EXPERIMENTAL PERFORMANCE EVALUATION

OF A GBPS FSO LINK

A Case Study

J. A. R. Pacheco de Carvalho

1,2

, N. Marques

1,3

, H. Veiga

1,3

, C. F. Ribeiro Pacheco

U. de Detecção Remota,

Dept. de Física,

Centro de Informática, Universidade da Beira Interior

6201-001 Covilhã, Portugal

A. D. Reis

1,2,4

Dept. de Electrónica e Telecomunicações, Instituto de Telecomunicações

Universidade de Aveiro, 3810 Aveiro, Portugal

Keywords: Wireless Network, Laser, FSO, Point-to-Point Link, Performance Measurements.

Abstract: Wireless communications have been increasingly important. Besides Wi-Fi, FSO plays a very relevant

technological role in this context. Performance is essential, resulting in more reliable and efficient

communications. A FSO medium range link has been successfully implemented for high requirement

applications at Gbps. An experimental performance evaluation of this link has been carried out at OSI layers

1, 4 and 7, through a specifically planned field test arrangement. Several results are presented and discussed,

as obtained from simultaneous measurements of powers received by the laser heads, TCP, UDP and FTP

experiments, resulting in determinations of TCP throughput, jitter, percentage datagram loss and FTP

transfer rate. Conclusions are drawn about link performance.

1 INTRODUCTION

Wi-Fi and FSO are wireless communications

technologies whose importance and utilization have

been growing.

Wi-Fi uses microwaves in the 2.4 and 5 GHz

frequency bands and IEEE 802, 11a, b, g standards.

Nominal transfer rates up to 11 (802.11b) and 54

Mbps (802.11a, g) are specified (IEEE Std 802.11-

2007). It has ben used in ad hoc and infrastructure

modes. Point-to-point and point-to-multipoint

configurations are used both indoors and outdoors,

requiring specific directional and omnidirectional

antennas.

FSO uses laser technology to provide

point-to-point communications e.g. to interconnect

LANs of two buildings having line-of-site. FSO was

developed in the 1960’s for military and other

purposes, including high requirement applications.

At present, speeds typically up to 2.5 Gbps are

possible and ranges up to a few km, depending on

technology and atmospheric conditions. Interfaces

such as fast Ethernet and Gigabit Ethernet are used

to communicate with LAN’s. Typical laser

wavelengths of 785 nm, 850 nm and 1550 nm are

used. In a FSO link the transmitters deliver high

power light which, after travelling through

atmosphere, appears as low power light at the

receiver. The link margin of the connection

represents the amount of light received by a terminal

over the minimum value required to keep the link

active: (link margin)

= 10 log

(P/P

min

), where P

and P

min

are the corresponding power values,

respectively.

There are several factors related to performance

degradation in the design of a FSO link: distance

between optical emitters; line of sight; alignment of

optical emitters; stability of the mounting points;

atmospheric conditions; water vapour or hot air;

strong electromagnetic interference; wavelength of

the laser light (Rockwell & Mecherle, 2001). A

redundant microwave link is always essential, as the

laser link can fail under adverse conditions. Several

studies and implementations of FSO have been

reported (D’Amico, Leva & Micheli, 2003). FSO

has been used in hybrid systems for temporary

multimedia applications (Mandl et al., 2007).

123

A. R. Pacheco de Carvalho J., Marques N., Veiga H., F. Ribeiro Pacheco C. and D. Reis A. (2010).

EXPERIMENTAL PERFORMANCE EVALUATION OF A GBPS FSO LINK - A Case Study.

In Proceedings of the International Conference on Wireless Information Networks and Systems, pages 123-128

DOI: 10.5220/0002978601230128

 SciTePress

Performance has been a very important issue,

resulting in more reliable and efficient

communications. Telematic applications have

specific performance requirements, depending on

application. New telematic applications present

special sensitivities to performances, when

compared to traditional applications. E.g.

requirements have been quoted as: for video on

demand/moving images, 1-10 ms jitter and 1-10

Mbps throughput; for Hi Fi stereo audio, jitter less

than 1 ms and 0.1-1 Mbps throughputs (Monteiro e

Boavida, 2002).

Several performance measurements have been

made for 2.4 and 5 GHz Wi-Fi (Pacheco de

Carvalho et al., 2008a, 2009). FSO and fiber optics

have been applied at the University Campus to

improve communications quality (Pacheco de

Carvalho et al., 2007, 2008b, 2008c). In the present

work we have further investigated that FSO link for

performance evaluation at OSI layers 1, 4 and 7.

The rest of the paper is structured as follows:

Chapter 2 presents the experimental details i.e. the

measurement setup and procedure. Results and

discussion are presented in Chapter 3. Conclusions

are drawn in Chapter 4.

2 EXPERIMENTAL DETAILS

The main experimental details, for testing the quality

of the FSO link, are as follows.

A 1 Gbps full-duplex link was planned and

implemented, to interconnect the LAN at the Faculty

of Medicine building and the main University

network, to support medical imaging, VoIP, audio

and video traffics (Pacheco de Carvalho et al., 2007,

2008b). Then, a FSO laser link at 1 Gbps full-

duplex, over a distance of 1.14 km, was created to

interconnect the Faculty of Medicine (FM) building

at Pole III and the Sports (SB) building at Pole II of

the University (Figure 1).

We have chosen laser heads from FSONA

(Figure 2) to implement the laser link at a laser

wavelength of λ= 1550 nm for eye safety, where

allowable laser power is about fifty times higher at

1550 nm than at 800 nm (Rockwell & Mecherle,

2001). Each laser head comprised two independent

transmitters, for redundancy, and one wide aperture

receiver. Each laser had 140 mW of power, resulting

in an output power of 280 mW (24.5 dBm). 1000-

Base-LX links over OM3 50/125 μm fiber were used

to connect the laser heads to the LANs.

For a matter of redundancy a 802.16d WiMAX

point-to-point link at 5.4 GHz (IEEE Std 802.16-

2004) was available, where data rates up to either 75

Mbps or 108 Mbps were possible in normal mode or

in turbo mode, respectively (Alvarion, 2007). This

link was used as a backup link for FM-SB

communications, through configuration of two static

routing entries in the switching/routing equipment

(Pacheco de Carvalho et al., 2007).

Performance tests of the FSO link were made

under favourable weather conditions. During the

tests we used a data rate mode for the laser heads

which was compatible with Gigabit Ethernet. At OSI

layer 1 (physical layer), received powers were

simultaneously measured for both laser heads. Data

were collected from the internal logs of the laser

heads, using STC (SONAbeam Terminal Controller)

management software (FSONA, 2006). At OSI layer

4 (transport layer), measurements were made for

TCP connections and UDP communications using

Iperf software (NLANR, 2005), permitting network

performance results to be recorded. . Both TCP and

UDP are transport protocols. TCP is connection-

oriented. UDP is connectionless, as it sends data

without ever establishing a connection. For a TCP

connection over a link, TCP throughput was

obtained. For a UDP communication, we obtained

UDP throughput, jitter and percentage loss of

datagrams. TCP packets and UDP datagrams of

1470 bytes size were used. A window size of 8

kbytes and a buffer size of the same value were used

for TCP and UDP, respectively.

A specific field test arrangement was planned

and implemented for the measurements (Figure 3).

Two PC’s having IP addresses 192.168.0.2 and

192.168.0.1 were setup as the Iperf server and client,

respectively. The PCs were HP computers, with 3.0

GHz Pentium IV CPUs, running Windows XP. The

server had a better RAM configuration than the

client. They were both equipped with 1000Base-T

network adapters. Each PC was connected via

1000Base-T to a C2 Enterasys switch (Enterasys,

2008). Each switch had a 1000Base-LX interface.

Each interface was intended to establish a FSO link

through two laser heads, as represented in Figure 3.

The laser heads were located at Pole II and Pole III,

at the SB and FM buildings, respectively. The

experimental arrangement could be remotely

accessed through the FM LAN. In the UDP tests a

bandwidth parameter of 300 Mbps was used in the

Iperf client. Jitter, which represents the smooth mean

of differences between consecutive transit times,

was continuously computed by the server, as

specified by RTP in RFC 1889. RTP provides end-

to-end network transport functions appropriate for

applications transmitting real-time data, e.g. audio,

WINSYS 2010 - International Conference on Wireless Information Networks and Systems

124

video, over multicast or unicast network services. At

OSI layer 7 (application layer) the setup given in

Figure 3 was also used for measurements of FTP

transfer rates through FTP server and client

applications installed in the PCs. Each measurement

corresponded to a single FTP transfer, using a 2.71

Gbyte file. Whenever a measurement was made at

either OSI layer 4 or 7, data were simultaneously

collected at OSI layer 1. Batch command files were

written to enable the TCP, UDP and FTP tests. The

results, obtained in batch mode, were recorded as

data files in the client PC disk.

3 RESULTS AND DISCUSSION

Large amounts of data were collected and processed

by averaging over several time intervals. The

corresponding results are shown for TCP in Figure

4, for UDP in Figure 6 and FTP in Figure 8. The

average received powers for the SB and FM laser

heads, mostly ranged high values in the 25-35 µW

interval which corresponds to link margins of 4.9-

6.4 dB (considering P

min

=8 µW). From Figure 4 it

follows that TCP average throughput (313.6 Mbps)

is reasonably steady, although some small peaks

arise for throughput deviation. Figure 5 illustrates

details of TCP results over a small interval. Figure 6

shows that UDP average throughput (125.5 Mbps) is

fairly steady, having a small steady throughput

deviation. The jitter is small, usually less than 1 ms,

while percentage datagram loss is practically

negligible. Figure 7 illustrates details of UDP-jitter

results over a small interval. Figure 8 shows that

average FTP throughput (344.5 Mbps) is very

steady, having low throughput deviation. Figure 9

illustrates details of FTP results over a small

interval. Transfer rates of the PC’s disks are always

a limitation in this type of FTP experiments. In all

cases, having high values of average received

powers, the quantities under analysis did not show

on average significant variations even when the

received powers varied. The results here obtained

complement previous work by the authors (Pacheco

de Carvalho et al., 2007, 2008b, 2008c). Generally,

for our experimental conditions, the FSO link has

exhibited very good performances at OSI layers 4

and 7.

Besides the present results, it must be mentioned

that we have implemented a VoIP solution based on

Cisco Call Manager (Cisco, 2004). VoIP, with

G.711 and G729A coding algorithms, has been

working over the laser link without any performance

problems. Tools such as Cisco IP Communicator

have been used. Video and sound have also been

tested through the laser link, by using eyeBeam

Softphone CounterPath software (CounterPath,

2007). Applications using the link have been well-

behaved.

Figure 1: View of the 1.14 km laser link between Pole II

(SB) and Pole III (FM).

Figure 2: View of the laser heads at FM (Pole III) and SB

(Pole II).

EXPERIMENTAL PERFORMANCE EVALUATION OF A GBPS FSO LINK - A Case Study

125

Figure 3: Field tests setup scheme for the FSO link.

Figure 4: TCP results.

Figure 5: Details of TCP results.

Figure 6: UDP results; 300 Mbps bandwidth parameter.

Figure 7: Details of UDP-jitter results; 300 Mbps

bandwidth parameter.

Figure 8: FTP results.

WINSYS 2010 - International Conference on Wireless Information Networks and Systems

126

Figure 9: Details of FTP results.

4 CONCLUSIONS

An FSO laser link at 1 Gbps has been successfully

implemented over 1.14 km along the city, for

interconnecting Poles of the University and support

high requirement applications.

A field test arrangement was planned and

implemented for FSO performance measurements at

OSI layers 1, 4 and 7. At OSI layer 1, received

powers were simultaneously measured in both laser

heads. At OSI layer 4, TCP throughput, jitter and

percentage datagram loss were measured. Through

OSI layer 7, FTP transfer rate data were acquired.

Under favourable experimental weather conditions,

when the measurements were carried out, the link

has shown to be very well behaved, presenting very

good performances. Applications such as VoIP,

video and sound, have been well-behaved. Further

measurements are planned under several

experimental conditions, such as environmental and

multimedia traffic.

ACKNOWLEDGEMENTS

Supports from the University of Beira Interior and

FCT (Fundação para a Ciência e a

Tecnologia)/POCI2010 (Programa Operacional

Ciência e Inovação) are acknowledged. We

acknowledge Hewlett Packard and FSONA for their

availability.

REFERENCES

Alvarion 2007, BreezeNET B-Making Point to Point

Connections Fast and Easy (pp. 1-2, 213695 rev. i),

retrieved June 5, 2007, from http://www.alvarion.com

Cisco 2004, Cisco Call Manager Version 4.1 (pp. 1-18,

204121.30_ETMG_JM_10.04), retrieved February 5,

2007, from http://www.cisco.com

CounterPath 2007, eyeBeam 1.5 for Windows User Guide

(pp. 1-71), retrieved February 10, 2008, from http://

www.counterpath.com

D’Amico, M, Leva, A & Micheli, B 2003, ‘Free Space

Optics Communication Systems: First Results From a

Pilot Field Trial in the Surrounding Area of Milan,

Italy’, IEEE Microwave and Wireless Components

Letters, vol. 13, no. 8, August, pp. 305-307.

Enterasys 2008, Enterasys SecureStack C2-Hardware

Installation Guide (pp. 1-78, 9033951-06), retrieved

April 5, 2008, from http://www.enterasys.com

FSONA 2006, SONABEAM Terminal Controller Software

User’s Guide (pp. 1-98, 95-0282 Rev. H), retrieved

March 10, 2008, from http://www.alliedtelesis.com

IEEE Std 802.16-2004, IEEE Standard for Local and

metropolitan area networks-Part 16: Air Interface for

Fixed Broadband Wireless Access Systems, retrieved

September 15, 2008, from http://standards.

ieee.org/getieee802

IEEE Std 802.11-2007, IEEE Standard for Local and

metropolitan area networks-Specific Requirements-

Part 11: Wireless LAN Medium Access Control (MAC)

and Physical Layer (PHY) specifications, retrieved

October 10, 2007, from

http://standards.ieee.org/getieee802

Mandl, P, Chlestil, Ch, Zettl, K & Leitgeb, E 2007,

‘Hybrid systems using Optical Wireless, Fiber Optics

and WLAN for temporary multimedia applications’, in

Proceedings of the 9th International Conference on

Telecommunications-Contel 2007, Zagreb, Croatia,

pp. 73-76.

Monteiro, E & Boavida, F 2002, Engineering of

Informatics Networks, 4th edn, FCA-Editor of

Informatics Ld., Lisbon.

NLANR 2005, Iperf User Docs, retrieved February 5,

2007, from http://dast.nlanr.net

Pacheco de Carvalho, J A R, Gomes, P A J, Veiga, H &

Reis, A D 2007, ‘Wi-Fi and Very High Speed Optical

Links for Data and Voice Communications’, in

Proceedings of the 2ª Conferência Ibérica de Sistemas

e Tecnologias de Informação, Universidade Fernando

Pessoa, Porto, Portugal, pp. 441-452.

Pacheco de Carvalho, J A R, Gomes, P A J, Veiga, H &

Reis A D 2008a, ‘Development of a University

Networking Project’, in Encyclopedia of Networked

and Virtual Organizations, eds GD Putnik, MM

Cunha, IGI Global, Hershey, PA, pp. 409-422.

Pacheco de Carvalho, J A R, Gomes, P A J, Veiga, H &

Reis, A D 2008b, ‘Experimental Performance

Evaluation of a Very High Speed Free Space Optics

Link at the University of Beira Interior Campus: a

Case Study’, in Proceedings of SEONs 2008- VI

EXPERIMENTAL PERFORMANCE EVALUATION OF A GBPS FSO LINK - A Case Study

127

Symposium on Enabling Optical Network and Sensors,

Porto, Portugal, pp.131-132.

Pacheco de Carvalho, J A R, Veiga, H, Gomes, P A J,

Ribeiro Pacheco, C F F P & Reis, A D 2008c,

‘Experimental Performance Study of a Very High

Speed Free Space Optics Link at the University of

Beira Interior Campus: a Case Study’, in Proceedings

of the ISSPIT 2008-8th IEEE International Symposium

on Signal Processing and Information Technology,

Sarajevo, Bosnia and Herzegovina, pp. 154-157.

Pacheco de Carvalho, J A R, Veiga, H, Gomes, P A J,

Ribeiro Pacheco, C F F P, Marques, N & Reis, A D

2009, ‘Laboratory Performance of Wi-Fi Point-to-

Point Links: a Case Study’, in Proceedings of WTS

2009-8th Wireless Telecommunications Symposium,

Prague, Czech Republic, pp. 143-147.

RFC 1889, RTP: A Transport Protocol for Real-Time

Applications, retrieved January 15, 2009, from

http://www.rfc-archive.org

Rockwell, D A & Mecherle, G S 2001, ‘Wavelength

Selection for Optical Wireless Communication

Systems’, in Proceedings of SPIE, 4530, pp. 27-35.

WINSYS 2010 - International Conference on Wireless Information Networks and Systems

128