Near-Earth RF Propagation – Path Loss and Variation with Weather

Ashley Larsson, Andrew Piotrowski, Timothy Giles, Darryn Smart

Department of Defence, DSTO, EWRD Edinburgh SA 5111, Australia

Ashley.Larsson@DSTO.defence.gov.au

Abstract— Most of the current radio frequency propagation

analysis is focused on signals travelling well above the earth’s surface. A recent surge in low power communication networks having RF signals propagating close to the earth’s surface has resulted in some networks using antennas operating at heights less than one metre from the surface. The resultant near ground RF propagation and the effect of obstructions that impinge upon the first Fresnel zone for short ranges is not well understood. It is unclear as to which propagation model to apply in this situation. Knowledge of the propagation conditions is essential to the design of radio communications for distributed radars such as those used for perimeter protection. DSTO is investigating RF propagation and evaluating the effects of weather for these propagation conditions. Propagation loss measurements were made at UHF frequencies using antennas set at heights varying between 0.3 m and 1.5 m in rural line of sight environments. Measured RF propagation loss results for these conditions are in very good agreement with the two-ray RF propagation model. In addition, the dew and cloud coverage can significantly affect the RF propagation loss.

Keywords—RF Propagation near ground; radar unattended ground sensor, propagation loss, atmospheric effects, wireless sensors

I. INTRODUCTION Most of the radio frequency (RF) propagation analysis has

generally been focused on signals that propagate between two antennas well above the earth’s surface [1]. These analyses resulted in the development of RF propagation models that enable prediction of RF signals at the receiver in various environments. However, there is an increased interest in analysing RF propagation for antennas at low heights, usually located near to the ground. This communication configuration is important for wireless sensor applications in areas such as wireless personal networks (Mobile Ad Hoc Networks) [2], unattended ground sensors and RADAR for border patrol and military, environmental [3], agricultural, mining and structural monitoring [4][5]. In these applications the antennas are small, located close to the earth’s surface usually at heights less than 1.0m, separated by maximum distances of a few hundreds meters with and without direct line of sight, and operate in a wide range of frequencies from VHF to UHF. Since these antennas are located near the ground, terrain obstructions and surface conditions significantly affect the RF propagation, which may also be affected by atmospheric phenomena such as dew, frost, moisture evaporation and solar radiance. Some of the work on near ground short range RF propagation was reported by [6-12]. These papers concentrate on propagation for the near ground antennas at frequencies of 1 GHz and above. For near ground RF propagation, the RF signal is subjected to multiple reflections and diffraction from large

terrain obstacles (compared to its wavelength), scattering from the roughness of the terrain surface and vegetation [6], absorption by vegetation, and refraction in ducts close to the ground surface that are affected by temperature, humidity and air pressure. DSTO has measured near ground RF propagation in two rural settings at frequencies 433 MHz, 650 MHz, 920 MHz, 1800 MHz and 2200 MHz with vertical polarisation. RF propagation loss was measured as a function of receive-transmit antenna separation using log-periodic antennas positioned at several heights up to 1.5 metres above the surface. These measurements enabled the influences on RF propagation loss to be observed, such as surface types and roughness, environmental parameters and the weather. This paper is divided into two parts. The first part (sections II to V) describes the effects of surface physical impairments to propagation, including the effects of the ground, distance, and antenna heights on the propagation loss. The second part (section VI) describes the effects of weather on propagation loss.

II. MEASUREMENTS RF propagation loss measurements for various transmit and

receive antenna separation distances were made in two rural locations, one on a bitumen road and the other in a freshly mown dry grassy field with vegetation no taller than 5 cm. The setup for these measurements is shown in Figure 1. The measurement system consisted of a calibrated log-periodic AH&S SAS-510-7 transmitter antenna permanently located at a fixed position and set at a height of 0.3 m, and a mobile receiver antenna system that was moved to different separation distances. The receive antenna consisted either of the calibrated log-periodic ARRONIA hyperlog antenna (enclosed in RF transparent plastic) set at a height of 0.5 m or the EDS 3142D log-periodic antenna set at a height of 1.5 m. The antennas were mounted for vertical polarisation. The transmit signal was generated by the Agilent N5182A vector signal generator (VSG) and amplified by either the ASD 2042CFRAAXLXX 1-1000 MHz or 2132CFFAAXLXX 500-2500 MHz amplifier. The received signal was when necessary (for the long separation distances) first amplified using a Lucix 10M060L4001 amplifier and then recorded using the Agilent N9342C spectrum analyser. Both transmit and receive systems were controlled using a laptop via optical fibre based Ethernet connections. The spectrum analyser was set with the centre frequency of the transmitted signal, a bandwidth of 1 MHz, a resolution bandwidth of 10 kHz and trace capture with maximum hold. The VSG generated CW RF signals with a bandwidth of 0.5 Hz at 433 MHz, 650 MHz, 920 MHz, 1800 MHz and 2200 MHz respectively, then amplified to 33 dBm (2 W) at the transmit antenna input. For calibration

978-1-4673-5178-2/13/$31.00 © 2013 Commonwealth of Australia RADAR 201357

purposes the level of a received signal passing through a 50 dB attenuator, instead of via the antennas and environment, was recorded. This measured received level was used for calibration by subtracting it from received levels passing through the environment.

For each measurement point, three signals of the same frequency were emitted and each was recorded. The median received power level of these three recordings was used to calculate the propagation loss.

Fig. 1. RF propagation measurement equipment setup.

III. MODELS One of the most widely used near ground models is the Egli

semi-empirical RF propagation model [13], based on VHF and UHF propagation data collected for television signals. For this model the 50% RF power propagation loss is represented by:

⎟⎟⎠

⎞⎜⎜⎝

⎛=

rxtx

sdB hh

fdPL40

log202

10)(50 , (1)

where f is the frequency in MHz, and hrx , htx and ds are the receive antenna height, transmit antenna height and antenna separation distance respectively, all in metres.

The two ray RF model is one of the most commonly used coherent RF propagation models that takes into account direct and reflected signals [4]. The direct signal travels distance r1 and the reflected signal travels from the transmit antenna to the ground and is reflected back to the receive antenna travelling a

distance r2, with 221 Δ+= sdr and 22

2 Σ+= sdr , where ds is the separation distance, Δ = hrx--htx is the difference in the receive and transmit antenna heights and Σ = hrx-+htx is the sum of the antenna heights. In the two ray model the RF propagation loss is expressed by the following equation

( ) ( ) ( )2

22

11

2

exp1exp14

jkrr

jkrr

PL v −Γ+−⎟⎠⎞

⎜⎝⎛= α

πλ

(2)

where ( ) ( ) ( )( ) ( ) ( )αεαε

αεαεα2

2

cossincossin)(

−Χ−+Χ−−Χ−−Χ−

=Γjjjj

v

is the reflection coefficient, Χ = 60λσ, α is the grazing angle, k (= λ

π2 ) is the wave number, λ is the wavelength, and ε and σ are the relative dielectric constant and conductivity of the reflecting surface, respectively. For antennas at low heights the reflection coefficient approaches -1 and is usually not affected by the type of reflecting surface [1].

Fig. 2. Reflection coefficient as a function of grazing angle for dry ground and dry bitumen.

TABLE I. THE REAL PART OF THE REFLECTION COEFFICIENT FOR DRY BITUMAN AND DRY GROUND FOR TWO GRAZING ANGLES

Electric properties Grazing angle

ε σ (Ώ/m) 0.17º 3.43º

Dry bitumen 4.3 0.00555 -0.9861 -0.7520

Dry ground 15.0 0.001 -0.9765 -0.6131

The magnitude of the reflection coefficients of dry bitumen and dry ground were calculated using (2) for the two ray model and are shown in Figure 2. The relative dielectric constants for dry bitumen and dry ground are shown in Table I. In our measurements the grazing angle ranged from 0.17 degrees to 3.43 degrees and the calculated real part of the reflection coefficient is shown in table 1. The imaginary part of the reflection coefficient was negligibly small. Consequently the phase of the reflection coefficient is near –π and for long distances, the reflection coefficient is close to -1.

The two ray model can be seen as two circular electromagnetic wave-fronts (one direct and one reflected) with different centres. The two wavefronts will constructively and destructively interfere depending on the differences in distances travelled by the two wavefronts. Consequently peaks and nulls occur. Equation 3 is a rearrangement of (2) for long distances so the reflection coefficient is -1.

( )( )2

122

2

1

2

1

exp14

rrjrr

rPL −−−⎟⎟

⎠

⎞⎜⎜⎝

⎛= λ

π

πλ

(3)

Amplifier

VSG

Laptop

Spectrum Analyser

Ethernet Switch

Fibre Optic

Convertor

TRANSMITTER RECEIVER

40dB Amplifier

58

Assuming a flat earth model and long distances, the difference between the direct and reflected signal paths can be calculated by [4] as:

s

rxtxd

hhss ddrrd 22222

12 ≅Δ+−Σ+=−=Δ . (4)

The nulls of (3) occur when the path difference is an integer value of the wavelength or zero. The peaks occur when the path difference is an odd integer multiple of the half wavelength. The last peak occurs when the path difference equals the half wavelength and is called the ‘break point’. Beyond this distance the path difference is smaller than half a wavelength, and destructive interference occurs. So the distance of the break point can be calculate by 2

λ=Δd and is

λrxtx

bpthh

d4

≅ . (5)

Thus, beyond λrxtxhh4 only destructive interference from the

reflected signal occurs. Figure 3 shows the calculated RF propagation loss for the two ray model against separation distance when the receive and transmit antennas are at the same height and the reflection coefficient Γ = -1. For these conditions sdrr ≈≈ 21 and (3) can be simplified to

( ) λ

λπ

πλ

/4

4sin

4

4

2

22

rxtxss

rxtx

s

rxtx

s

hhdfordhh

dhh

dPL

>≅

⎟⎟⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛=

. (6)

For long separation distances, (6) is similar to the Egli model. The six curves in Figure 3 represent the propagation losses for a 400 MHz signal when the receive and transmit antennas are both set at 5.0 m, 3.0 m, 2.0 m, 1.0 m, 0.5 m and 0.25 m, with the ‘break points’ marked with diamonds. At ranges shorter than the ‘break point’ RF propagation losses exhibit the interference pattern and increase with a second power distance dependence. Beyond this point the propagation losses increase with a fourth power distance dependence and no interference pattern occurs. All measurements in this paper were beyond the break point.

The two ray model though informative does not take into account obstructions that are within the Fresnel ellipsoid. A prominent obstruction in the Fresnel ellipsoid causes diffraction of the direct and reflected signals. The reflected signal is also affected by electrical properties of the surface as well as the surface roughness resulting in scattering that, depending on roughness of the surface, can cause coherent and diffused scattering. To help analyse the effect of the obstruction on the RF propagation the first Fresnel ellipsoid is used. This ellipsoid is defined as the ellipsoidal volume with the receive antenna and the transmit antenna as foci. When there are no obstacles impinging on the first Fresnel ellipsoid volume the two ray model would be sufficient. However, when the first Fresnel ellipsoid is obstructed, the path loss becomes

significantly greater than line-of-sight propagation. Assuming a flat earth model, the radius of the nth Fresnel ellipsoid can be calculated by equation 7 [1], hence the first Fresnel ellipsoid radius is

txrxs

txrx

txrxn

ddfordF

ddddnF

==

⎟⎟⎠

⎞⎜⎜⎝

⎛+

=

,21λ

λ (7)

where dtx and drx are distances from the transmit and receive antennas respectively and ds=dtx+drx. The Fresnel ellipsoid model is valid only for line-of-sight cases where there is a direct signal path or the direct path is partially obscured between the transmitter and receiver, and for this reason the model is applied only to line-of-sight topographies. For obstructed topographies, a single regression path loss needs to be used [4].

-110

-100

-90

-80

-70

-60

-50

-40

-30

-20

1 10 100 1000

Separation distance (m)

Prop

agat

ion

loss

(dB

)

htx=hrx=5.0m htx=hrx=3.0m htx=hrx=2.0mhtx=hrx=1.0m htx=hrx=0.5m htx=hrx=0.25m

Fig. 3. Propagation loss at 400 MHz for the two ray model for both antennas set at 5 m, 3 m, 2 m, 1 m, 0.5 m and 0.25 m height.

The knife edge diffraction of an obstruction at height, Hob, relative to the line-of-sight between the transmit and receive antennas modifies the electric field (Eob) in comparison to no obstructed values (E) according to [18]

⎟⎟⎠

⎞⎜⎜⎝

⎛+=

⎟⎠⎞

⎜⎝⎛ −+⎟

⎠⎞

⎜⎝⎛ −=

rxtxob

ob

ddHwith

SCE

E

112

)(21)(

21

21 22

λν

νν, (8)

where C(ν) and S(ν) are Fresnel integrals, ν is the knife edge diffraction parameter, dtx and drx are distances of the obstruction from the transmit and receive antennas respectively and ds=dtx+drx. Equation 8 can be approximated to [1]

59

( ) .1062.05.0log20

10}{

−≥>−−≈

−<≈

νν

ν

for

fordBE

Eob

(9)

For the current test set-up, the ground surface partially obstructs the Fresnel ellipsoid. For distance shorter than 10 m diffraction loss is at most 2.36 dB. The worst case obstruction occurs at 433 MHz frequency with the receive and transmit antennas 100 m apart and at heights of 0.5 m and 0.3 m respectively. The obstruction is 3.345 m into the first Fresnel ellipsoid. From (8), this obstruction results in a knife edge diffraction parameter ν= -0.13 and a reduction in electric field strength by 4.7 dB. This reduction was not included in the subsequent two ray path modelling, but these calculations provide an indication of the maximum amount of loss expected in the obstructed Fresnel ellipsoid.

Surface roughness increases the total energy spread and results in the reduction of the energy reflected in the specular direction. The Rayleigh and sometimes the Fraunhofer criteria [14] are used to consider if the roughness of the surface is sufficient to cause scattering for the critical surface height, ψ for an angle of incidence θ, as shown in equation 10.

( )

( ) Fraunhofer

Rayleigh

θλψ

θλψ

cos32

cos8

≤

≤ (10)

To account for surface roughness the reflection coefficient is modified to decrease the amount of energy in the specular direction, but this coefficient does not include the spread of energy into other directions. Ament [15] was one of the first to derive the rough surface scattering loss factor coefficient,

⎟⎟⎠

⎞⎜⎜⎝

⎛⎟⎠⎞

⎜⎝⎛−=

2)cos(8expλ

θψπρ sr . (11)

This equation was later modified for Gaussian like surface roughness [15],

,)cos(8

.)cos(8exp

2

0

2

⎥⎥⎦

⎤

⎢⎢⎣

⎡⎟⎠⎞

⎜⎝⎛

⎟⎟⎠

⎞⎜⎜⎝

⎛⎟⎠⎞

⎜⎝⎛−=

λθψπ

λθψπρ

I

sr

(12)

where I0 is the modified zeroth order Bessel function.

For our test set up the incident angle varies from 89.82 to 86.56 degrees and using (10) with the test frequency of 2200 MHz the critical surface height σ is in the range 5.8 m to 0.29 m, and 1.5 m to 0.07 m, for the Rayleigh and Fraunhofer criteria respectively. These critical surface heights are significantly higher than the variation in the test surface heights and thus the effect of surface roughness does not have to be taken into account in our modelling. Vegetation may also scatter the RF signal when their height is comparable with the

wavelength of the RF signal [1], however our field test site had very sparse and dry vegetation.

IV. MEAN PATH LOSS EXPONENT The mean path loss exponent is one of the simple measures

used by engineers for modelling RF propagation loss. The mean path loss exponent n determines the rate at which propagation loss increases with distance, as represented in (13).

( ) ( )

( )s

n

o

sdBsdB

dnPL

dddPLdPL

100

100

log10

log10

+

=⎟⎟

⎠

⎞

⎜⎜

⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛+=

(13)

where d0 is a reference distance, typically 1 m, and the propagation loss at this distance is PLdB(d0).

For free space the loss exponent is n=2, for rural areas it is between n=3 to 5 and for suburban areas n=3.5 to 6 [1].

V. RESULTS The RF propagation measurements as a function of

separation distance were conducted for a fixed transmit antenna height of 0.3 m, receive antenna heights of 0.5 m or 1.5 m, and test frequencies 433 MHz, 650 MHz, 920 MHz, 1800 MHz and 2200 MHz.

Fig. 4. RF propagation loss as a function of distance on an bitumen road for test frequencies 433 MHz, 650 MHz and 920 MHz, with htx=0.3 m and hrx = 0.5 m and 1.5 m.

Figure 4 shows measured propagation loss, marked as symbols, against separation distance for the two receive antenna heights (1.5 m and 0.5 m) at frequencies 433 MHz, 650 MHz and 920 MHz on the dry bitumen road. In addition to these results, the two ray propagation modelling results for the same conditions are shown in Figure 4 as lines. The mean path loss exponent calculated from the least squares fit for these results is between 3.51 and 3.83 for the 1.5 m receive antenna height and between 3.82 and 3.89 for the 0.5 m receive antenna height. The measured results are within 4.8 dB of the least square fit. The biggest deviation between the mean path

60

exponent and experimental results occurs for short separation distances where the incident angle is not close to zero and the real part of the reflection coefficient Γ is furthest from -1. These experimental results correlate well with the two ray model, with a maximum error between the two ray model and the experimental results of 5.4 dB. It should be noted that the experimental results follow the reduction in propagation loss for short separation distances (less than 15 m) as predicted by the two ray model. The best agreement between the experimental and modelling results was achieved for the dry bitumen conditions.

Fig. 5. RF propagation loss as a function of distance on a grassy field for test frequencies 433 MHz, 650 MHz and 920 MHz, with htx=0.3 m and hrx = 0.5 m and 1.5 m.

Figure 5 shows measured propagation loss, marked as symbols, against separation distance for the two receive antenna heights (1.5 m and 0.5 m) at test frequencies 433 MHz, 650 MHz and 920 MHz on the grassy field. The two ray modelling results for the same conditions are marked as lines. The least squares fit of the mean path loss exponent from the measured data is between 3.28 and 3.71 for the 1.5m receive antenna height and between 3.41 and 3.85 for the 0.5m antenna height. The measured results were within a maximum of 6.4 dB from the best line fit for both antenna heights, with the biggest errors occurring at the short separation distances. The maximum error of the mean path exponent occurs at the shortest separation distances. The experimental results correlate well with the two ray model, with a maximum error between the model and the experimental results of 5.9 dB. It should be noted that the experimental results follow the reduction in RF propagation loss for short separation distances (less that 15 m) as predicted by the two ray model. The spread of the experimental results is bigger for the field measurements than for the road measurements and results in the results spread wider than predicted differences in RF loss with frequency.

In a similar fashion Figure 6 shows the experimental and modelled RF propagation loss for the 0.5 m receive antenna height at test frequencies 1.8 GHz and 2.2 GHz on both the grassy field and dry bitumen road as well as the two ray modelling results. The mean path loss exponent of the least squares fit of the results is between 3.47 and 3.67 for the field

and 3.81 and 2.92 for the road. The measured data is within 4.2 dB of the least squares fit. The experimental results correlate well with the two ray model results, with a maximum error between the model and the experimental results of 6.1 dB. The experimental results again follow the decrease in RF propagation loss for short separation distances (less than 15 m) as predicted by the two ray model.

Fig. 6. RF propagation losses as a function of distance on graasy field and dry bitumen for test frequencies 1.8 GHz and 2.2 GHz, with htx=0.3 m and hrx = 0.5 m.

For all frequencies and surfaces tested, the measured results are within 4-6 dB of the modelled results. The measured path loss exponents are given in Table II. All mean path exponent values are between 3.41 and 3.89, except for the test field at 433 MHz with the receive antenna at 1.5 m for which the exponent is 3.28. The experimental results are also fitted with (2), and the experimental results show a similar trend to the two ray model at distances shorter than 15 m. The best least square fits of the exponent of the experimental data and the two ray model results at the separation distances of the measured results are shown in Tables II and III respectively.

TABLE II. MEAN PATH LOSS EXPONENT FROM MEASUREMENTS

Site hrx (m)

433 MHz

650 MHz

920 MHz

1800 MHz

2200 MHz

Road 1.5 3.51 3.77 3.83 Road 0.5 3.82 3.85 3.89 3.92 3.81 Field 1.5 3.28 3.64 3.71 Field 0.5 3.41 3.82 3.85 3.67 3.42

TABLE III. MEAN PATH LOSS EXPONENT FROM 2 RAY MODEL

Site hrx (m)

433 MHz

650 MHz

920 MHz

1800 MHz

2200 MHz

Road 1.5 3.54 3.47 3.36 Road 0.5 3.81 3.80 3.78 3.71 3.65 Field 1.5 3.36 3.24 3.13 Field 0.5 3.71 3.80 3.68 3.59 3.53

VI. VARIATIONS OF PROPAGATION LOSS WITH WEATHER The near ground RF propagation channel is affected by

local atmospheric effects very close to the ground which are

61

characterised by major changes of meteorological elements with height such as the vertical gradients of wind speed, air temperature, and air humidity. These gradients originate from surface phenomena driven by the interactions between soil, vegetation, air and solar radiation. Of the three atmospheric variables that influence atmospheric refraction (temperature, moisture, and pressure), moisture – or more specifically, water vapour – has the greatest effect on refraction. Temperature has the next greatest effect on refraction, followed by pressure.

An empirical formula for the refractive index of air [16] is

mr c

cTT

Pn =+⎟⎠⎞

⎜⎝⎛×+×−⎟

⎠⎞

⎜⎝⎛×= −−− 1107305.1106.5106.77 386 ρρ (14)

where P is the atmospheric pressure in millibars, T is the temperature in Kelvin, ρ is the water vapour concentration (g/cm3), c is the speed of light in a vacuum and cm is the speed of light in the medium.

From this equation a simple rule of thumb can be derived that the refraction increases with the moisture content but decreases with the temperature, but atmospheric pressure variations alone provide no significant change. Normal refraction occurs under normal (standard) atmospheric conditions in which moisture, temperature, and pressure all decrease with altitude. Normal refractive conditions are found in areas with very weak (or no) temperature inversions, high humidity, moderate to strong winds, and very unstable, well-mixed conditions. The area with normal refraction can be characterised by frequent showers, and distinct cloud elements (cumulus/cumulonimbus, open convective cells, wave clouds, streaks, or convective cloud lines). Abnormal refractive conditions occur in hot, dry areas (temperature > 30°C, relative humidity < 40%) where solar heating produces a homogenous surface layer, sometimes up to heights of hundreds of metres. Sub-refractive areas are also formed by warm, moist air moving over a cooler, drier surface, and near warm fronts because of warmer temperatures and an influx of moisture [17].

The effects of weather conditions on RF propagation loss were measured 3rd July 2012 using a fixed receive and transmit antenna separation distance of 100 m with fixed antenna heights of 1.5 m and 0.9 m, respectively. RF propagation loss measurements were performed on a bitumen road on a dewy morning with test frequencies 100 MHz, 160 MHz, 316 MHz and 751 MHz. Results are shown in Figure 7. Measurements started at 09:00a.m. with the temperature near the surface well below 5°C with dew lying on the ground. The temperature eventually rose to 17°C.

Temporal changes in RF propagation loss were also measured on 5th December 2012 on a grass field, again using a fixed receive and transmit antenna separation distance of 100 m and receive and transmit antenna heights of 1.5 m and 0.9 m respectively, for a wide range of frequencies. On this day the sun’s radiation was blocked by blanket cloud cover all morning until 12:30 p.m. when it was swiftly blown away allowing the sun’s radiation to heat the earth and transit from one stable thermal state to another. The wind direction was 300º north, blowing with a steady average speed of 3 m/s varying within the range of ±1 m/s. Atmospheric pressure was 1177.5 mbar slowly decreasing to about 1116 mbar by

3:00 p.m. and then increasing again. The measured RF propagation loss, solar radiance, air temperature and relative humidity for 5th December as a function of time are shown in Figure 8.

Fig. 7. Temporal RF propagation loss on a grassy field at 100 m for test frequencies 100 MHz, 160 MHz, 316 MHz and 751 MHz, with htx=0.3 m and hrx = 0.9 m and 1.5 m on 3 July 2012.

In the morning the measured solar radiance was 662 W/m2 and steadily increased from 12:30 p.m. until 1:25 p.m. to about 1017 W/m2 as the cloud was blown away. Similarly, the air temperature was affected by the sun’s radiation and increased from 18.3ºC to reach a maximum of 19.9ºC at 1:30 p.m. then later very slowly decreased to 19.6ºC. The relative humidity was also affected by the sun and decreased from an initial 50.5% reaching a minimum of 41.7% at 1:30p.m. followed by a slow increase to 47.0%.

Figure 9 shows the RF propagation losses measured for various frequencies on the 5th December as a function of time. Before noon propagation loss was steady. It then increased to a maximum around 1:20 p.m. before decreasing again until 14:20. After 2:20 p.m. the RF propagation losses were the same as before noon. The RF propagation loss varied between 2 to 5 dB during this transition state of the near earth’s atmosphere. The sudden change of cloud cover rapidly increased the ground temperature between 12:30p.m. to 1:30 a.m.. Given that the air temperature and humidity measurements were made at a height of 2 metres and the humidity slowly decreased with time and temperature slowly rose with time, it is surmised a duct near the surface was formed and grew in depth. This duct had a refraction index 9.7% smaller than when it was overcast. However modelling these effects requires additional measurements that were not conducted during our tests. For modelling the effect of weather on the RF propagation, not only the current weather state but also the past states have to be included.

62

Fig. 8. Temporal average solar radiance, air temperature and relative humidity for 5th December 2012.

Fig. 9. Temporal RF propagation loss for 5th December 2012.

VII. CONCLUSION RF propagation loss measurements for UHF frequencies,

using antennas at heights 0.3 to 1.5 m and separation distances less than 100 m have been made for a bitumen road and a dry grassy field. The type of surface had a small impact on the propagation loss for measurements of the same antenna heights. Only the 433 MHz measurements with the receive antenna at height 0.5 m had the Fresnel ellipsoid slightly obstructed by the ground. Two types of model were used to fit the experimental data; the two ray propagation and the simplified mean path loss propagation models. The experimental results are in excellent agreement with the two ray model with the maximum error of 6.1 dB not including knife edge diffraction loss.

Temporal change in the RF propagation loss was also measured. The experimental results suggest that dew on the ground can reduce the propagation loss by about 10 dB, while a sudden increase of solar radiance (changing from overcast to

clear sky) can have a significant impact on the propagation loss by 2-5 dB. These measurements have shown that the recent meteorological history of a site has an effect on the propagation loss. Further investigation of the diurnal and seasonal changes of RF propagation loss will be conducted to explore in detail various weather effects.

REFERENCES [1] Seybold J.S., “Introduction to RF Propagation”, John Wiley and Sons

2005 [2] Wyne, S.,Santos T., Singh A.P., Tufvesson F., Molisch A.F.,

“Characterisation of a time–variant wireless propagation channel for outdoor short-range sensor networks”, IET communications 2010, Vol4, Iss 3, pp253-264

[3] Tahmoush D; Silvious J; Burke, E, “A radar unattended ground sensor with micro-Doppler capabilites for false alarm reduction”, SPIE Proceedings, Vol 7833, Unmanned Systems Technologies II

[4] Feuerstein M.J., Blackard K.L., Rappaport T.S., Seidel S.Y. and Xia H.H., “Path loss, Delay Spread and Outage Models as Functions of Antenna Height for Microcellular System Design”, IEEE Transactions on Vehicular technology 43 (3) August 1994, 487-497.

[5] Xia H.H., Bertoni, H.L., Maciel L.R., Lindsay-Lewis A. and Rowe R., “Radio Propagation Characteristics for Line of Sight Microcellular and Personal Communications”, IEEE Transactions on Antennas and Propagation 41 (10) Oct 1993,1439-1447

[6] Laskarzewski Z., Brachman A., “Measurement based model of wireless propagation for short range transmission”, Proceedings of HET-NETs conference, 405-418

[7] Nicholas DeMinco, “Propagation Loss Prediction Considerations for Close-In Distances and Low-Antenna Height Applications”, NTIA Report TR-07-449

[8] T. S. Rappaport, “Characterization of UHF multipath radio channels in factory buildings”, IEEE Trans. Antenn. Propagation. vol. 37, pp.1058-1069, Aug. 1989.

[9] Merrill, W., Liu, H., Leong, J., Sohrabi, K., and Pottie, G., “Quantifying Short-Range Surface-to-Surface Communications Links”, IEEE Antennas and Propagation, Magazine, 46, 3, June 2004, pp. 36–46.

[10] H. Masui, K. Takahashi, S. Takahashi, K. Kage, and T. Kohayashi, “Difference of Path-Loss Characteristics Due to Mobile Antenna Heights in Microwave Urban Propagation”, IEICE Trans. Fundamentals, E82-A, 7, July 1999, 1144-1150.

[11] Patwari, G. D. Durgin, T. S. Rappaport, and R. J. Boyle,” Peer to-Peer Low Antenna Outdoor Radio Wave Propagation at 1.8 GHz,” 49th IEEE Vehicular Technology Conference, Houston, TX, May, 1999.

[12] K. Sohrabi, B. Manriquez, and G. J. Pottie, “Near Ground Wideband Measurement in 800-1000 MHz”, 49th IEEE Vehicular Technology Conference, Houston, TX, May, 1999.

[13] Egli John J. (Oct 1957) “Radio Propagation above 40 MC over Irregular Terrain”, Proceedings of the IRE (IEEE) 45 (10),1383-1391.

[14] Ogilvy, J.A., Theory of wave scattering from random rough surfaces, IOP Publishing 1991, 2

[15] Ament, W., “Towards a theory of reflection by a rough surface”, IRE Proc., Vol. 41, 142-146, 1953.

[16] Wert R.A., Goroch A.K., “Near-Earth Radio Frequency Propagation, Electronics and Electromagnetics”, 2007 NRL REVIEWv p153-155.

[17] ITU-R Recommendations Attenuation due to cloud and fog, ITU-R 840-3 Geneva, 1999.

[18] Anderson H. R., Fixed Broadband Wireless System Design, Wiley, 2003.

ACKNOWLEDGMENT We would like to thank Peter Wilinski and Barbara

Szumylo for assisting in field measurements and DSTO EWRD for supporting this program of work.

63