Professional Documents
Culture Documents
Practical Use
15th March 2002
Lee Kong Wah
Objectives
To choose appropriate propagation model
for different environment
To simplify the selected models to provide
the RF planners a quick-and-dirty coverage
estimate on sites
Content
1.
2.
3.
4.
5.
6.
Introduction
Large Cell
Small Cell
Microcell
Hilly Environment
Sea Environment
7. In Building
8. Tunnel
9. Model tuning
Introduction
RF Propagation Model = Mathematical
Representation of a Radio Transmission
Environment
Empirical Model
Deterministic Model
Mixed
Introduction
Empirical Model
Based on measurements
All environmental influence taken implicitly
Accuracy depends on similarity between
measurement and estimation environments
Efficient computation
Usually applied in large outdoor cell
E.g. Okumura-Hata, COST231 Walfisch-Ikegami
and etc
Introduction
Deterministic Model
Based on physics principles
All Individual propagation are considered
More Accurate and applicable for different
environment
Less efficient in computation
Usually Applied in microcell and Indoor
E.g. Ray Tracing
Large Cell
Large Cell
Okumura-Hata Model
PL = 69.55 + 26.16 log(f) - 13.82 log(hb) - a(hm) + [44.9 - 6.55 log(hb)]
log(d) dB K
Where :
f - operating frequency (MHz)
hb - Base station antenna height (m)
hm - mobile station antenna height (m)
a(hm) - correction factor for mobile station antenna height
(dB)
d - distance from Base station (Km)
Large Cell
Type Of
Area
Open
Suburban
Mediumsmall City
Large City (f
> 400)
a(hm)
[1.1 log(f) 0.7]hm
[1.56 log(f) 0.8]
K
4.78(logf)2
18.33logf 40.94
2(log(f/28))2 + 5.4
0
3.2(log11.75hm)2
4.97
Large Cell
Model Limitation
Large Cell
Simplify Model for quick estimate on sites
f = 925MHz
hb = 30m (countryside, village)
hm = 1.5m (primary coverage objective)
K = 10dB (suburban category)
a(hm) = 0 (primary coverage objective)
Large Cell
Plot Graph to estimate d vs hb (fix PL)
PL calculated from Link Budget (130dB)
Large Cell
To estimate PL at close proximity of BTS
g = 4hbhm/
PLd<g = 32.4 + 20log(d) + 20log(f)
PLd>=g = 42.6 + 26log(d) + 20log(f)
Large Cell
To estimate at close proximity but with single
diffraction
Large Cell
Previous profile is normalized to be :
v = h((2(d1 +d2)/(d1d2))1/2
where h = hB -hb (approx.)
Large Cell
Prx = ERP 101.9 - 26log(d) + Ldiff
Small Cell
Small Cell
COST231-WI
PLNLOS = Lfs + Lrts + Lmsd
(or PLNLOS = Lfs for Lrts + Lmsd <= 0)
where
with
Small Cell
Small Cell
Lmsd = multiscreen diffraction loss
= Lbsh + ka + kd*log(d) + kf*log(f) 9*log(b)
with Lbsh = 18*log(1 +Hb HB)
=0
ka = 54
for Hb > HB
for Hb <= HB
for Hb > HB
= 54 0.8*(Hb HB)
= 4 + 1.5*(f/925 1)
for H b > HB
for Hb <= HB
for medium sized cities and
suburban centres with
moderate tree density
for metropolitan centres
Small Cell
With a free LOS between base and mobile (Street
Canyon)
Microcells (Base station antennas below roof top
level)
PLLOS = 42.6 + 26*log(d) + 20*log(f)
for d >= 0.020 km
Model Limitation
Small Cell
Simplify Model for quick estimate on sites
f = 925MHz
w = 25m (typical street width)
b = 50m (typical buildings separation)
PL allowed = 130dB (typical IB link budget)
Small Cell
For hb <= hB
Small Cell
(38-15(hb-hB)/hB)log(d)1.6log(hb-hB)d=12.6-20log((hB-1.5)
for d < 500m
(38-15(hb-hB)/hB)log(d)0.8log(hb-hB)d=12.6-20log((hB-1.5)
for d >= 500m
Microcell
Microcell
COST 231-Microcell
PLNLOS(GSM 900)=101.7+26log(d)+20n
d > 0.020 m
Where n = number of corners encountered in propagation
Hilly Environment
Irregular elevation of the surrounding land
Often NLOS between BTS & MS
Various models developed by extending OH
Hilly Environment
Terrain Clearance Angle Method (TCA)
Additional terrain loss based on TCA is
obtained and added to OH PL
Developed by EBU, adopted by CCIR
PL = PLOH + CTCA
Where
PLOH = Path Loss (dB) estimated using classic Okumura-Hata model
CTCA = attenuation/gain in dB due to terrain
= 14.9 (6.9 + 20log(((v-0.1)2+1)1/2+v-0.1))
v = -93.1
= Terrain Clearance Angle (radian), refer to next diagram
Hilly Environment
Hilly Environment
Simplify for use, PLOH is found from below
graph
155
150
145
140
135
130
125
120
hb=15m
hb=20m
hb=25m
hb=30m
hb=35m
Distance (Km)
3.4
3.2
2.8
2.6
2.4
2.2
1.8
1.6
1.4
1.2
hb=40m
1
PL (dB)
PL vs Distance
Hilly Environment
TCA loss is found from below graph
Hilly Environment
E.g. to estimate RxLev at 2Km away from a
BTS in hilly area assuming
ERP = 56dBm, hb=30m, TCA = -3 degree
PL = PLOH + CTCA
PLOH = 139dB
(from PL vs Distance Graph)
CTCA = 10dB
(from Clearance vs graph)
RxLev = ERP-PL = 56-149=-93dBm
(50% coverage area probability)
Sea Environment
Environment with huge water body
surrounding the BTS
Sea, big lake or river
E.g. National sea territory, ferry routes
No empirical model encountered
Measurement done and recommendation
available in ITU-R P.370
PL curves available and can be used
Sea Environment
Original curves
between 10 &
1000Km
Extrapolation done
between 1Km to
10km
2 slopes observed
Tx ERP = 60dBm
Hm = 10m
Cold sea
Y-axis in dBm/m
PL (dB) = ERPP(dBm)
= 60 E(dBV/m)
+138.7
= 198.7 E(dBV/m)
Sea Environment
Sea surface coverage estimate using curves
E.g. 140dB PL allowed (from Link Budget),
37.5m height BTS Ant, MS Ant height 3m
140dB PL yields 59dBv/m on the 37.5m curve,
E(dBV/m)=198.7-PL(dB)
Apply hm correction, Cm(dB)=-0.67*20log(hm/10) if
hm<10m
Sea Environment
From ITU-R P. 370, dh = 4.1(hb)1/2
hb(m)
Cell Range(Km)
dh(Km)
37.5
15
25
70
25
75
As a rule of thumb, Cell Range<dh for hm=3m
If hm is higher, e.g. 10m, cell range is then
closer to dh
Same curves used for RxLev estimate
In Building Environment
Antenna placed indoor
Cell Range dictated by building geometry,
materials, furniture & etc.
E.g. corporate building, shopping complex,
stadium or airport
Model can be deterministic or empirical
Empirical model for first level design, used
to determine number of BTS/Ant required
In Building Environment
ITU-R P.1238
LID = 20 log(f) +Nlog(d) +Lf (n) 28dB + X
where
N = distance power loss coefficient
f = frequency (MHz)
d = separation distance (m) between the BTS/antenna and the
MS (where d >1 m)
Lf(n) = floor penetration loss factor (dB)
n = number of floors between BTS/antenna and MS (n 1).
Xcoverage probability margin to overcome indoor shadow
fading
In Building Environment
Power loss coefficients, N, for indoor transmission loss calculation
Residential
Office
Commercial
33
20
1.8-2 GHz
28
30
22
Floor penetration loss factors, Lf(n) (dB) with n being the number of floors
penetrated, for indoor transmission loss calculation (n >=1)
Floor penetration Lf(n) (dB)
Frequency
(GHz)
Residential
Office
Commercial
900 MHz
9 (1 floor)
19 (2 floor)
24 (3 floor)
1.8-2 GHz
4n
15+4(n-1)
6+3(n-1)
In Building Environment
Simplify model assuming f=925Mhz, X=8dB
LID = 39 +Nlog(d) +Lf (n)
E.g. ERP=39dBm, Omni antenna at 3rd floor,
building of 50m x 50m x 30m (LxWxH)
P4 = 39dBm PLID = 33log(25) + 0
= -46dBm
P3 = 39dBm PLID = 33log(27) + 9
= -56dBm
P2 = -68dBm
P1 = -73dBm
Tunnel Environment
E.g. Road, Train, Underground tunnel
Tunnel Propagation considered is between a
BTS Ant and the MS located inside the
vehicle throughout the tunnel
No discussion on leaky propagation
Train tunnel propagation expected to
encounter higher signal blockage due to the
effect of the train itself
Tunnel Environment
Wave guide model most commonly used
Deterministic Model considering
Reflection, scattering and diffraction effect
Computer Aided computation
E.g. http://www.ubka.uni-karlsruhe.de/
Measurement and prediction result available
from the above
Can be used for first level tunnel design
Measurement is important and critical for
higher confidence design
Tunnel Environment
Straight, rectangular, single lane, concrete
12dBi directional antenna placed in tunnel
/4 receiver on a lorry, no penetration loss
Higher signal fluctuation at
short d
Steeper loss at short d
compared to higher d
Ducting effect observed
overall where loss increase
gradually with d
Tunnel Environment
Circular, curved, single lane, concrete
Same Tx and Rx setup as previous one
Tunnel Environment
Higher signal fluctuation at short d
Steeper loss at short d
Tunnel Environment
For installation & maintenance purpose, BTS Ant
preferred outside of tunnel
Prediction done with the deterministic model (tuned)
To examine effect of internal/external BTS Ant
BTS Ant at 2.5m above ground, on tunnel axis, inside
& 30m away
Omni directional Ant used
Rectangular tunnel measuring 10m x 5m
Tunnel Environment
At d=0, RxLev of internal Ant is much higher than
external Ant
This is due to less energy coupled into tunnel for
external antenna
Higher fluctuation in
short d for internal Ant
PL vs d is less steep for
both cases at large d
At large d, both curves
merged
Tunnel Environment
In external BTS Ant installation, exact
height & location also affect propagation
Prediction done & effect examined
Dipole vertical Ant, 1GHz
First, Ant at 20m from tunnel, centric
position, 0.5m, 2.5m, 4.5m above ground
Tunnel Environment
BTS Ant height of 2.5m yields least PL
Due to its centric position, provides most direct
illumination into tunnel
Ant height of 4.5 causes highest PL due to
ground reflection
Tunnel Environment
Second, Ant fixed at 2.5m height, distance from
tunnel varied from 1m, 20m, 50m, 80m, 120m
Same Tx & Rx settings, prediction run and result
observed
Further external Ant is,
lower RxLev at tunnel
entrance
Steeper loss for all cases
at short d
All curves except 120m
merged at 800m
Tunnel Environment
Dominating Tx mode
Beyond Breakpoint
Tunnel Environment
Last, different lateral Ant position set and run
prediction
Ant height set to 2.5m, 2 distance of 20m &
80m from tunnel set, 3 lateral Ant position I.e.
central, 5.5m to left & right
Left position with least
overall PL, due to most direct
illumination
Lateral effect found less
profound with further Ant
distance
Tunnel Environment
Measurement done in a straight road tunnel by
Huawei
10dBi Directional antenna at 50m from tunnel, 8m
lateral shift, 2.6m Ant height
Rx antenna inside the DT car
140
120
100
80
Series3
60
40
20
2457
2268
2079
1890
1701
1512
1323
1134
945
756
567
378
189
Tunnel Environment
It is doubtful that waveguide model works for
train tunnel
Beyond breakeven point, where dominating ray is
direct ray, it is believed that higher PL encounter
due to train-inflicted penetration loss
To be sure, do measurement
If not, as temporary solution, use the waveguide
model or measurement curve only for road tunnel
For Train, do not exceed 675m
Dominating Tx mode
Beyond Breakpoint
Breakpoint
Model tuning
CW Test & Model Tuning
Flow Chart For Model Tuning
Test Preparation
Test Preparation
Propagation Test
Data Processing
Model Tuning
Propagation Test
Equipment
Transmitter Setup
(CW)
Coordinate
Correction
Import CW
measurement
Data Processing
Transmission
Method
Receiver Setup
Measurement
Averaging
Plot PL vs d, find
K2
Format
Conversion
Adjust K1 to
reduce error
Model Tuning
Site Selection
Channel Selection
(CW)
Power Setting
(CW)
Drive Route
Determination
Measurement
Collection
Transmitter
dismantle (CW)
Model tuning
CW Test Preparation
CW or Live GSM
CW Transmitter, Fast Scanner, Ant
Select suitable test sites (clearance, surrounding clutter,
accessibility)
RF channel & Tx Pwr selection
Drive route determination
CW Test
Model tuning
Measurement Data Processing
Correct DT data coordinates with respect to GIS
Model tuning
Load CW data
Run analysis to get report
Plot PL vs d graphs with ASSET or EXCEL
Obtain K2 from graph & adjust K1 to minimize mean
error
Filter out NLOS data & analyze them
Adjust K7 to minimize mean error
Analyze & obtain error/clutter report
Enter clutter offset based on reported error
K3, K4 usually not tuned unless different varying Hm
faced
K5, K6 usually not tuned unless sufficient measurement
with various Hb
Model tuning
E.g. Prediction Error Distribution before & after
tuning
Narrower error SD, 0 mean error
Model Error Distribution
6000
5000
Number of Bins
4000
3000
2000
1000
0
-48 -44 -40 -36 -32 -28 -24 -20 -16 -12 -8 -4 0 4 8 12 16 20 24 28 32 36 40 44 48
-1000
Error (dB)
Calibrated
Non-Calibrated