Availability of Free Space Optics (FSO)
and hybrid FSO/RF systems
Isaac I. Kim and Eric Korevaar
Optical Access Incorporated
10343 Roselle Street
San Diego, CA 92121
ABSTRACT
Free Space Optics (FSO) has become a viable, high-bandwidth wireless alternative to fiber optic cabling. The primary
advantages of FSO over fiber are its rapid deployment time and significant cost savings. The disadvantage of FSO over
fiber is that laser power attenuation through the atmosphere is variable and difficult to predict, since it is weather
dependent. This factor limits the distance at which FSO should be deployed. Using historical weather data collected at
airports, the link availability as a function of distance can be predicted for any FSO system. These availability curves
provide a good indication of the reasonable link distances for FSO systems in a particular geographical area. FSO link
distances can vary greatly from desert areas like Las Vegas to heavy-fog cities like St. Johns NF. Another factor in
determining FSO distance limitations is the link availability expectation of the application. For enterprise applications, link
availability requirements are generally greater than 99%. This allows for longer FSO link ranges, based on the availability
curves. The enterprise market is where the majority of FSO systems have been deployed. The carriers and ISPs are another
potential large user of FSO systems, especially for “last-mile” metro access applications. If FSO systems are to be used in
telecommunication applications, they will need to meet much higher availability requirements. Carrier-class availability is
generally considered to be 99.999% (“5 nines”). An analysis of link budgets and visibility-limiting weather conditions
indicates that to meet carrier-class availability, FSO links should normally be less than 140 m (there are cities like Phoenix
and Las Vegas where this 99.999% distance limitation increases significantly). This calculation is based on a 53 dB link
budget. This concept is extended to the “best possible” FSO system, which would have a 10 W transmitter and a photo-
counting detector with a sensitivity of 1 nW. This FSO system would have a 100 dB link margin, which would only
increase the 99.999% link distance to 286 m. A more practical solution to extending the high availability range would be to
back-up the FSO link with a lower data rate radio frequency (RF) link. This hybrid FSO/RF system would extend the
99.999% link range to longer distances and open up a much larger metro/access market to the carriers. It is important to
realize that as the link range increases, there will be a slight decrease in overall bandwidth. To show the geographical
dependence of FSO performance, the first map of FSO availabilities contoured over North America is presented. This map
is the first step to developing an attenuation map for predicting FSO performance, which could be used in a similar fashion
to the International Telecommunication Union (ITU)/Crane maps for predicting microwave performance.
Keywords: free space optics, FSO, laser communication, lasercom, optical wireless, infrared, atmospheric attenuation,
visibility, availability, link distance, link range, enterprise, carrier, internet service provider, ISP, telecommunications, last
mile, radio frequency, RF, microwave, International Telecommunication Union, ITU, Crane, bandwidth
1. INTRODUCTION
Free Space Optics (FSO) is gaining market acceptance as a functional, wireless, high-bandwidth access tool.
Fiber-optic cabling is still the preferred media for long haul, high-bandwidth transport. However, because of FSO’s lower
cost and significantly shorter installation time, FSO is now considered a viable option to fiber for short-haul access
distances of 4 km or less. As the awareness of FSO technology increases, and as the installation base of FSO links
increases, FSO is overcoming the early market barriers that faced this new access technology.
_____________________________________
Further author information
I.I.K. (correspondence): Email: ikim@opticalaccess.com; Web: http://www.opticalaccess.com;
Telephone: 858-792-8501; Fax: 858-792-8503
Fiber Optic cable
Multimode: lose 2 to 3 dB/km
Singlemode: lose .5 to .2 dB/km
Free-space Optics
Clear: lose .5 dB/km
Haze: lose 3 dB/km
Fog: 30 to 350 dB/km
Figure 1 Laser power through the atmosphere attenuates exponentially, in a similar fashion to
attenuation in fiber optic cable. Current multimode fiber attenuates at 2 to 3 dB/km, and singlemode fiber
attenuates at 0.5 to 0.2 dB/km. In clear conditions, the atmosphere attenuates at .5 dB/km (similar to
singlemode fiber), and in haze the attenuation is 3 dB/km (similar to multimode fiber). However, in fog
or heavy snow, the attenuation can increase to 30 to 350 dB/km. For longer link distances, these large
attenuation values can reduce FSO link availability. More detail about atmospheric attenuation in
different types of weather and precipitation can be found in Table 1.
One barrier that still exists in the acceptance of FSO is the concern of the effect of weather on the up time of an
FSO link. Figure 1 shows the attenuation of laser power through the atmosphere as compared to fiber. The most
significant difference between FSO and fiber optic transmission is the unpredictability of laser power attenuation in the
atmosphere. Fiber optic cable attenuates at a constant, predictable rate. Current multimode fiber attenuates at 2 to 3
dB/km, and singlemode fiber attenuates at 0.2 to 0.5 dB/km. On the other hand, atmospheric attenuation of laser power is
quite variable and difficult to predict, since it depends on the weather. Atmospheric attenuation can vary from 0.2 dB/km in
exceptionally clear weather to 350 dB/km in very dense fog. These large attenuation values in dense fog and heavy snow
can potentially reduce the up-time or link availability of FSO systems. The amount of potential down-time due to
visibility-limiting weather can be estimated from the link budget of the FSO system, and historical visibility data from
airports.
1
A sample availability calculation as a function of link range for a short-range (System A) and a long-range
(System B) FSO system are shown in Figure 2. These availability curves are calculated from historical visibility data from
the San Francisco International Airport.
These availability estimates are very useful in aiding potential users during the evaluation of FSO as an access
technology for a particular location and link range. Users have more confidence in deploying FSO if presented with
knowledge beforehand that the link availability meets or exceeds their requirements. Availability requirements generally
depend on whether the FSO system will be deployed in an enterprise or carrier network. Carrier-class availability
requirements of 99.999% (or “5 nines”) are generally much greater than enterprise-class availabilities of greater than 99%
(although this is not always the case). It is immediately apparent from the availability curves in Figure 2 how the distance
limitations of FSO systems depend on the user’s required availability. To meet carrier-class availability, FSO is limited to
short link distances. Depending on the geographical location of the FSO link, enterprise-class availabilities can extend the
possible FSO link ranges to much longer distances.
This paper examines the distance limitations of FSO systems for both carrier and enterprise applications. It will be
shown that carrier-class availability is achievable for much longer link distances if the FSO link is combined with a radio
frequency (RF) backup. Since FSO distances depend heavily on the geographic location of the link, the first map of FSO
link distances contoured over North America will be shown. This map will eventually evolve into an attenuation map for
predicting FSO performance in a manner similar to the Crane/ITU maps for microwave performance. This more scientific
and quantitative approach of describing the effects of weather on FSO links has provided users with a higher level of
knowledge and awareness of the capabilities of FSO systems. As long as FSO is used within its capabilities, it will provide
a sound and reliable “pipe” for users requiring high-bandwidth access.
FSO Availability vs. Link Range
95
96
97
98
99
100
0 1000 2000 3000 4000
Link Range (m)
Availability (%)
System A
System B
Figure 2 FSO estimates of availability for two example FSO Systems. These availability as a function
of link range curves were calculated from the visibility verses link range curves (see Figure 3) and
historical visibility data that is available for most airports worldwide.
1, 4
2. ENTERPRISE-CLASS FSO LINK RANGES
The following example of a typical enterprise FSO application best demonstrates the value of these availability
calculations. An enterprise user has just acquired a second office building near to their current building in the San
Francisco area. Both buildings are wired to run Fast Ethernet (100 Mbps) between their computers. However, all of the
servers are in the original building and the only conventional data access to the new building is a slow 1.5 Mbps T-1 line,
which is leased on a monthly basis. The user is considering deploying an FSO system, which will provide 100 Mbps data
access to the new building. The user is looking at two FSO systems: System A, which has a maximum range of 1100 m,
and System B, with a maximum range of 4000 m.
Figure 2 shows availability estimates as a function of link range for the two FSO products. These availability
curves are based on historical visibility data from the San Francisco International Airport. If the user’s two buildings are
separated by 400 m, both FSO System A and System B will provide approximately the same availability of 99.8%. Since
there is no performance advantage at this link range, the user would most likely choose the less costly of the two systems
(System A). If the link range is extended to 900 m, there begins to be a performance difference between the two systems.
System A’s availability drops to 99.3%, while System B’s availability is 99.6%. Even though the maximum range of
System A is quoted at 1100 m, at ranges close to the maximum specified range, it is wiser to choose the next longer-range
system (System B).
If the link range were extended to 2 km, System B would be the logical choice. The estimated availability of
99.3% at 2 km will still meet most enterprise-class requirements. From our experience, most enterprise data applications
are satisfied with availabilities of greater than 99%. However, if mission-critical data or voice is transmitted through the
FSO link, the availability requirement can increase to the carrier-class levels of 99.999%. Carrier-class link distance
calculations will be discussed in the next section.
These availability charts provide much more information than only the maximum specified range of the FSO
systems, and can assist the FSO user in making a more informed decision. These availability calculations are estimates
based on the power link budgets of the FSO systems, and the historical visibility data from airports.
1
Since the link budgets
used are conservative, typically the actual availability observed in the field exceeds the predicted availability. Therefore the
estimated availability can be considered a minimum expected availability. Quantitative studies comparing actual FSO
availabilities to estimated FSO availabilities will be published in the near future.
Visibility
50 m
200 m
500 m
1 km
1.9 km
2 km
2.8 km
10 km
4 km
20 km
23 km
50 km
Weather
condition
Dense fog
Thick fog
Moderate fog
Light fog
0 m
Thin fog
Haze
Light
Haze
Clear
Very
Clear
Precipitation
Heavy rain
Medium rain
mm/hr
770 m
Cloudburst
100
25
12.5
18.1 km
Drizzle
0.25
5.9 km
Light rain
2.5
Snow
FSO
System A
Range
330 m
610 m
880 m
1.02 km
1.08 km
1.13 km
1.26 km
1.26 km
1.03 km
1.28 km
1.22 km
780 m
1.25 km
1.18 km
110 m
FSO
System B
Range
7.42 km
7.23 km
8.22 km
440 m
890 m
1.67 km
2.59km
3.30 km
4.07 km
2.68 km
6.08 km
1.32 km
7.08 km
5.04 km
140m
0 m 0 m
dB/km
Loss
(785 nm)
-84.9
-34.0
-14.2
-7.1
-4.6
-3.0
-0.53
-0.46
-6.7
-0.21
-1.1
-20.0
-0.6
-1.8
-339.6
Table 1 International Visibility Code weather conditions
2
and precipitation
3
along with their visibility,
dB/km loss at 785 nm,
6
and effective FSO link range for Systems A and B.
Another important consideration is the distance from the FSO deployment to the airport where the visibility data
was collected. Because of microclimates, the weather can change in as little as one-half mile from the airport. Since the
weather in downtown San Francisco can be different than at the airport, the FSO availability downtown will be slightly
different than what is predicted from airport data. However, the San Francisco airport data still gives a pretty good
approximation to the availabilities in the general San Francisco area. This question of proximity to the data-collecting
airport will be addressed in more detail in the section discussing the map of FSO availability.
One final comment needs to be made about the maximum range of FSO products, which, without proper
explanation, can be a misleading specification (see Figure 3). The maximum range of an FSO system is a direct function of
the visibility or weather (see Table 1 and Figure 3).
1
Therefore, any maximum range of an FSO system needs to specified
to a particular weather condition or more specifically a dB/km rate of attenuation. In this example, hazy weather with 4 km
visibility (which corresponds to 3 dB/km attenuation) is the standard for the maximum range specifications of 1100 m for
System A and 4000 m for System B (see solid arrows in Figure 3).
How the maximum range can be misleading is demonstrated in Figure 3. System C is from another FSO vendor.
This vendor uses clear weather with 10 km visibility (which corresponds to 1 dB/km attenuation) as their maximum range
weather standard. Given this clear weather standard, the maximum range of System C is 4600 m (see dashed arrow in
Figure 3) and appears to go further than the 4000 m maximum range of System B (which uses the hazy weather standard).
However, it is obvious from Figure 3 that System B extends to longer ranges than System C in all visibility conditions.
When comparing FSO system maximum ranges, the user needs to make sure that the visibility standards used to determine
the maximum range of all FSO systems are the same. This will result in an equitable comparison between FSO systems.
Better yet, examining the FSO availabilities gives a much better indication of FSO performance than only comparing the
specified maximum range.
FSO Link Range vs Visibility
0
1
2
3
4
5
6
7
8
9
10
0 1000 2000 3000 4000 5000 6000
Link Range (m)
Visibility (km)
System A
System B
System C
Figure 3 Link range as a function of visibility for FSO Systems A, B, and C. As the visibility decreases,
the effective link range decreases. Generally, for short ranges, the link range is approximately equal to or
greater than the visibility. The solid arrows show the maximum link range for System A (1100 m) and
System B (4000 m) using the 4 km visibility standard (3 dB/km attenuation). The dashed line is
maximum link range of System C (4600 m) using a 10 km visibility standard (1 dB/km). By changing
the visibility standard, the maximum range of System C appears to be greater than System B - even
though, compared to System C, System B has a longer link range in all visibility conditions.
V
isibility (miles) 0 >=1/4 >=5/16 >=1/2 >=5/8 >=.75 >=1 >=1.25 >=1.5 >=2 >=2.5 >=3 >=4 >=5 >=6 >=10
New York City, NY 100 99.7 99.6 99.5 99.1 99.1 98.7 98 97.7 96.4 94.3 92 88.3 83.3 77.5 55.2
Los Angeles, CA 100 99 98.6 98.4 98 97.9 97.2 96.3 95.7 93.7 90.9 88.6 83.3 76.3 69.7 49.3
Chicago, IL 100 99.7 99.5 99.4 98.9 98.9 98.6 97.7 97.6 96.4 94.4 93.4 90.6 85.5 80.6 59.4
Washington DC 100 99.8 99.6 99.6 99.4 99.3 99.1 98.7 98.4 97.6 96.3 94.9 91.3 87 82.3 61.8
San Francisco, CA 100 99.6 99.58 99.5 99.45 99.35 99.2 99 98.9 98.4 97.5 96.9 94.7 92.6 90.2 73.3
Philadephia, PA 100 99.4 99.2 99 98.6 98.6 98 97 96.7 95.3 93.2 90.8 85.7 80.3 74.2 50.5
Boston, MA 100 99.6 99 98.9 98.3 98.2 97.7 96.9 96.8 95.6 93.9 92.5 89.7 85.7 81.4 66.8
Detroit, MI 100 99.5 99.2 99.2 98.7 98.6 98 96.9 96.7 95 92.7 90.8 87.7 82.4 76.4 55.1
Dallas, TX 100 99.8 99.7 99.7 99.5 99.5 99.2 98.8 98.7 98.3 97.5 97.2 95.9 94.6 92.7 64.5
Toronto, ON 100 99.4 99.2 99.1 98.8 98.7 98.3 97.6 97.1 95.9 94.3 92.6 89.2 85.7 82 70.1
Houston, TX 100 99.3 98.9 98.9 98.5 98.4 98.1 97.4 97.3 96.5 95 94.2 92.1 89.2 85 59.2
Miami, FL 100 99.9 99.9 99.8 99.8 99.7 99.7 99.6 99.5 99.4 99.1 98.9 98.4 97.7 96.4 50.2
Phoenix, AZ 100 100 100 99.9 99.9 99.9 99.9 99.9 99.9 99.9 99.8 99.8 99.7 99.5 99.2 98.1
San Juan, PR 100 100 100 100 100 100 100 99.9 99.9 99.9 99.8 99.8 99.7 99.5 99.1 95.2
Las Vegas, NV 100 100 100 100 100 100 99.9 99.9 99.9 99.9 99.9 99.8 99.8 99.7 99.6 99.1
Honolulu, HI 100 100 100 100 100 100 100 100 100 99.9 99.9 99.8 99.7 99.6 99.4 98.6
Tucson, AZ 100 100 100 100 100 100 100 99.9 99.9 99.9 99.9 99.9 99.8 99.8 99.7 99.4
Table 2 Historical tabulated visibility data used to calculate FSO availability curves.
4
The data is a
percentage of the frequency of visibility (miles) from hourly observations and is available for most
airports in the world. The data for San Francisco was used for the availability curves in Figure 2. For
most cities (except for the last five) the resolution is not good enough to determine 99.999% availability.
