taken to be the rms delay
spread Tm of Ac ( ).
2.3.3

Doppler Power Spectrum and Channel Coherence Time

The Doppler effect can be characterized by taking the Fourier transform of AC (f ; t)

relative to t:
Z
SC (f ; ) =
AC (f ; t)ej2t dt
(51)

In order to characterize Doppler at a single frequency, we set f to zero and then define
SC () , SC (0; ). It is easily seen that
Z
SC () =
AC (t)ej2t dt
(52)

where AC (t) , AC (f = 0; t).

Capacity of Wireless Channels (Chapter 4)

The channel capacity formula in AWGN:

C = B log2 (1 + )

3.1
3.1.1

(53)

Capacity of Flat Fading Channels

Channel Side Information at Receiver - Shannon (Ergodic) Capacity
Z
C=
B log2 (1 + )p()d
(54)
0

3.1.2

Channel Side Information at Transmitter and Receiver - Shannon Capacity

The capacity of the fading channel with transmitter and receiver side information is.
Z
Z
C=
C p()d =
B log2 (1 + )p()d
0

(55)

with average transmit power constraint P :

Z
P ()p()d P
0

(56)

The optimal power adaptation policy that maximize the capacity for the fading channel
defined as:
Z
P ()
B log2 (1 +
C=
)p()d
(57)
R max
P
P (): P ()p()d=P 0
is given by:
P ()
=
P
3.1.3

(
1/0 1/
0

0 ,
< 0

(58)

Channel Side Information at Transmitter and Receiver - Zero-outage Capacity and Channel Inversion

Fading channel capacity with channel inversion is just the capacity of an AWGN channel with
SNR :


1
C = B log2 [1 + ] = B log2 1 +
(59)
E[1/]
3.1.4

Channel Side Information at Transmitter and Receiver - Outage Capacity

and Truncated Channel Inversion

The power adaptation policy for truncated channel inversion is given by:
(
/ 0
P ()
=

P
0
< 0

(60)

Pout = p( < 0 ) , = 1/E0 [1/], where

Z

E0 ,
0

1
p()d

(61)

The outage capacity:


C(Pout ) = B log2

1
1+
E0 [1/]


p( 0 )

(62)

p( 0 )

(63)

The maximum outage capacity is:


C = max B log2
0

3.2

1
1+
E0 [1/]

Capacity of Frequency-Selective Fading Channels - Time-Invariant Channels

The capacity for parallel set of channels is:

X

C=

max Pj :

B log2 (1 +

j Pj P

|Hj |2 Pj
)
N0 B

(64)

The optimal power allocation is:

Pj
=
P

(
1/0 1/j
0
8

j 0
j < 0

(65)

where j = |Hj |2 P/N0 B. The cutoff value must satisfy:

X 1
1
( )=1
0 j

(66)

Digital Modulation and Detection (Chapter 5 )

Denote si the signal constellation point corresponding to the signal si (t), and r as the received vector associated with r(t). In the receiver, we always want to maximize the posterior
probability: p(si |r), by Bayes rule:
p(si |r) =

p(r|si )p(si )
p(r)

(67)

Since the denominator doesnt matter, maximize p(si |r) equals to find si :
arg max
si

p(r|si )p(si )
= arg max p(r|si )p(si ),
si
p(r)

i = 1, . . . , M,

(68)

Further assuming equally likely messages (p(si ) = 1/M ), thus:

arg max p(r|si )p(si ) = arg max p(r|si ),
si

si

i = 1, . . . , M.

(69)

If define the likelihood function:

L(si ) = p(r|si )

(70)

N


1
1 X
2
p(r|si sent) =
p(rj |mi ) =
(r

s
)
exp

j
ij
N0
(N0 )N/2
j=1
j=1

(71)

and recall that:

N
Y

maximize the likelihood function equals to maximize the log likelihood function, defined as
l(si ) = ln[(N0 )N/2 L(si )] , this yields:
N
1 X
1
l(si ) =
(rj s2ij ) = kr si k2
N0
N0

(72)

j=1

This equals to find the minimum Euclidean distance, i.e., the signal constellation closest to r
is determined as the constellation point si satisfying
arg min
si

4.1

N
X

(rj sij )2 = arg min kr si k2

si

j=1

(73)

Amplitude and Phase Modulation - MPAM

The transmitted signal over one symbol time is given by:

si (t) = Re{Ai g(t)ej2fc t } = Ai g(t) cos(2fc t),
9

0 t Ts  1/fc

(74)

4.2

Amplitude and Phase Modulation - MPSK

The transmitted signal over one symbol time Ts is given by.

si (t) = Re{Ag(t)ej2(i1)/M ej2fc t }


2(i 1)
= Ag(t) cos 2fc t +
M




2(i 1)
2(i 1)
cos 2fc t Ag(t) sin
sin 2fc t
= Ag(t) cos
M
M

4.3

(75)

Amplitude and Phase Modulation - MQAM

The transmitted signal is given by:

si (t) = Re{Ai eji g(t)ej2fc t }
= Ai cos(i )g(t) cos(2fc t) Ai sin(i )g(t) sin(2fc t),

4.4

0 t Ts

(76)

Frequency Modulation - MFSK

In MFSK the modulated signal is given by:

si (t) = A cos[2fc t + 2i fc t + i , ],

4.5

0 t < Ts ,

(77)

Frequency Modulation - CPFSK

The modulated signal is given by:


Z
si (t) = A cos 2fc t + 2


u( )d = A cos[2fc t + (t)]

(78)

5
5.1

Performance of Digital Modulation over Wireless Channels

(Chapter 6 )
AWGN Channels

Approximate symbol and bit error probabilities for coherent modulations

Modulation
Ps (s )
Pb (b )

BFSK
Pb = Q( b )

BPSK
Pb = Q( 2b )

QPSK, 4-QAM
Ps 2Q( q
s )
Ps Q( q
2b )
2(M 1)
2(M 1)
6
s
b log2 M
MPAM
Ps = M Q( M 2 1 )
Pb M log M Q( 6M
2 1 )
2
p

MPSK
Ps 2Q( 2s sin( M )) Pb log M Q( 2b log2 M sin( M
))
2
q
q
3

log
M
3
s
b
2
Rectangular MQAM
Ps = 4Q( M 1 )
Pb log4 M Q(
)
2
q M 1
q
3

log
M
b
s
2
Nonrectangular MQAM
Ps 4Q( M31
)
Pb log4 M Q(
M 1 )
2

10

5.2

Error Probability Approximation for Coherent Modulations

Many of the formulas for Ps for coherent modulation are in the following form:
p
Ps (s ) M Q( M s ).

(79)

and Pb can be expressed as:

q
Pb (b ) =
M Q( M b )

(80)

where
M = M / log2 M and M = (log2 M )M

5.3

Alternate Q-Function Representation

The Q-function is defined as.

Z

Q(z) = p(X z) =
z

1
2
ex /2 dx
2

An alternate representation of the Q-function is given by:



Z
1 /2
z 2
d,
Q(z) =
exp
0
2 sin2

5.4

z > 0.

(81)

(82)

Fading - Outage Probability

The outage probability relative to 0 is defined as.

Z
Pout = p(s < 0 ) =

ps ()d

(83)

5.5

Fading - Average Probability of Error

when Ts Tc , the average error probability is:

Z
Ps =
Ps ()ps ()d

(84)

If the fading is Rayleigh distribution,

ps () =

1 /s
e
.
s

In particular, for BPSK in Rayleigh fading:

r


1
b
1

Pb =
1

2
1 + b
4b

(85)

(86)

For binary FSK in Rayleigh fading:

r


1
b
1

Pb =
1

2
2 + b
4b
For DPSK:
Pb =

1
1

2(1 + b )
2b
11

(87)

(88)

5.6

Fading - Moment Generating Function Approach to Average Error

Probability

The MGF for a nonnegative random variable with distribution p (), 0, is defined as
Z
p ()es d
(89)
M (s) =
0

n Rayleigh:

Ms (s) = (1 ss )1

(90)



Kss
1+K
exp
1 + K ss
1 + K ss

(91)

n Rician with factor K:

Ms (s) =
n Nakagami-m:
Ms (s) = (1

ss m
)
m

(92)

The moments E[ n ] of can be obtained from M (s) as:

E[ n ] =

n
[Ms (s)]|s=0 .
sn

(93)

If we change the notation in (79), setting = M and g = .5M and also use the alternate
Q-function representation in (82), we get that:

Ps =

/2


g
d,
exp
sin2


(94)

Thus the average error probability in fading for modulations with Ps = Q( 2gs ) in AWGN
is given by:


Z
g
/2
Ps =
Ms
d,
(95)
0
sin2
n Rayleigh:

Ms

g
sin2


=

1+

g s
sin2

1
(96)

n Rician with factor K.





g
(1 + K) sin2
Kg s
Ms 2
exp
=
sin
(1 + K) sin2 + g s
(1 + K) sin2 + g s

(97)

n Nakagami-m:

Ms

g
2
sin


=

g
2
sin

12


=

g s
1+
m sin2

m
(98)

5.7

Doppler Spread

Correlation coefficients for different Doppler power spectral models

Type
Doppler power spectrum SC (f ) C = AC (T )/AC (0)
Rectangular
P0 /2BD , |f | < BD
sinc(2BD T )

2 /B 2
2
f
D
Gaussian
(Pq
e(BD T )
0 / BD )e
2 f 2 , |f | < B
Uniform scattering
P0 / BD
J0 (2BD T )
D
2
2
1st-order Butterworth
P0 BD /(f + BD )
e2BD T
where BD is the Doppler spread of the channel.
By utilizing alternate forms of the Macrum Q-function, the resulting average bit error
probability for DPSK is:




K b
1 1 + K + b (1 C )

exp
Pb =
(99)
2
1 + K + b
1 + K + b
where C is the channel correlatin coefficient after a bit time Tb , K is the fading parameter
of the Rician distribution, and b is the average SNR per bit. For Rayleigh fading (K = 0)
this simplifies to


1 1 + b (1 C )

Pb =
.
(100)
2
1 + b
Letting b , yields the irreducible error floor:
(1 C )eK
Pfloor =
2

DPSK.

(101)

Diversity (Chapter 7)

The array gain Ag is defined as the increase in the average combined SNR over the average
branch SNR :

Ag =
(102)

6.1

Receiver Diversity - Selection Combining

For M -branch diversity, the cumulative distribution function (cdf) of is given by:
P () = p( < ) = p(max[1 , 2 , . . . , M ] < ) =

M
Y

p(i < )

(103)

i=1

The distribution for in Rayleigh fading:

p () =

M
[1 e/ ]M 1 e/

(104)

The average SNR of the combiner output in independent and identically distributed Rayleigh
fading is:
M
X
1
=
.
(105)
i
i=1

13

6.2

Threshold Combining

Denote the SNR on the ith branch by i and the SNR of the combiner output by , the
threshold level T . Then for two-branch diversity with i.i.d. branch statistics, the cdf of the
combiner is:
(
< T
P1 (T )P2 ()
(106)
P () =
p(T 2 ) + P1 (T )P2 () T
For Rayleigh fading in each branch with i = (i = 1, 2), this yields:
(
1 eT / e/ + e(T +)/ < T
P () =
1 2e/ + e(T +)/
T

(107)

The outage probability Pout associated with a given 0 is obtained by evaluating P () at

= 0 :
(
1 eT / e0 / + e(T +0 )/ 0 < T
Pout (0 ) = P (0 ) =
(108)
1 2e0 / + e(T +0 )/
0 T
Hence, the distribution is:
(
(1 eT / )(1/
)e/
p () =
(2 eT / )(1/
)e/

6.3

< T
T

(109)

Maximal-Ratio Combining

The output SNR of the combiner is:

P
2
1 ( M
i=1 ai ri )
=
P
M
2
N0
i=1 ai

(110)

The optimal weight factor is: a2i = ri2 /N0 .

The distribution of is:
M 1 e/
,
p () =
M (M 1)!

(111)

The corresponding outage probability for a given threshold 0 is given by:

Z

0 /

p ()d = 1 e

Pout = p( < 0 ) =
0

7
7.1

M
X
(0 /
)k1
k=1

(k 1)!

(112)

Adaptive Modulation and Coding (Chapter 9)

Variable-Power Techniques

The power adaptation for channel inversion is given by:

P ()

P
14

(113)

where the average power constraint implies

Z
Z
P ()

p()d = 1
p()d =

P
For truncated channel inversion, the power adaptation is
(
/, 0
P ()
=
P
0,
< 0

(114)

(115)

In truncated channel inversion, since the channel is only used when 0 , given an average
power P we have = 1/E0 [1/], where
  Z
1
1
E0
,
p()d.
(116)

7.2

Variable-Rate Variable-Power MQAM - Adaptive Rate and Power

Scheme

A bound oon bit error probability for MQAM in an AWGN channel is:
Pb 2e1.5/(M 1)

(117)

A tighter bound, good to within 1 dB for M 4 and 0 30 dB, is:

Pb .2e1.5/(M 1)
If using the tighter bound given by (118), Pb under a certain power adaptation is:


1.5 P ()
Pb () .2 exp
M 1 P

(118)

(119)

Adjusting M and P () to maintain the target Pb yields:

M () = 1 + K

P ()
P

(120)

where

1.5
< 1.
ln 5Pb
The power adaptation policy that maximizes the average spectral efficiency
Z
KP ()
)p()d
E[log2 M ()] =
log2 (1 +
P
0
K=

(121)

(122)

is given by:
KP ()
=

(
1/K 1/
0

K
K

where K is a cutoff fade depth and the cutoff must satisfy the power constraint
Z
1
1
(
)p()d = K.

K
K
15

(123)

(124)

7.3

Variable-Rate Variable-Power MQAM - Channel Inversion with Fixed

Rate

The spectral efficiency for full channel inversion is:



R
1.5
= log2 1 +
B
ln(5Pb )E[1/]
The maximum spectral efficiency for truncated channel inversion is:


R
1.5
p( > 0 )
= max log2 1 +
0
B
ln(5Pb )E0 [1/]

7.4

(125)

(126)

Variable-Rate Variable-Power MQAM - Discrete-Rate Adaptation

If restrict the adaptive MQAM to a limited set of constellations with squared constellation
size: M0 = 0, M1 = 2, and Mj = 22(j1) , j = 2, . . . , N 1. The power adaptation policy is:
(
M
(Mj 1)(1/K) Mj < /K
Pj ()
j+1
=
(127)
P
0
Mj = 0
= as in (123). The spectral efficiency for this discrete-rate policy is:
where K
K



N
1
X
R

=
log2 (Mj )p Mj < Mj+1
B
K

(128)

j=1

7.5

Variable-Rate Variable-Power MQAM - Discrete-Rate Adaptation with

Channel Inversion Policy

If the constellation must be chosen from a fixed set of possible constellation M = M0 = 0, . . . , MN 1 .

For total channel inversion, the spectral efficiency is.


1.5
R
= log2
1+
(129)
B
ln(5Pb )E[1/] M
For truncated channel inversion, the spectral efficiency is obtained by optimizing relative
to the cutoff level 0 :


R
1.5
= max log2 1 +
p( > 0 )
(130)
0
B
ln(5Pb )E0 [1/]

Cellular System and Infrastructure-Based Wireless Networks

(Chapter 15)

For diamond-shaped cells, the reuse distance D for a given number of intermediate cells NI
between co-channel cells, assuming the cell radius is R:
D = R + 2RNI + R = 2R(NI + 1)

16

(131)

For diamond-shaped cells, assuming a cluster with K cell on each side, the reuse factor
N = K 2 and we also have N = .25(D/R)2 .
If NT denotes the total number of channels and Nc denotes the number of channels per
cell, then N = NT /Nc .
If we assume that all transmitters send at the same power Pt and assume the user is at
distance d < R and further assume there are M interferers at distance di , i = 1, . . . , M, from
the intended receiver, the resulting SIR is then.
dI
SIR = PM O
i=1 di

(132)

where I denotes the in-cell path-loss exponent and O denotes the out-of-cell(intercell) pathloss exponent.
The SIR can also be written as a function of N as:
SIR = a1 (a2 N )/2

(133)

where a1 and a2 are two constants depending on the cell shape. Thus for a given target SIR
value SIR0 , the reuse factor could be computed by:
1
N
a2

SIR0
a1

2/
(134)

The user capacity Cu is defined as the total number of active users per cell:
Cu =

B
G
NT
=
=
N
N Bs
N

(135)

where G = B/Bs is the ratio of the total system bandwidth to the bandwidth required for an
individual user.

17

Table of Q-Function
z
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9

Q(z)
0.50000
0.46017
0.42074
0.38209
0.34458
0.30854
0.27425
0.24196
0.21186
0.18406
0.15866
0.13567
0.11507
0.09680
0.08076
0.06681
0.05480
0.04457
0.03593
0.02872

z
2.0
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3.0
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9

Q(z)
0.02275
0.01786
0.01390
0.01072
0.00820
0.00621
0.00466
0.00347
0.00256
0.00187
0.00135
0.00097
0.00069
0.00048
0.00034
0.00023
0.00016
0.00011
0.00007
0.00005

The definition of the Q function is:

Z
Q(z) =
z

1
2
ey /2 dy
2

Two important properties of Q(z) are

Q(z) = 1 Q(z)
1
2
The Q-function is related to the erf and erfc functions by




1
z
1
z
Q(z) =
1 erf( ) = erfc
2
2
2
2

erfc(z) = 2Q( 2z)

erf(z) = 1 2Q( 2z)

Q(0) =

18