AIRCOM_CW-Drivetest

Wireless Facilities Incorporated

Introduction to Field Measurements Ver. 1.1

Proprietary of WFI, Do not Copy

1

Outline

n

Introduction, scope and objectives

n

Sampling Issues in Field Measurement

n

Drive Test for Model Optimization – Test Site Selection – Transmitter Set up – Drive Test Guidelines

n

Drive Test for Coverage Verification – Contour vs. Area Reliability

n

Appendices: – Review of dB units – Review of Safety Remarks


2

Introduction, Scope & Objectives

Depending on the design phase drive test is performed with different objectives: n Predesign drive test for measurement integration: – This is at beginning of design when no site has been built or even selected. All test sites are temporary/dummy sites possibly put on a crane. – At this point drive test is performed mostly for characterization of propagation and fading effects in the channel. The objective is to collect field data to optimize or adjust the prediction model for preliminary simulations. – Depending on the measurement tools it may also provide us with statistics of fading and delay spread in the channel to be used as part of LBA and future optimization studies. n Postdesign drive test for site verification/optimization – At this point the initial set of site are selected. – Drive test is performed to verify if they meet the coverage objectives. – The hand-off areas are also checked.


3

Outline

n


n


n


n


n



4

Propagation Loss and Fading

Variations in the received signal in a fading channel results from three types of phenomena: n

Propagation Loss: – Depending on the environment it is about 20-50 dB loss per decade. – Example 40dB corresponds to inverse 4th power law i.e. Loss = K/d4, K=Constant, d= Distance.

n

Slow Fading: – It is due to shadowing. – It is models as a lognormal random process

n

Fast Fading: – It is due to multipath effects. – It is modeled by a Rayleigh or Ricean Distribution.


5


n n n

In field measurements we are interested only in variations due to propagation and slow fading. The received signals are typically sampled and averaged over spatial windows called bins. There are several sampling issues to be considered: – Sampling Rate – Averaging Window – Number of bins to be measured


6

Sampling Intervals

RS < v/0.8λ

d > 0.8λ

d distance n n

n n

n

To have a meaningful average on each window we have to measure enough number of independent samples. The measurement points or sampling intervals have to be far enough to result in independent samples. Mathematically, to ensure small correlation, the sampling points have to be at least half the carrier wavelength, i.e. λ/2, apart. In practice the number 36 samples in every 40 wavelengths, i.e. 0.8λ is used. Thus the minimum window size is roughly 36 samples or 40λ which for 1900MHz is about 7 meters. Depending on the speed of the vehicle during the drive test, v , the minimum sampling interval in time or the maximum sampling rate RS can be selected. In particular to ensure independent samples, measurement have to be stopped whenever the vehicle is not moving.


7

Averaging Window

W n

n

n

n

n

In field measurements we are mostly interested in local averages of received signals or the overall effects of propagation combined with shadowing. The size of averaging window have to be small enough to capture slow variations due to shadowing and large enough to average out the fast variations due to multipath. The maximum size of this window is a function of the variations in the terrain, e.g. hilly or dense urban area vs. flat rural areas. A typical range is 20-1500 meters. Therefore the bin size is typically selected in 40λ < W < 1500meters Typically the post processing tool takes care of averaging the collected data over different bins and performing the so called measurement integration.


8

Number of Bins to Measure W W

n

n

n n

By comparing the predicted and measured signal strengths for all measured bins along the drive routs we estimate the best set of correction factors to minimize the prediction errors. Obviously we can not drive test all bins within the coverage area, so a large enough sample set should be considered. The reliability of estimated correction factors depends on the amount of data, i.e. the number of bins covered in the drive test. The more number of bins included, the larger is the confidence level of results. For acceptable confidence in the process model optimization, it is recommended that at least 300-400 of bins, not samples, be measured and considered.


9

Outline

n


n


n


n


n



10

RX-Power P

Model Adjustments

Measured Data: Pm

P(r,C): Default Model

P(r,C+∆ ∆C): Model after Adjustments distance r

n n n

n

We typically start with a default model, e.g. a variation of Hata’s Propagation Model. This model is a parametric model with some coefficients. The correct choice of these coefficients for a given environment can enhance the accuracy of our predictions. One of the main objective of drive testing is to provide real data based on which the the best set of coefficient that minimize the prediction error are estimated. Different propagation tools such as PlaNet and CellCAD provide means of setting these coefficients.


11

Test Site Selection

n n

The test site should represent all the characteristics, e.g. height and antenna of a typical site. The test site should – – – –

n

n

provide easy access to the roof. have access AC power source. Preferably have a building engineer be free of close obstructions

Using cranes is costly but sometimes inevitable. It is preferable to reserve them in advance from a an operator with some experience in RF test sites. Also for crane sites, power generators should be arranged in advance. For predesign drive test usually omnidirectional antennas are used and measurements are made all around the site.


12

Test Site Selection (cont.)

n

n

n

Propagation characteristics depends on the environment and land usage or morphology, e.g. urban, suburban or rural. In a market with significantly diverse morphological areas one has to drive test and adjust the model for each area separately. For example 3 test sites in urban and 3 sites in suburban can be used as test sites for drive test. The model used for urban sites is optimized based on the drive test on the urban sites measurements whereas the model used for suburban site is adjusted based drive tests for suburban sites.


13

Site Setup Guidelines

n

Verify Antenna Placement – It should be mounted at least 2-3 feet away from metallic structures. – It has be as close as possible to the final location. – For roof top sites the antenna has to have some clearance above the roof so that the roof fringes do not block the propagation.

n n

n n

n

n

Do not hold the transmit antenna with bare hand while its transmitting Never operate a transmitter without a load , e.g. an antenna or a resistive terminator, attached to it. Open circuit operation may damage the internal amplifiers. Do not transmit with the receiver in the proximity, i.e less than 50 feet distance. This may desensitize the front end circuits. Put the transmitter set in an area which provides it with some shelter. Also maintain backup transmitters and receivers in case of failure. To verify proper operation, calibrate the receiver before testing and check transmit power both before and at the end of drive test. Absolutely do not climb towers.


14

Losses and Gains in the Forward Link Budget GA ERP Power Amplifier

PA

Combiner Loss, Connector Loss & Cable Loss

LCCC In dB scale:

n

n

n

Input Power To Antenna

ERP = PA - L CCC + GA

The Emitted Radiated Power (ERP) of the antenna equals the power amplifiers output minus all combiner, connector and cable losses plus the antenna gain. The objective of this measurement is to estimate the EPR of the test site. In order to estimate this value we need to find the power that is entering into the antenna and for that we need to put a Wattmeter in the loop as close as possible to the antenna. Usually for pre design drive tests omnidirectional antennas are used.


15

Transmitter Set up

Cable A

Cable B Watts

Dummy Load

Cable A Watts

n n n

n

n

This figure shows the measurement setup system. Set the transmitter to the correct frequency.This frequency should not be in use in the existing systems in the area. The powermeter reads the input power to the antenna provided the loss due to cable B is small. To ensure small loss for cable B we need to put the powermeter as close as possible to the antenna i.e. make cable B as short as possible (e.g. maximum 4-5 feet). Alternatively, if a dummy load (50 Ohm resistor) is available, we can replace cable B and the antenna with the load. Either way, note down power reading of the Wattmeter.


16

Reflected Power

Cable A

Cable B Watts

Dummy Load

Cable A Watts

n n n

Using the same setup with bi-directional wattmeters we can read the reverse power, reflected from the antenna. This power has to be less that 10 percent of the power in the forward direction. Otherwise impedance matching has to be corrected.


17

Drive Test Plan

n

n n

n

In a drive test plan both radial and azimuthal routs have be tested. In urban area the effect of street orientations have to considered. Both Line Of Site (LOS) and non-LOS points have to be included in the drive test. Measurements have to be unbiased, i.e. the data collected should represent all typical coverage scenarios. The selection of drive test routs should be based on the: – – – – –

n

The terrain variations subscriber distributions Major highways and thoroughfares Critical Areas Potential Shadowing areas and hand-off regions

Distance from site to be driven depends on the cell size, typically 2-3 time the cell radius is drive tested.


18

Drive Test Guide lines

n n

n

Engineer should study the drive test plan ahead of time and highlight the intended drive test routs. It is recommended that for each drive test a team two people, including one engineer, get involved. For safety reasons the driver should not read maps while he is driving. Also drive testers should always stay vigilant and careful about the measurement process they should turn off the measuring device behind traffic lights or whenever the sampling and measurements look suspect.


19

Drive Test Deliverables: Data Files

Lat. Long. RSSI Freq. .....

X1 Y1 M1 F X2 Y2 M2 F X3 Y2 M3 F ........

n

n

n

As a result of drive test and data collection a set of tables which correlated the location of each measurement point with its corresponding signal strength and frequency is recorded. The location information, latitude and longitude, along the drive test rout is constantly measured by navigation equipment such as GPS, LORAN or ETAK navigator. The location information is used by postprocessing tools as a reference of correlation between measured vs. predicted signal levels for measurement integration. It is also used for data display for diagnosis and testing purposes.


20

Outline

n


n


n


n


n



21

Drive Test for Coverage Verification

n

n n

n

In order to verify that a candidate site, selected based on its acceptable predicted coverage area, is actually covering all objective areas for enough confidence we need to perform drive test. Since drive testing every possible point is impossible we need to resort to sampling and statistical analysis. The drive test also tells us about potential interference problems or even hand-off problems for the site. At this point the test site has to be placed, if it is not already built, at the selected location with all the parameters that have been determined based on simulations. Unlike measurement integration the site may not necessarily have an omni configuration.


22

Nonuniform Sampling Least Critical More Critical

More Critical n

Less Critical

More Samples

Most Critical

Although, theoretically, samples have to be taken uniformly in the cell area, in practice we prioritize areas. – Areas close to cell boundaries or under shadow, and hand-off areas are sampled more closely than the areas very close to the base station were very high signal levels are expected. – Areas were most calls are made, i.e. regions with high density of potential subscribers have to be more densely sampled and inspected. – Like predesign drive tests all major roads and critical areas, e.g. intersections, convention centers major business areas have to be included in the drive test. – Also coverage along streets of different orientations has to be inspected.


23

Checking Coverage Objectives •Sufficient Signal Strength or Ec/It •Sufficient overlap for Proper (soft)hand-off •Least interference to other cells

n

n n

n n n n

Depending on the type of receiving environment, e.g. on street, in car or in building coverage the objective coverage thresholds are different. The objective coverage threshold is usually determined after link budget analysis. Also the coverage area of each cell in conjunction with other cells, is roughly estimated and defined based on composite coverage plots resulted from simulations The actual coverage of the cell has to be tested against the objective coverage based on the corresponding threshold. Also (soft)hand-off areas have to be tested for sufficient overlap. The site should not cause too much interference outside its objective coverage area. The coverage objectives are typically given in terms of area reliability.


24

Contour Reliability

n

n

n

Due to the shadowing and variations in the terrain even if at a fixed distance signal level changes. These random changes are modeled by log-normal distributions. Because of these random fluctuations, for any given coverage T threshold and distance R there is a probability that the signal is above that threshold. This probability is called Contour Reliability. For example 90% contour reliability on a circle means that, as the mobile moves on the perimeter of that circle, at 90% of the points it receives signals above the coverage threshold.


25

Area Reliability J.J. J. J. J. J. J. J.L. J. J. J. J. J. J. J. L. L. J. J. J. J. J. J. J. L. J. J. J. L. n

n n n

n

80%

90%

Another way of expressing coverage reliability which is more realistic and is specified as part of design objectives or customer requirements is the area reliability. Area Reliability is the percentage of area where the received signal is above the threshold. It can be thought of as the average of contour reliability's for all circles of radii r, 0 < r < R. As we reduce the radius of the circular contour the corresponding points become closer to the base station and the contour reliability increases. Area reliability is always larger than the contour reliability. 99% 97%

95% Proprietary of WFI, Do not Copy

94% 90%

26

#Samples vs. Confidence Interval 1.5 2

9.6

9

8.4

7.8

0.5% 7.2

6.6

6

5.4

3.6

3

2.4

1.8

1.2

4.8

C=99%

0.5%

4.2

0.001031 0.001594 0.00242 0.00361 0.005291 0.007617 0.010774 0.014969 0.020432 0.027397 0.036089 0.046702 0.059369 0.074143 0.090962 0.10963 0.129801 0.150974 0.172508 0.19364 0.21353 0.231314 0.246164 0.6

0

0 0.2 0.3 0.4 0.6 0.8 0.25 1 1.2 1.4 0.2 1.6 1.8 2 0.15 2.2 2.4 2.6 0.1 2.8 3 3.2 0.05 3.4 3.6 3.8 0 4 4.2 4.4

e=2.575σ z= 2.575 n

n

Given the cell is divided into N bins, what is the minimum number of measured bins Nm needed to warrant p% area coverage reliability with C% confidence: Nm = max { Z 2 P(1-P) /e2 , 5/(1-P)} – Z=Standard Deviation Unit (e.g. Z=2.575 for C=99% confidence) – P=Claimed Area Coverage Reliability – e=Half the length of confidence interval (e= zσ)


27

Confidence Level 1.5 2

9.6

7.8

7.2

6.6

6

5.4

4.8

4.2

3.6

3

2.4

1.8

1.2

9

(1-p)/2

(1-p)/2

8.4

0.001031 0.001594 0.00242 0.00361 0.005291 0.007617 0.010774 0.014969 0.020432 0.027397 0.036089 0.046702 0.059369 0.074143 0.090962 0.10963 0.129801 0.150974 0.172508 0.19364 0.21353 0.231314 0.246164 0.6

0

0 0.2 0.3 0.4 0.6 0.8 0.25 1 1.2 1.4 0.2 1.6 1.8 2 0.15 2.2 2.4 2.6 0.1 2.8 3 3.2 0.05 3.4 3.6 30. 8 4 4.2 4.4

z

Efrc ( z ) = Erfc( x ) = n n n n

1 2π

1− p 2

∞ −t /2

∫x

e

dt

We use a gaussian distribution with unit variance as a reference. This figure shows the relationship between p and z for such a normalized guassian distribution. For a given p, the z that gives the required one sided (1-p)/2 area can be read from error function tables. For the actual distribution e=z.σ σ has to be considered,


28

Case Study n

n n

For a Cell of radius r=4km, to ensure p=90% cell area reliability with C=95% confidence we accept a confidence interval length of e=+/-5% The number of bins to be measured Nm=137 (table) Total number of bins N > 10 Nm=1370

for 90% area coverage reliability e% : Conf. Interval length

#bins for 95% Confidence Level

1

#bins for 99% Confidence Level 5940

2

1485

857

3

660

381

4

311

180

5

238

137


3428

29

Outline

n


n


n


n


n



30

Appendix: A:Review of dB units B: Review of Safety Remarks


31

Review of dB Units n n n

dB is a logarithmic unit for power gains and losses. Gain G=P/P0 = g dB where g=10 log (P/P 0) Examples: – A gain of 100 is equivalent to 20dB gain – A 10 times attenuation in power = -10 dB loss

n n

By fixing P 0 as a reference power, e.g.... to 1 Watt or 1Miliwatt, one can define similar units for power. Examples: – (P) dBw = 10 log P/(1Watt) – (P) dBm = 10 log P/(1mW)

n n n n n

dB is also a logarithmic unit for voltage gains and losses. Gain G=V/V0 = g dB where g=20 log (P/P 0) Since power is proportional to voltage squared the two definitions are consistent. Similarly by fixing V0 as a reference voltage, e.g.... to 1 volt or 1microvolt, one can define similar units for voltage. Examples: – (V) dBV = 10 log V/(1Volt) – (V) dBu = 10 log V/(1µW)


32

Common Mistakes Regarding dB units dBm+dBm dB*dB dBm/dBm dBm+dB=dBm dBm-dB=dBm dB+dB=dB n

n n

Incorrect

Correct

Remember the difference between dB as a unitless measure of gain or loss and dBm as a unit of power or voltage. Also note that addition in the logarithmic scale e.g.... in dB domain is like multiplication in the linear scale. Therefore the following are meaningless and not correct: – Adding two signal levels in dBm domain. – Multiplication of gains or losses expressed in dB’s – Looking at the ratio between two signal in dB domain.


33

dB to linear conversion & vice versa n n n

Let’s consider two signals S1 with power P1 Watts or Q1 dBm S2 with power P2 Watts or Q2 dBm

n

From Watts to dBm – (Q 1)dBm =10 log (P1 W/1mW)=10 log(103 P1)=30 + 10 log P1

n

From dBm to Watts – (P1)mW= 10(Q1/10) and (P1)W=10 -3 x 10(Q1/10)

n

Adding two signals has to be in the linear domain: – S1+S2 = P 1 + P 2 = 10 Q1/10+ 10 Q2/10

n

Q1 +Q2

The ratio between two signals or signal to noise ratio has to be calculated in the linear domain – S1/S2= P1 / P 2 = (Q1-Q2) dBm


Q1/Q2

34

Test Site Selection

n n

The test site should represent all the characteristics, e.g. height and antenna of a typical site. The test site should – – – –

n

n

provide easy access to the roof. have access AC power source. Preferably have a building engineer be free of close obstructions

Using cranes is costly but sometimes inevitable. It is preferable to reserve them in advance from a an operator with some experience in RF test sites. Also for crane sites, power generators should be arranged in advance. For predesign drive test usually omnidirectional antennas are used and measurements are made all around the site.


35

Safety Remarks

n n

n n

n

n n n

Do not hold the transmit antenna with bare hand while its transmitting Never operate a transmitter without a load , e.g. an antenna or a resistive terminator, attached to it. Open circuit operation may damage the internal amplifiers. Do not transmit with the receiver in the proximity, i.e less than 50 feet distance. This may desensitize the front end circuits. Put the transmitter set in an area which provides it with some shelter. Also maintain backup transmitters and receivers in case of failure. To verify proper operation, calibrate the receiver before testing and check transmit power both before and at the end of drive test. Absolutely do not climb towers. Engineer should study the drive test plan ahead of time and highlight the intended drive test routs. It is recommended that for each drive test a team two people, including one engineer, get involved. For safety reasons the driver should not read maps while he is driving.


36

AIRCOM_CW-Drivetest

Recommend Documents