coverage predictions cpp06r3b

Coverage Predictions

Chapter 6

This chapter is designed to provide the student with an overviewof elaborate radio wave coverage predictions.

OBJECTIVES:Upon completion of this chapter, the student will be able to:

• Explain why different models are used in differentenvironments

• Discuss more elaborate radio wave propagation models andthe predictions they can make

GSM Cell Planning Principles

EN/LZT 123 3314 R3B

ll aa nn kkBB

iioonnaall llyy

ttnneettnnII

6 Coverage Predictions

EN/LZT 123 3314 R3B – i –


Table of Contents

Topic Page

INTRODUCTION..................................................................................51

FLAT CONDUCTIVE EARTH ..............................................................52

KNIFE EDGE DIFFRACTION ..............................................................53

FIELD MEASUREMENTS AND SEMI-EMPIRICAL MODELS ............54

ALGORITHM 9999...............................................................................57

URBAN MODEL...................................................................................58

MICROCELL MODELS..............................Error! Bookmark not defined.

OVERVIEW.................................................................................................................. 59

LINE-OF-SITE MODELLING ....................................................................................... 59

NON LINE-OF-SITE MODELLING .............................................................................. 59

IN-BUILDING MODELLING ......................................................................................... 60

OPEN AREA MODELLING.......................................................................................... 60

POINTS AFFECTED BY MULTIPLE SOURCES......................................................... 60


– ii – EN/LZT 123 3314 R3B

ll aa nn kkBB

iioonnaall llyy

ttnneettnnII


EN/LZT 123 3314 R3B – 51 –

INTRODUCTION

It is important to be able to estimate cell coverage, not only todetermine the size of the cell, but also to be able to estimateinterference. The definition of coverage is usually the following:an area is considered covered if in 95 percent of that area, thesignal received by the mobile station is larger than somerequired value, e.g., -90 dBm. In order to achieve this, thepredicted signal strength at the cell border must be larger thansome design value, e.g. SS design = -85 dBm i.e.,

3LQ06�

�SUHGLFWHG��≥�66GHVLJQ

The signal strength required and design values are estimated byadding margins to the MS receiver sensitivity. These are fast andslow fading margins, interference margins, margins for bodyloss, and possibly additional margins for in-car and indoorcoverage. The margins depend on the type of environment andoperator requirements.

It is very important to be able to estimate the signal strength inall parts of the area to be covered, i.e., to predict the pathloss.There are more elaborate models than the one discussed inchapter three, “Radio Wave Propagation”. Improvements can bemade by taking into account:

• the fact that radio waves are reflected towards the earthsurface (the conductivity of the earth is then an importantparameter)

• the transmission losses due to obstructions in the line ofsight

• the finite radius of the curvature of the earth

• the terrain type in a real case, as well as the differentattenuation properties of different land usages such asforests, urban areas, etc.

The best models used are semi-empirical, i.e. based onmeasurements of path loss/attenuation in various terrains. Theuse of such models are motivated by the fact that radiopropagation cannot be measured everywhere. However, ifmeasurements can be performed in typical environments,parameters of the model can be adjusted so that the modelbecomes a good approximation for that particular type of terrain.This chapter briefly discusses a few of these more elaboratemodels.


– 52 – EN/LZT 123 3314 R3B

FLAT CONDUCTIVE EARTH

MobileBase

h1

h2

d

Figure 6-1 Radio wave propagation over flat conductive earth

In Figure 6-1, reflections against the surface of the earth aretaken into account. If we assume an unobstructed propagationthrough free space, the signal at the receiving antenna can beseen as the sum of one direct signal and the reflected signal, ifwe also assume that the earth is a perfect conductor (hardly agood assumption, except possibly for sea water), i.e. loss freereflection, this yields (for the received power at the receivingantenna) the interference term:

( )3

3* *K KG

GU

W U W

=

λ

πλ

π

2 2 1 2

2

2

2

sin

which is the squared sum of the field amplitudes from the directand reflected wave. (See chapter three, “Radio WavePropagation” for the explanation of the symbols.) Assuming thath1h2<<λd (i.e., small angles), the sine function can be replacedwith its argument (radians) and so

( )3

3* * K K

GU

W U W= 1 2

2

4

or

( ) ( )/33

GK K

* *W

U

U W=

=

− −10 20 10 10

2

1 2

log log log log

The expression 20log(G��K�K�)) corresponds to the path loss.


EN/LZT 123 3314 R3B – 53 –

KNIFE EDGE DIFFRACTION

Additional path loss due to objects obstructing the line of sightcan be taken into account by calculating the (Fresnel) diffractionpattern at the receiver. The intensity is a function of the height ofthe obstruction above (or below) the line of sight as well as thedistances transmitter-object and receiver-object (Figure 6-2).

TX RXh

d1 d2

Figure 6-2 Knife edge diffraction

Derivations of the expression is somewhat lengthy, so here wemust be satisfied with expressing the additional attenuationcaused by these so-called “knife edges” in a diagram (Figure 6-3). The additional attenuation is read as a function of theparameter ν, which is given as

( )ν

λ=

+K

G G

G G1 2

1 2

ν

0

Figure 6-3 Knife edge diffraction loss as a function of ν


– 54 – EN/LZT 123 3314 R3B

FIELD MEASUREMENTS AND SEMI-EMPIRICAL MODELS

The models discussed previously do not take into account thetopographical variations in a real environment nor the differentattenuation properties of different land usages such as forests,urban areas, etc. Although calculations taking all details intoaccount are possible, they are tremendously time consuming andnot practical to use for the cell planner. Indeed, empirical datacan be used. An example of such data is shown in Figure 6-4.This figure is a depiction of measurements made in 1968 by aJapanese engineer, Okumura. It is interesting to note that thefree space model yields consistently higher field strengths. Thatis, it yields lower path loss than the measurements.

Okumura made measurements in various types of terrain, eachyielding a new set of curves. However, the diagrams can only beused as a rough guide since terrain types differ from place toplace, and local variations in the topography as well as in theland usage cannot be accounted for.

Figure 6-4 Okumura’s field measurements displayed as thefield strength 1.5 m above ground as a function of thelogarithmic distance between base and mobile. Curves areshown for different effective base station antenna heights, h1.


EN/LZT 123 3314 R3B – 55 –

Empirical data can be used to improve more elaborate models.In particular, the attenuation as a function of terrain type andland usage, e.g., in 1980, Hata presented a number of semi-empirical formulas based on the field measurements made byOkumura.

As an example the expression for the path loss in urban areas isgiven as:

/S(urban) = 69.55 + 26.16log I - 13.82logK

E + (44.9 - 6.55logK

E)log G

- a(KP),

where

a(KP) = (1.1 log I - 0.7)K

P - (1.56 log I - 0.8)

I = carrier frequency in MHz (150 - 1000 MHz)

KE = the base station antenna height in meters (30 - 200 m)

G = distance in km from the base station (1 - 20 km)

KP = mobile antenna height in meters above ground (1 - 10 m)

Note that this particular formula is strictly valid only for urbanareas in Japanese “quasi-smooth” terrain, but it is still useful forrough estimates of cell coverage. For the same type of terrain,this formula can be adjusted to yield

/S(suburban) = /

S(urban) - 2 [log (I/28)]2 - 5.4

/S�open) = /

S(urban) - 4.78 (log I)2 + 18.33 log I - 40.94

for two other types of land usage.

Another model worth mentioning is the Cost 231-Hata modelwhich can be used when the carrier frequency is in the interval1500-2000 MHz, as in the case of GSM 1900. Here

/S�= 46.3 + 33.9logI - 13.82logK

E�+ (44.9 - 6.55logK

E)logG - a(K

P)

Ericsson has developed simple Okumura-Hata type modelsbased on wave propagation measurements which can be used toestimate the coverage for both GSM 900 and GSM 1800. Here

/S�= A - 13.82logK

E�+ (44.9 - 6.55logK

E)logG - a(K

P)


– 56 – EN/LZT 123 3314 R3B

where

a(KP) = 3.2(log 11.75K

P)2 - 4.97

and

A(900) = 146.8 and A(1800) = 153.8 for urban areas

A(900) = 136.9 and A(1800) = 146.2 for suburban areas

A(900) = 118.3 and A(1800) = 124.3 for open areas


EN/LZT 123 3314 R3B – 57 –

ALGORITHM 9999

The prediction model used by Ericsson, Algorithm 9999 (validbetween 0.2 - 100 km), is based on ideas similar to Hata’s in thatempirical data is used to fit parameters for the attenuationcaused by different land usages (clutters). Regardless of whatmodel is actually used, the accuracy of the prediction dependsheavily on the field measurements performed.

Semi-empirical models are used because radio propagationcannot be measured everywhere. Algorithm 9999 takes intoaccount the land usage by the use of digitized map data as wellas knife-edge diffraction and effects of the earth’s curvature.The algorithm, which is incorporated in the EricssonEngineering Tool (EET) packet and TEMS Cell Plannercalculates the path loss for radio waves between two spatialcoordinates. From the antenna, the path loss is calculated in alldirections for an arbitrary distance from the antenna. Thesepredictions give an accuracy of about ±5 dB if the parameters inthe model have been optimized from field measurements ofsignal strengths. From these predictions, various “arrays” can becalculated. The arrays are based on cell data, e.g., the outputpower of the TRXs, the antenna gains, and the frequencyallocation to the different TRXs are quite fast to calculate sincethe path loss is known and does not change unless the antenna ismoved.

These arrays are very useful and EET/TEMS Cell Planner canplot them on maps to obtain a graphical display of thepredictions. Signal Strengths (SS), Carrier-to-Interferer (C/I),and Carrier-to-Adjacent (C/A) ratios and other information canbe plotted, thus helping the cell planner verify the nominal cellplans and/or improve the system design.

In addition to Algorithm 9999, EET/TEMS Cell Planner alsosupports other models including the Cost 231-Walfish-Ikegamimodel (valid between 0.02 - 5 km) which can be used in urbanareas because it uses street orientation, building heights,building separations, and road widths. However, for urbanenvironment the preferred model is the Urban Model, se below.


– 58 – EN/LZT 123 3314 R3B

URBAN MODEL

In an urban environment, there are mainly two paths for radiowave propagation:

• Over the roof tops

• Along the street

At a far distance from the site, the first part dominates but nearthe site, the second one dominates. The Urban Model is aconcept of two different wave propagation algorithms:

• Half-screen model

• Recursive microcell model

The half-screen model is used for calculating the propagationabove the roof tops. Obstacles such as buildings and treesbetween transmitter and mobile are replaced with a number ofscreens with heights correlated to the heights of the obstacles.The path loss is then calculated by using a multiple knife-edgeapproach.

The recursive microcell model calculates the propagation overopen areas, e.g., along streets. The exact locations of thebuildings are used for defining the propagation paths. The pathloss is calculated by determining the so-called illusory distancebetween transmitter and mobile in a street system.

This model is incorporated in EET/TEMS Cell Planner

coverage predictions cpp06r3b

Documents