COVERAGE PROBABILITY VERIFICATION FOR CELLS IN URBAN RADIO NETWORK

PLANNING
S. BOUZOUKI
(1)
, I PANOUTSOPOULOS
(1)
, C. IOANNOU
(1)
S. KOTSOPOULOS
(1)
, C. SORAS
(2
(1)
Wireless Laboratory
(2)
Laboratory of Electromagnetics
Dept. of Electrical and Computer Engineering
University of Patras
Rio
26500 Patras-Greece
ABSTRACT
A methodology for verification coverage probability in
small cell structure in urban environments under
multipath and shadowing conditions is presented. The
shadowing-fading channels in urban areas with a large
number of slow-moving pedestrians and vehicles, follows
a composite lognormal shadowing/Rayleigh fading
distribution. The estimation of the cell radius is based on
received signal strength measurements and both the short-
term and long-term signal distributions. The presented
methodology is useful for optimal radio network planning
since it effectively determines the geographic extent of
reliable RF coverage. This information is crucial for
handoff decisions and frequency allocation pattern in
order to delimitate the small cells in the best possible way.
KEYWORDS
RF coverage, quality, cell edge reliability, coverage
validation
1. Introduction
With the third generation of mobile cellular systems
shifting to higher frequencies, there is an increasing
migration to smaller cells that is further driven by the
growing demand for wireless service. A successful cell
planning is characterized from two requirements: a) The
absence of co-channel interference and b) the minimum
number of required handoffs [1]. To satisfy the above-
mentioned requirements, the decision for implementing
particular cell sizes is crucial. The upper bound of the cell
size is determined by the traffic demand per area and the
limited transmission range due to power limitations
(frequent handoffs) and infrastructure cost (base
stations/km
2
). The minimum Quality of Service (QoS)
within the cell is achieved when the received signal power
has a certain level in at least 90%-95% of the coverage
area. This QoS parameter is often referred to as the
coverage reliability.
The coverage is defined in terms of the cell edge
probability P
R
, i.e., the probability that sampled signal
strength is above a particular threshold at the cell edge and
the cell area probability P
A
, which is the probability that
the received signal strength is above a threshold for entire
cell area.
For optimal cell planning the signal prediction is
imperative. The propagation models currently in use give
a prediction for the large-scale path loss and do not take
into account the actual multipath fading or shadowing
variation. To verify the coverage both short-term and
long-term fading should be included with the appropriate
confidence margins.
The cell planning in urban areas is based in small cell
structure. The implementation of a small cell can be
achieved by using BTS (base transceiver station) antennae
located above the medium height levels but below the
maximum height of the surrounding rooftops. The
maximum range for small cells is typically less than 1-3
km. The focus of this paper is to verify the cell edge
probability. Knowing the cell edge probability the cell
area probability is automatically verified. The proposed
methodology takes into account both short-term and long-
term fading as well as parameters that depend on the
existing environment.
2. Proposed Methodology
The proposed method verifies coverage using a closed
form expression for the signal reception probability, which
combines lognormal shadowing and Rayleigh fading
distribution to estimate the coverage area reliability. In the
present examination the cells are assumed to be circular. It
should be emphasized that this approach does not in any
way require that the true cell edge be circular. The
propagation path loss over the urban area is modeled using
a hybrid propagation model. A fade margin P
FM
based on
the actual signal variation within each cell is calculated to
ensure the desired cell edge reliability.
2.1 Cell Edge Estimation
The cell edge is defined as the region on or around the cell
radius R. The estimation of cell radius can be performed
by utilizing coverage prediction algorithms whose
parameters are optimized for the specific terrain of
interest. Signal strength measurements have to be taken
uniformly over the area of interest and the path loss
coefficient is then determined via a linear regression fit
with constant term. The received power P
r
(dBm) at the
edge of a cell R is given by the formula [2]:
P
r
=P
t
+ G
t
+ G
r
-L
m
-L
D
-PL
=P
t
+G
t
+G
r
-L
m
-L
D
-C-10nlog
10
R (1)
Where P
t
is the transmitting power of the base station, G
t
the transmitter antenna gain, G
r
the receiver antenna gain,
L
m
represents intermediate losses, L
D
is the diffraction
losses, C an unknown constant and n the path loss
exponent, both estimated from the measured RF data via
linear regression. The factor Lm is taken into account
regardless of the environmental type (rural, urban,
suburban, open area) and is derived from the equation:
(2)
ln0.5
100
yleigh percent_Ra
ln
10log
100
g percent_lo
2 1
1
erf
2
sd
en
L
m
L
1
1
1
1
]
1

,
_

+
,
_

+
In equation (2) L
en
represents the environmental losses (or
land cover correction factors [3],) and sd is the lognormal
standard deviation (typical 6 to 10 dB range). The terms
percent_log and percent_Rayleigh correspond to the
statistical desired radio coverage. L
en
is a value that
minimizes the overall mean square error between
predicted field values and measurements taken over a
limited area. The diffraction losses L
D
in urban and
suburban environments are calculated by the UTD-based
approach for cellular mobile propagation. The
propagation model used in this paper along with its
performance is shown in [2].
If A = P
t
+ G
t
+G
r
- L
m
-L
D
- C, the probability that the
received signal level will exceed a certain value T can be
calculated from the cumulative density function [4,5]:

,
_

,
_

>
2
) log 10 (
2
1
2
1 log 10 (
] ) ( [
10 10
sd
r n A T
erf
sd
r n A T
Q T r P P
R R
(3)
The fade margin P
FM
that ensures the desired cell edge
reliability P
R
can be approximated as:
R FM
P sd P 2 (4)
In the cell border according to the desired threshold and
the fade margin, the received power P
rb
will be:
FM rb
P T P + (5)
Using the equations (1) to (5) the distance R
edge
from a
base station at any desired signal strength threshold and
cell edge probability can be estimated. The derived
formula is:

,
_

n
A P
edge
rb
R
10
10 (6)
From the cell planning engineer point of view, it is
important to have the knowledge of the probability of
handoff request P
H
to complement the coverage
probability.
) ( 1
edge R H
R P P (7)
The requirement for absence of co-channel interference,
co-channel cells must be physically separated by a
minimum distance to provide sufficient isolation due to
propagation. In urban areas the size of each small cell is
approximately the same and the base stations transmit the
same power so the co-channel interference ratio is
independent of the transmitted power and becomes a
function of the radius of the cell and the distance (D)
between the centers of the nearest co-channel cells. In this
case, the signal to interference ratio (S/I) is given by the
equation:
( )
0
/
i
R D
I
S
n

(8)
where i
0
is the number of co-channel interfering cells.
2.2 Cell Area Estimation
The cell area probability P
A
is dependent on the cell edge
probability P
R
. This conclusion is derived from the fact
that usually the signal strength degrades the most at the
cell border. So, if there is sufficient coverage at the cell
edge then there will be coverage in the whole cell.
Reudink first established the formula used to estimate the
reliability of RF coverage over a circular area, [4]:
1
]
1

1
]
1

,
_

,
_

+
b
b
erf
b
b
erf T P
A

1
1
2 1
exp ) ( 1
2
1
) (
2
(9)
where:
2
) log 10 (
10
sd
R A T

(10)
2
log 10
sd
e n
b
(11)
3. Simulation Results
The performance of the proposed methodology was testing
in the urban area of Patras (Greece). Patras is a city with
quite narrow streets, having building heights in the range
of 20-30 m in the old part and center. The main building
materials all over the city are concrete blocks and bricks.
Also Patras has a particular characteristic, namely a large
number of squares (small open areas). Extensive
experimental tests were performed in the geographic area
of Patras and especially in the city center. The aim of
these tests was to investigate and to study the RF
propagation in different directions relative to the streets.
The results of those measurements were used to define the
standard deviation, the path loss exponent and the path
loss coefficient C [6]. It is worth to notice that in the
center of the city, the handoffs were occurred at a
threshold in the order of 70 dBm. Theoretically the cell
planning is designed for a threshold of 97 dBm.
The cell area probability P
A
can be controlled with the
transmitter parameters but depends also on the
propagation channel. Figures 1 to 3 show the impact of
the path loss exponent and standard deviation on P
A
with
P
R
(or P
H
) as a parameter. Figure 4 and 5 show PH as a
function of the distance between mobile and base stations
with n as a parameter. These results are derived from a
GSM 900 system with 10 W transmission power, base
station antenna height 25-35 m and mobile station antenna
height 1.8 m.
Figure 1: The cell area probability PA versus path loss exponent n for
different cell edge probability with threshold 97 dBm and standard
deviation 6.17 dB
Figure 2: The cell area probability PA versus path loss exponent n for
different cell edge probability with threshold 70 dBm and standard
deviation 6.17 dB.
Figure 3: The cell area probability PA versus standard deviation sd for
different handoff probability, threshold 97 dBm and n=4
Figure 4: The handoff probability versus the distance between mobile
and base station for different n, threshold 97 dBm and standard
deviation 6.5 dB. (The dash, the dot, the solid and the dashdot line is
respectively for n=2, 3, 4, 5)
Figure 5: The handoff probability versus the distance between mobile
and base station for different n, threshold 70 dBm and standard
deviation 6.5 dB. (The dash, the dot, the solid and the dashdot line is
respectively for n=2, 3, 4, 5)
It was also found that the small cells, which are
responsible to cover the center of Patras have different
effective radius and standard deviation as shown in table I.
P
H
PR=90%
PR=80%
PR=70%
PR=60%
PR=50%
PA
P
H
PA
PR=90%
PR=80%
PR=60%
PR=70%
PR=50%
EXPERIMENTAL
RADIUS
THRESHOLD STANDARD
DEVIATION
0,620 km -93 dBm 10
0,816 km -78 dBm 10
1,091 km -95 dBm 8,5
1,127 km -71 dBm 7,63
1,308 km -75 dBm 7,6
1,664 km -83 dBm 9
Table I: Measurement results for the small cells of a GSM900 provider
in the center of Patras
As far as the reuse distance, the simulation results using
the formula (8) for a cluster of 7 cells and for the same
type of environment (sd=6 dB, n=3,6) are shown in table
II:
CELL EDGE
PROBABILITY P
R
THRESHOLD REUSE DISTANCE
(km)
70% -97 dBm 3,1
80% -97 dBm 3,5
70% -70 dBm 2
80% -70 dBm 2,24
Table II: Reuse distances for different cell edge probability and
threshold
4. Conclusions
The disadvantage of existing RF prediction tools used by
GSM providers is the lack of information about the fading
and shadowing variation. An optimum cell planning
requires higher quality link budget estimation. The present
methodology to estimate the probability of adequate
reception at a particular location satisfies the above
requirement. This achievement was obtained due to
information of the statistical channel parameters for
specific environment, which have been taken into account
along with the prediction of handoffs regions.
REFERENCES
[1] Theodore S. Rappaport: Wireless Communications
Principles & practice, Prentice Hall Communications
Engineering and Emerging Technologies Series
[2] S. Bouzouki, S. Kotsopoulos, G. Karagiannidis: On
the Optimum Approximation of the Personal
Communication Networks RF cell area, IEEE VTC99,
Houston, Texas
[3] RMTP/RB/R105, ISSUE 1.0, APPENDIX 2, 20
th
November 1990
[4] D. O. Reudink, Microwave Mobile Communications,
W. C. Jakes, Ed. Pitcataway, NJ: IEEE Press, 1993, ch. 2,
pp 126=128
[5] Pete Bernardin, et al: Cell Radius Inaccuracy: A new
measure of coverage reliability IEEE Tran. On Veh.
Techn. Vol. 47, No 4, November 1998
[6] S. Bouzouki, S. Kotsopoulos, G. Karagiannidis: RF
Modeling: Propagation Characteristics in the Frequency
Band of 900/1800 MHz for Patras Area, Int. Workshop
on Mobile Com. Focused on UMTS and IMT 2000,
June 24-26 1999, Chania, Crete, Greece