Exercise 5.

5: Large-scale log-normal fading

Since the system is designed to handle a propagation loss of 135 dB, outage will happen
when the propagation loss is 8 dB higher than the deterministic loss of 127 dB (135−127 =
8). With a σF = 7 dB log-normal fading, the outage probability, Pout , becomes
 dB   dB   
MLS 1 MLS /σF 1 8/7
Pout = Q = erfc √ = erfc √ = 12.7% (1)
σF 2 2 2 2

i. Increase TX power : YES, this will work, since it increases the fading margin. Spec-
trum regula ons may, however, prevent us from doing so.
ii. Decrease the deterministic path loss: YES, this may be possible. It can be done by
making a BS taller, for instance.
iii. Change the antennas: YES, this will also work.If the signal can be focused better the
total loss will become smaller. The focusing of energy will, however, make the system
more sensitive to the direction in which the TX/RX is in. Spectrum regulations,
limiting EIRP may also prevent us from doing this.
iv. Lower the σF : NO, this will not work in general. We would need to re-design the
radio propaga on environment, which is not exactly realistic.
v. Build a better RX : YES, this will work, in principle. The system can tolerate a lower
received power if a be er receiver is designed. It may, however, not be a practical
solution, since better receivers tend to be more expensive from several points of view.

Exercise 5.6: Small-scale Rayleigh fading

RX Sensitivity is rmin (signal amplitude) which leads to

rZmin  2 
r
Pout = Pr{r ≤ rmin } = cdf(rmin ) = pdf(r)dr = 1 − exp − min
2
(2)
rrms
0

1
Inserting the relation between the signal amplitude r and the instantaneous received
power C
C = Kr2 , K > 0 (3)
yields the expression
r2
     
Cmin /K Cmin
Pout = 1 − exp − min
2
= 1 − exp − = 1 − exp − (4)
rrms C̄/K C̄

Using the definition of the fading margin (eq. (30.5) in the book), M = C̄/Cmin , we get
 
1
Pout = 1 − exp − (5)
M
which gives us the required fading margin, in dB, as
 
M = −10 log10 − ln (1 − Pout ) (6)

Exercise 5.7: Approximation of (Rayleigh) fading margin

For small values of rmin , the outage probability, Pout , can be approximated as
 2  2
r rmin
Pout = Pr{r ≤ rmin } = cdf(rmin ) = 1 − exp − min ≈ (7)
2σ 2 2σ 2
2
rmin 2
rmin 1
Pout ≈ = = (8)
2σ 2 2
rrms M̃dB
which leads to
M̃dB = −10 log10 (Pout ) (9)

Define the error  as

   
 = MdB − M̃dB = −10 log10 − ln (1 − Pout ) − − 10 log10 (Pout ) , Pout > 0 (10)

And the largest error, −0.11 dB, occurs at the edge of the interval, namely where
Pout = 0.05 = 5%. Given how large uncertainties we normally have when designing
radio systems, this error is insignificant.

2
Exercise 5.9: Level crossing rate and average duration of fades

The level crossing rate of Rayleigh fading under isotropic scattering conditions (Jakes’
fading), expressed in terms of the received amplitude r threshold (eq. (5.53) in the book),
√ r − r
2
NR (r) = 2πνmax √ e 2Ω0 (11)
2Ω0
√
where 2Ω0 is the RMS value of the received signal amplitude. Hence, r2 /2Ω0 = C/C̄ =
1/M and the level crossing rate can be written
√ 1 1
NR (M ) = 2πνmax √ e− M (12)
M
When expressed in terms of amplitude, the product NR (r)ADF (r) = cdf (r) of the Rayleigh
distribution. Now, when expressed in terms of power instead of amplitude, the distribution
is exponential instead. Hence, the average duration of fades, ADF becomes
√
1 − e−1/M M  1/M 
ADF (M ) = =√ e −1 (13)
NR (M ) 2πνmax

Exercise 5.10: Large- and small-scale fading

• Distance (max) between antennas: dmax = 5 km

• Base-station height: hTX = 20 m
• Mobile-station height: hRX = 1.5 m
• Carrier frequency: f = 450 MHz
• Large scale log-normal fading: σF = 5 dB
RX = −122 dBm
• RX Sensitivity: Cmin
• Antenna gains: GRX = GTX = 2.15 dB
The propagation loss, L, is given by
2
d2
   
fMHz
L = Lground + LEgli’s = 20 log10 + 10 log10 = 139.4 dB (14)
hTX hRX 1600
The pathloss model is only valid if the distance is smaller than the radio horizon, dh
p p  √ √ 
dh ≈ 4100 hTX + hRX = 4100 20 + 1.5 ≈ 23 357 m > 5 km (15)

so we are good to go.

3
 Figure 1: Link budget

With the parameters given in the problem statement, the link budget becomes as shown
in Figure 1.

The fading margin required for the small scale fading, relative to the average received
power, with Pout = 5% is
 
Msmall = −10 log10 − ln (1 − Pout ) = 12.9 dB (16)

Since the propagation loss in the problem gives a median value, we need to compensate for
the difference between median and average power of the Rayleigh fading signal.
 2
2σ 2
 
r̄
Msmall,median = Msmall −10 log10 2 = 12.9−10 log10 = 12.9−1.6 = 11.3 dB
r50 1.182 σ 2
(17)

To obtain 95% boundary coverage, we need a fading margin Mlarge against σF = 5 dB

log-normal fulfilling  
Mlarge
Q = 5% (18)
5

4
which gives
Mlarge = 5 · 1.645 = 8.2 dB. (19)

The fading margin to be inserted in the link budget, given that we add the respective
fading margins, is

M = Msmall,median + Mlarge = 11.3 + 8.2 = 19.5 dB (20)

With this value inserted in the link budget we can calculate the required transmit
power.
PTX = −122 + 19.5 − 2.15 + 139.4 − 2.15 = 32.6 dBm. (21)

Exercise 6.7: CDMA Signals

See Section 6.2.2. on pages 104-105 in the book.

Ts = 1/W = 0.26 · 10−6 s (Inverse of the bandwidth).
Narrow-band if 1/W >> τmax , else wide-band.
With a maximum excess delay, τmax , of 1.3 µs and a chip duration, Ts , of 0.26 µs, the
multipath components fall in τmax /Ts = 5 delay bins. This means that we experience
leakage of energy between chips and the channel is therefore wide-band.
If the excess delay instead is 100 ns, then the channel is much shorter than a chip and we
do not experience any significant leakage of energy between chips. The channel is therefore
considered as narrow-band.

Exercise 7.2: Okumura and Okumura-Hata channel models

• Medium-size city
• hBS = 40 m
• hMS = 2 m
• f = 900 MHz
• d = 2 km

5
a: Okumura

Read values of figures in Appendix 7 in Molisch, and/or look in the lecture slides.
LOku = Lfree + Corrloss + CorrBS + CorrMS (22)
4π2 · 103
 
= 20 log10 + 24.5 + 10 + 2 (23)
3 · 108 /900 · 106
= 20 log10 (24π · 103 ) + 36.5 = 134 dB (24)

b: Okumura-Hata
   
a(hMS ) = a(2) = 1.1 log10 (900) − 0.7 · 2 − 1.56 log10 (900) − 0.8 ≈ 1.29 (25)
A = 69.55 + 26.16 log10 (900) − 13.82 log10 (40) − a(2) ≈ 123.40 (26)
B = 44.9 − 6.55 log10 (40) ≈ 34.41 (27)
C=0 (Medium-size city) (28)
These values inserted to the Okumura-Hata model yields
LOku-Hata = A + B · log10 (d) + C = 123.40 + 34.41 log10 (2) + 0 = 133.7 dB (29)

The two models agree well, which is quite natural since the second is a parameterization
of the first. The range of validity for the Okumura-Hata model is, however, a bit smaller.

Exercise 9.1: User influence on coverage

Let’s assume that the MS antenna is a patch antenna, since helix antennas are very rare
on terminals these days. Figure 9.1 then shows that the median user (User 21) decreases
the radiation efficiency by about 3.5 dB. Since the propagation exponent is 3.2, a 3.5 dB
reduction in the link budget gives
 3.2
d
10 log10 = −3.5 dB. (30)
37
Solving for d the new coverage distance becomes
d = 29 km. (31)