Free-Space Optics & high-speed RF for Next Generation

Networks – Propagation Measurements

Vikas Kukshya†, T.S. Rappaport7, H. Izadpanah†, G. Tangonan†, R.A. Guerrero>, J.K. Mendoza>, and Brey Lee>

† HRL Laboratories, LLC. 7 Wireless Networking and > Department of Physics

3011 Malibu Canyon Road, Communications Group Ateneo de Manila University
Malibu, CA 90265 433A ENS Building, Loyola Heights,
[email protected] U. of Texas, Austin, TX 78715 Quezon City, Philippines
Tel: (310) 317 5492 [email protected] [email protected]

Abstract – Due to consistently growing demand for high- 2. FSO - System and Channel Measurement
speed wireless data access, future communication systems will
necessarily have to enable real-time data access, high Atmospheric losses in a free space laser system are mainly
reliability as well as superior weather resilience. Hybrid due to free space loss, clear air absorption, scintillation,
scattering and refraction [1]. Beam divergence is the main
communication networks that integrate free space optics and
contributor to the free space loss, with typical beam
RF systems are uniquely positioned to satisfy all the divergence attenuation values of 20dB/km. Scintillation,
bandwidth, reliability, security and ‘all-weather functionality’ caused by random variations in refractive index along the
requirements. Unfortunately, the technology has been growing propagation path due to solar heating, can also cause the
at a pace that far exceeds our understanding of the effect of received signal amplitudes to fluctuate rapidly by as much as
wireless propagation channels on these hybrid systems. This 30dB [1]. Elaborate FSO link operational requirements and
budget analysis are presented in [5]. This section of the
paper attempts to present some of the preliminary propagation paper focuses only on attenuation by refraction and scattering
measurement results for typical pico-cell scenarios and their of a free-space laser while it propagates through water
impact on the design of hybrid communication systems. In droplets (rain). The point-to-point propagation measurements
addition to clear-sky and rain attenuation measurements, the were performed in the Metro Manila, Philippines during
paper also presents frequency diversity measurements for moderate to heavy rain events.
fixed RF systems at 38GHz and 60GHz bands of frequencies. 2.1 System parameters
1. Introduction A simplistic schematic of the experimental setup is shown in
Figure 1. The semiconductor laser used in the experiments
The multibillion-dollar optical-fiber backbones for high-
operates at 830nm and has a maximum output power rating
performance data access, management, and Internet based
of 40mW. Collimation is performed with an objective lens
multimedia services have been around for quite some time
(0.25 NA) mounted on a micrometer stage and the detector is
now. While the service providers have been immensely
a high-speed silicon photodiode (185-1100 nm) with peak
successful at developing robust, high capacity networking
sensitivity at 850 nm. The FSO propagation measurements
backbones, a cost-effective solution to extend the high speed
were performed during clear-sky (for calibration and
data access to individual subscribers has not yet materialized.
reference path loss statistics) as well as rain conditions (for
Free-space optics (FSO) is an emerging, cost-
rain attenuation measurements). The weather statistics and
effective, and rapidly deployable last-mile wireless access
rain rates were collected from the POST-PARTNERS satellite
technology that can offer true broadband (data rates up to
station at the Ateneo de Manila for higher reliability and
tens of Gbps) to end-users in densely populated pico-cell
accurate system performance characterization.
environments. Unfortunately, the technology has been
growing at a pace that far exceeds our general knowledge of Mirror
FSO propagation characteristics in clear-sky and other
weather conditions. Adverse weather conditions such as mist, FREE SPACE
fog and rain are generally acknowledged to severely affect Collimated Detector
the performance of free-space laser links [1] and Beam
consequently, have prevented wide-scale acceptance of the
technology. Innovative hybrid architectures that exploit the
complementary nature of weather sensitivities of FSO and RF Lens
technologies are necessarily the only solution to extremely
high-speed and weather-resilient communication networks.
The significance, global applications, development and
performance evaluation of such hybrid communication Laser Diode
architectures and systems are presented in [2], [3] and [4].

This paper attempts to qualitatively and quantitatively Figure 1 Simplistic Schematic for the FSO measurements
characterize the performance of hybrid wireless links,
specifically FSO and MMW (millimeter wave - 38 & 60 GHz 2.2 Propagation Measurement Results
frequency bands) during different weather events for pico-cell
scenarios. The wireless channel statistics, calculated from This section reports propagation measurements conducted
extensive wideband propagation measurements during during heavy rain events. Due to heavy rain conditions (up to
different weather events, will enable accurate channel 80 mm/h), the FSO propagation link length was limited to
modeling and robust designs for weather-resilient hybrid about 20m for these first set of propagation measurements.
communication systems.

0-7803-7467-3/02/$17.00 ©2002 IEEE. 616

The results derived from these reference measurements not information rates and therefore can serve as alternatives to
only validated the functionality of the experimental setup but FSO in the hybrid architecture. Wideband (800MHz)
also served as a reference for our future longer range clear- propagation measurements were conducted at three different
sky and rain measurements. The absolute rain attenuation cross-campus locations in Virginia Tech over a period of four
statistics (relative to clear-sky free space loss), recorded from months, and comprised of extensive path loss and rain
the optical wireless propagation experiments, are summarized attenuation measurements. In addition to vertical and
in Figure 2. In the Figure, two sets of data points are plotted horizontal polarization measurements at 38GHz and 60GHz
– average rain attenuation and maximum rain attenuation for carrier frequencies, the measurement campaigns also
a particular rain rate. The average attenuation values involved frequency diversity propagation measurements at
correspond to the statistical mean of all the rain attenuation 38GHz and 60GHz bands of frequencies to investigate
values recorded for a given rain rate and the maximum achievable gains in a broadband wireless system capable of
attenuation corresponds to the maximum value out of all the frequency diversity [6].
recorded measurements for a particular rain rate.
3.1 System Parameters
9
The propagation measurements were conducted by using a
8 fully-automated wideband direct-sequence spread spectrum
7 (DSS) sliding correlator channel sounding system ([6], [7])
Attenuation (dB)

Maximum capable of operating in a wide range of frequencies (cellular,

6
PCS, MMDS, and LMDS). The channel sounder is currently at
5 the University of Texas.
4
For the 38 GHz measurement setup, a horn transmitting
3 antenna with a maximum gain of 19dB and half power beam
Average
2 widths of 45° and 6.5° (in azimuth and elevation respectively)
was used. The receiver antenna was a parabolic reflector with
1 a maximum gain of 39dB and half power beam widths of 1.5°
0 in both azimuth and elevation. For the 60 GHz setup, an
0 10 20 30 40 50 60 70 80 90 open-ended wave-guide with a maximum gain of 6.7dB and a
Rain Rate (mm/h) half power beam width of 90° in azimuth was used as the
transmitting antenna. The receiver antenna was a horn with a
Figure 2 Rain Attenuation vs. rain rate at 850nm maximum gain of 29dB and a half power beam width of 7°.

As can be seen from the figure above, at 79mm/h rain rate, In order to accurately model a wireless channel, one must
the average rain attenuation is 4.7dB and the maximum eliminate small-scale fading effects from the recorded
attenuation level is as high as 8.2dB. At about 30mm/h, the propagation data, before correct channel statistics can be
average rain attenuation is 3.8dB and as low as 1dB at calculated. Recording multiple snapshots of Power Delay
10mm/h. As a part of the experiment, individual fluctuations Profiles (PDPs) over a large period of time or over a local
in received signal strength, relative to the rain rate, were also area, and performing statistical average on the data,
recorded and a maximum of 284 spikes per second were facilitates accurate temporal characterization and removal of
observed at 79 mm/h. small-scale fading effects. In this measurement campaign,
These propagation measurement results assume the Sliding Correlator Channel Sounding system was
significance for two reasons – First, the performance of FSO configured to record multiple snapshots of Power Delay
links during moderate to heavy rain events have not been Profiles (PDPs) during each measurement sequence.
widely published in literature and second, - a general Specifically, the data acquisition software of the channel
perception among the FSO community is that rain does not sounder was configured to record 25 PDP snapshots back-to-
significantly affect the performance of FSO links. The back for every measurement sequence. The data post
statistics presented above clearly indicate that rain can processing software then performs temporal averaging of 25
severely affect the performance of laser links even over short PDP snapshots, eliminates small-scale signal fading effects,
link lengths and can be a limiting factor for the hybrid and calculates the desired statistics from the averaged PDP.
architecture in rain intensive regions of the world. For accurate localized recording of weather statistics, a rain
As the next step in accurate characterization of FSO links gauge and weather monitoring station was setup. The
during rain events, currently, there are some experiments monitoring station was used to record real-time rain rates,
being conducted at our facilities to record the rain attenuation atmospheric temperature, wind direction, wind velocity,
statistics over longer propagation links (500m). humidity and atmospheric pressure. It was directly connected
to a laptop computer to facilitate ease of configuration and
3. MMW - System Design and Channel Measurements automatic data recording for extended periods of time [6].

The high-speed MMW communication systems are crucial to 3.2 Measurement Plan
the design of future hybrid communication networks. The
complimentary functionality of millimeter wave (MMW), with The measurement locations and propagation links were
respect to optical wireless links in hybrid architectures chosen to emulate actual downtown pico-cell scenarios with
(parallel FSO and RF), facilitates extremely robust designs medium and high-rise buildings. The transmitter and receiver
and superior end-to-end connectivity statistics. channel sounding subsystems were placed significantly above
Elaborate propagation measurement campaigns were the ground level and highly directional antennas were used
planned and conducted at two important bands of frequencies for the measurements. The three point-to-point wireless links
– the 38GHz and the 60GHz band. These are the only license- chosen for propagation measurement were 531m, 418m and
free LMDS frequency bands (in U.S.) that can support Gbps 161m long corresponding to typical pico-cell scenarios. In
addition to simple path loss propagation measurements in

0-7803-7467-3/02/$17.00 ©2002 IEEE. 617

different prevalent weather conditions (rain, snow, fog), 2.5
frequency diversity propagation measurements were also
conducted at both 38GHz and 60GHz bands of frequencies. Measured Rain Attenuation
vs. CCIR/SAM Model
2

Rain Attenuation (dB)

1.5

*** Measured Values

SAM Model
0.5

Rain Rate (mm/hr)

0
Figure 3 Measurement Location (418m long) during snow 0 10 20 30 40 50 60
Figure 5 Horizontal Polarization at 38GHz (~ 1000 PDPs)

2.5
Measured Rain Attenuation
vs. CCIR/SAM Model
2

Rain Attenuation (dB)

1.5

*** Measured Values

SAM Model
0.5
Figure 4 Measurement Location (161m long) during rain

3.3 Rain Attenuation Results Rain Rate (mm/hr)

0
As many as 1300 PDPs were recorded at all the three 0 10 20 30 40 50 60
measurement locations during clear-sky conditions at 38GHz Figure 6 Vertical Polarization at 38GHz (~ 875 PDPs)
and 60GHz. Simple clear-sky propagation measurements at
38GHz yielded a path loss exponent value of 2.02 for 4.5
horizontal polarization and 2.04 for vertical polarization. Since Measured Rain Attenuation
all the links had a clear LOS, the experimental results are in 4 vs. CCIR/SAM Model
agreement with the expected path loss exponent values of
two (2). At 60GHz, the path loss exponent value calculated 3.5
Rain Attenuation (dB)

from the recorded power delay profiles is relatively higher at

2.12. This excess free space attenuation can be easily 3
explained by the oxygen absorption bands at 60GHz. The
oxygen present in the atmosphere tends to absorb 2.5
significantly more energy from 60GHz RF signals as compared
to signal at other frequencies.
2
For rain measurements at 38GHz and 60GHz, as many as
2000 PDPs were recorded for the measurement location 1.5 *** Measured Values
SAM Model
shown in Figure 4 over a path length of ~ 160m. The rain
attenuation (in excess of free space loss) statistics, calculated 1
from the recorded PDPs at 38GHz and 60GHz are presented in
figures above. At 38GHz, for horizontal polarization, rain 0.5
attenuation values range from 1.2dB to 2.4dB for rain rates ~ Rain Rate (mm/hr)
30-52 mm/h. For vertical polarization measurements, the 0
attenuation levels are 0.1dB-1.9dB for rain rates ~ 4- 0 10 20 30 40 50 60 70 80
46mm/h. Figure 7 Vertical Polarization at 60GHz (~ 625 PDPs)

0-7803-7467-3/02/$17.00 ©2002 IEEE. 618

These statistics were compared with two popular rain models conditions, frequency diversity does not seem to enhance the
– the Global Crane Rain Model and the CCIR Rain Attenuation system performance in a significant manner.
Model. The CCIR Rain Attenuation Model is also referred to as
the Simple Attenuation Model (SAM). It must be noted that Similar measurements were performed for the 60GHz
while the Crane Global Model has different attenuation models frequency band. As expected, for an unobstructed LOS radio
for horizontal and vertical polarization, the SAM model is link with extremely directional antennas, frequency diversity
independent of the signal polarization. The comparison of rain measurements did not exhibit any significant performance
attenuation statistics with the empirical models showed that enhancement. The difference between received signal power
the SAM model is a better fit for 38GHz pico-cell links. Figures levels at all carrier frequencies was 0.7dB or less.
5 and 6 show the comparison of measured rain attenuation

37.05 37.15 37.25 37.35 37.45 37.55 37.65 37.75 37.85 37.95 38.05 38.15 38.25 38.35 38.45
values with the CCIR/SAM model. The Global Crane model for
horizontal polarization tends to overestimate rain attenuation
at low and moderate rain levels by > 0.5dB for pico-cell radio
links. On the other hand, the Global Crane model for vertical
polarization tends to underestimate the attenuation values.

At 60GHz, as shown in Figure 7, the rain attenuation

increases significantly with rain rate even for a short link
length of ~ 160m. The rain attenuation is as high as 3.5dB
(relative to free space path loss) during heavy rain events
with rain rates on the order of 60 mm/hr. The highest
recorded rain attenuation value is 3.90dB for a rain rate of
76.2 mm/hr. Comparison of measured rain attenuation values

Propagation Frequency (in GHz)

Location 1 (531m LOS radio link)
with values predicted by Crane and SAM Models indicates that
both models propose rain attenuation values that are
appreciably close (less than 1dB) to the actual measured
values. However, the SAM/ CCIR Model serves as a better fit
for the values as compared to the Global Crane Model. The
measured rain attenuation values are almost in complete
agreement with the predicted rain attenuation values of the
SAM Model. The Crane Model, on the other hand, tends to

Location 2 (418m LOS radio link)

underestimate the rain attenuation values by about 0.5dB.

3.4 Frequency Diversity Measurements

Frequency Diversity propagation measurements were also

performed at 38GHz and 60GHz bands of frequencies. The
motivation behind these measurements was to find out
whether parallel broadband sub channels can exhibit
appreciable frequency diversity gain. The measurements were
performed over two LOS radio links (531m and 418m) during
clear-sky conditions and Power Delay Profile (PDP) data files
were recorded to capture path loss statistics at 24 different
frequencies. These carrier frequencies (12 for each band)
were separated by 100MHz intervals in the 38GHz and 60GHz
bands.

In order to validate and quantitatively characterize the

functioning of the channel sounder, extensive free-space
calibration measurements were performed at all the 24
Relative Path loss values (dB)
carrier frequencies of interest. By quantifying any non-
uniform gain behavior of the channel sounder at different
carrier frequencies, we were able to compensate for
44.5

43.5
45

44

frequency selectivity of the received signals.

The path loss values calculated from PDP data files
Figure 8 Frequency Diversity measurements at 38GHz band
recorded at each of the 12 different carrier frequencies in the
of frequencies for two LOS locations in clear sky conditions.
38GHz frequency band are presented in Figure 8. The path
The path loss values are relative to free-space calibration
loss statistics are compared with each other in order to
path loss over a 4m link.
evaluate the performance of a communication system that is
capable of frequency diversity. The relative path loss statistics
indicate that the received signal power is fairly constant at all 3.5 Measurements in Snow and Fog
12 carrier frequencies. The difference between received signal
power levels at all carrier frequencies was 0.5dB or less. This Propagation measurements were also conducted in light snow
is expected because the channel sounding hardware makes and moderate to dense for environments. Though there was a
use of highly directional antennas and these propagation slight loss of signal power (< 0.3dB) due to light snow events
measurements are performed over unobstructed LOS radio in individual PDP snapshots, but the excess attenuation
links. Consequently, all the reflected multipath components values calculated from averaged PDP were negligible. In the
are discounted at the receiver. In the absence of any case of moderate to heavy fog events, again, the excess
multipath component at the receiver, during clear sky attenuation was found to be negligible (< 0.1dB).

0-7803-7467-3/02/$17.00 ©2002 IEEE. 619

Conclusions

The paper presented several propagation measurements that

can be critical in the design of hybrid communication system
consisting of FSO and MMW high speed links. Contrary to the
common perception, the rain attenuation measurement
results indicate that moderate to heavy rain events can
severely limit the performance of FSO communication links.
The results also showed that the millimeter wave links
(38GHz and 60GHz) may operate satisfactorily even during
high rain conditions for pico-cell deployments. However, for
long range deployments in rain intensive regions of the world,
hybrid architecture communication systems must explore
lower RF frequencies (preferably below 10GHz) as backup for
the FSO links during rain events.

Acknowledgement

The authors would like to acknowledge the MPRG’s industrial

affiliates program and partial support of NGI contract
#N00014-99-C-0331 (DARPA) for the research project. The
authors from Ateneo de Manila University would like to
acknowledge the invaluable assistance of Mr. David Persechini
of HRL Laboratories with the experimental equipment setup,
and the support of POST-PARTNERS program sponsored by
the Ministry of POST, Japan.

References

[1] Heatley, D.J.T.; Wisely, D.R.; Neild, I.; Cochrane, P.,

“Optical wireless: the story so far”, IEEE Communications
Magazine, Volume: 36, Issue: 12, Dec. 1998, Page(s): 72 -
74, 79-82.

[2] Kukshya, V.; Izadpanah, H.; Dolezal, F., “Smart

Broadband Wireless Connectivity – Design Challenges and
Performance Evaluation”, Broadband Wireless Access Systems
Conference, World Wireless Congress 2002, May 02, San
Francisco, USA.

[3] Clark, G.; Willebrand, H.; Achour, M, “Hybrid free space

optical/microwave communication networks: A unique
solution for ultra high-speed local loop connectivity”,
LightPointe Communications, Inc. (www.lightpointe.com)

[4] Kim, I.; Korevaar, E., “Availability of Free Space Optics

and hybrid FSO/RF systems”, Optical Wireless
Communications IV, August 2001, Denver, USA.

[5] Dodley, J.P. et. al., “Free Space Optical Technology and
Distribution Architecture for Broadband Metro and Local
Services”, Optical Wireless Communications III, November
2000, Boston, USA.

[6] Kukshya, V.; Rappaport, T.S., “Wideband Propagation

Measurements and Channel Modeling at 38GHz and 60GHz
bands of frequencies for pico-cell scenarios”, Technical Report
and Masters Thesis, June 2001, MPRG, Virginia Tech, USA.

[7] Durgin, G.; Kukshya, V.; Rappaport, T.S., “Wideband

Measurements of Angle and Delay Dispersion for Outdoor and
Indoor Peer-to-Peer Radio Channels at 1920 MHz”, IEEE
Transactions on Antennas and Propagation, June 2002.

0-7803-7467-3/02/$17.00 ©2002 IEEE. 620