Qualification of Fiber Optic Networks

Qualification of fiber optic networks 1 Qualification of Fiber Optic Networks José Manuel dos Santos Duarte, Instituto Superior Técnico Abstract the theme Qualification of Fiber Optic Networks, presented as a project report, has the objective to describe the related testing activities held since the mid -90s until today, aiming to qualify the networks according to international and client specifications or just finding the operating state of a network in operation, detecting eventual problems and bottle necks, and its capacity to support high transmission rates (10 to 40 Gb/s). In an introductory way, we will go through optical fiber communications and pulse propagation on fiber optics, identifying the factors that determine it, and which will be the appropriate test method. The results of tests on various networks, new and recent or old ones, implemented with different types and generation of optical fiber and cables were presented and discussed. The testing methods and specifications carry on, were a suit of the International Standards to fiber optic testing. The work is concluded with discussion of the problem of testing of fiber optic networks and its benefits to the network owner or operator. Index Terms Chromatic dispersion, fiber optics, insertion loss, polarization mode dispersion, network qualification tests and spectral attenuation. I. INTRODUCTION HE costumers have been putting on Ttelecommunications services providers, increasing demands and challenges for the services rendered (accessibility everywhere, high speed, new kind and more services, quality assurance, and low prices). On the other hand, network owners/operators aims to increase the results of its business (increase share of the market and improve its profits, based on high quality, innovative services and highly controlled network maintenance costs and investments). The networks where the services are rendered are submitted to great demands, namely high availability, quality of service, accessibility and transmission capacity. Telecom service providers fight each other to capture customers using aggressive trade policies and innovative and differentiated services, where being the first at providing something new is regarded, by them, as a major competitive advantage. Fiber optic telecom networks assumed for years a leading role in the development and expansion of telecommunications, contributing to the objectives mentioned above were reached. The optical fiber was invented by Narinder S. Kapany in 1952 [1, 2]. In 1966 Dr. Charles Kao presented the requirements for use of fiber optics in telecommunications. Some years later (1970) a group of scientists from Corning Incorporated Company, composed by Donald Keck, Peter Schultz and Robert Mauer, produced the first optical fiber that meets the requirements presented by Dr. Charles Kao. Today fiber optics is the way to perform the communication between two points, with higher quality of service (QoS), higher bandwidth and ability to transport information at less cost than other solutions. In Portugal, the first fiber optic telecom networks appeared in the 80s. The transmission based on optical fibers, have several advantages over copper cable or other, namely the signal can be transmitted over long distances (200 km) without the need of signal regeneration; the transmission is virtually insensitive to electromagnetic disturbances; the capacity of an optical communications system is much higher. A fiber optic cable usually has a guarantee period, provided by cable manufacturer and installer, from 20 to 30 years, related with fiber characteristics. During this period, it is expected that physical and environmental phenomenon and external interventions affect network performance, resulting in a degradation of their initial characteristics. Therefor, it is important to follow the evolution of the network, identifying any existing or potential problems, as well as their capacity and capability to support the new technological development of transmission solutions and increased traffic demands. A rational approach points to the need for network verification / testing, as a way to ensure that it can meet the quality and operating conditions, accordingly to international standards. Usually operators have a supervision system to monitor basic link characteristics, fault detection and optimize routing, focusing attenuation issues, link length and path, and do it in a regular basis. Full test activities (e.g. spectral attenuation, dispersion and other issues) are usually contracted to test laboratories (3 rd part laboratories). The light propagation in a fiber optic is affected by phenomena like attenuation, dispersion and nonlinearities. The attenuation in the fiber is caused by intrinsic causes, including light absorption and scattering and proceeds along the length of the link, or extrinsic reasons, related to the aging process and installation thereof. In the installation process, bad practices associated with noncompliance of its mechanical specifications (e.g, traction, tension) or radius of curvature that

Qualification of fiber optic networks 2 a fiber have been subjected, can also affect the attenuation characteristics of the fiber, increasing it significantly. The attenuation depends on the wavelength of the propagated light and the type of fiber used and is directly proportional to the length. The dispersive phenomena cause delays in the arrival of the various components of the signal, which is a distortion factor and introduces limitations in terms of this transmission speed. Lower speeds mean higher transmission times and less useful bandwidth available. When the delay difference is manifested selectively, accordingly the spectral components of the light signal, it is said that we are facing the phenomena of chromatic dispersion (CD); when the difference in arrival times of signal components were related to the polarization of the modes, it is said we are dealing with phenomena of polarization mode dispersion (PMD). A qualification process of a fiber-optic network, regardless of the binary debt it is expected to endure, taking in consideration all the factors and in order to evaluate all attributes of the link, must be performed under international standard methods and procedures. The following tests set are recommended: Insertion loss and return loss (at fixed wavelengths); Attenuation coefficient (at fixed wavelengths); Spectral attenuation (range 1260 nm-1640 nm); Chromatic dispersion (range 1260 nm-1640 nm); Polarization mode dispersion (at fixed wavelengths). The test activities presented in this work, brought decisive contributions to ensure that, either in the case of new networks, or in the case of networks in the regular operation, defined specifications were met and installation faults were removed, or trouble sectors of the link that requires preventive or reconstructive maintenance actions have been identified. This work is thus based on the results of the tests performed (field and laboratory) and qualification activity of fiber optic networks carried out since the late 80s of last century until 2011. II. PROPAGATION IN FIBER OPTICS A communication system can be generically described by three blocks: the transmitter, the transmission channel and receiver. In the case of fiber optic communications will have at the extremes, the blocks of optical transmission and optical reception, and the role of transmission channel is performed by the optical fiber (Fig. 2.1). the designation of core; the outer cylindrical structure is called cladding; the protection layer or primary coating is called jacket Core Cladding Jacket Fig. 2. 2. Structure of an optical fiber The refractive index of the core is higher than the cladding. The way the refractive index changes, from the center to core periphery, as illustrated in Fig. 2.3, defines the fiber type: step index, if the core index remains unchanged and graded index fibers, if the core index decreases gradually, from center to periphery. Step index a n a b b d Graded index Fig. 3. Index profile changes in optical fibers The light propagation in an optical fiber can be described by the optic geometric approach, as represented in Fig. 2.4. (n 0, n 1 and n 2, are respectively refractive index of air, fiber core and fiber cladding; i is the incident angle at fiber surface; r is the refractive angle at fiber surface; is the reflection angle at core-cladding interface.) a a n b b d Electrical signal Optical Tx. Optical fiber Optical signal Optical Rx Electrical signal Fig. 4. Light propagation in fiber, at air-core and corecladding interfaces, to a step-index fiber Fig. 1. Optical communication system block diagram The optical fiber, as showed in Fig. 2.2, is formed by a double concentric cylinder structure, protected externally by a protective layer structure. The inner cylindrical structure has As it can be seen, to each i corresponds one travel path, so in fiber core we will have multiple path. Multiple path result in pulse spread and consequently lower available bandwidth. This kind of fibers is named multimode fibers. This behavior wasn t in favor of telecommunication needs, and later on a new kind of fiber was developed, where only one beam path

Qualification of fiber optic networks 3 was enable. This was possible by substantially reducing the core diameter and manipulation of the refractive index of core and cladding material. This kind of fiber was designated as single-mode fiber and one of its most important characteristic is the Numerical Aperture (NA), which represents the maximum angle that the beam can have at the fiber interface, which assures that only one ray is propagated along core fiber (fiber axis), and is expressed by [1, 2] NA n0 sen imax n1 2 n2 2 ½. (1) Aceptance cone i max Fig. 5. Fiber optic acceptance cone illustration The use of geometrical optics approach to explain the guiding mechanism or light propagation in optical fiber has advantages. It enables a simplified understanding, but it has also severe limitations, because it is only valid when the fiber core radius (a) it is much larger than operating wavelength (). This is the case of multimode fibers, where a is in the range of 25-100 m and around 0,85 m; but not for single-mode fibers, where a is about 9 m and in the range of 1,310-1,625 m. As single-mode fibers are the most used and the only one in telecom sector, and this work is related to single-mode fiber testing in telecom industry, in the discussion of the propagation of light we must consider an optical fiber as a waveguide, so the wave theory approach must be used. The wave theory in conjugation with Maxwell s equations, explains how optical signals propagate through fiber, and it will help us gain an understanding of the phenomena that are important in the design of optic communication systems. Applying Maxwell s equations and assuming that the core and cladding regions of the silica are locally responsive, isotropic, linear, homogeneous and lossless, the wave equations for E (or H) are [1, 2] 0, (2) where is the Laplacian operator, E represents the electric field vector in frequency domain, is the signal frequency, n() is the index of refraction (frequency dependent) and c is the light velocity in vacuum. Taking in consideration the cylindrical structure of an optical fiber, as depicted in Fig. 2.6, assuming k 0 =, and expressing electrical field vector in cylindrical coordinates we have [1, 2] 0. (3) n2 n1 x Direction of wave propagation Fig. 6. Fiber optic cylindrical coordinates The electric and magnetic field vectors in the core and in the cladding, must satisfy the wave equation (2) expressed with respect to E or H, but the solutions are not independent, because they are related at boundary conditions at corecladding interface. Every pair of solutions of these wave equations that satisfies boundary conditions is a fiber mode. So, in general, and unless some construction rules are imposed, the light propagation in fiber is done with more than one mode. Assuming the direction of light propagation is z (Fig. 6), and that fiber properties (a, n 1 and n 2 ) are independent of z, the z-dependence of E or H for each fiber mode is of the form e iz, where is called propagation constant of the mode and determines the speed at which pulse in a mode is propagated in the fiber. The propagation constant =n/c = 2n/. Introducing the term k (wave number) as 2, results =kn. Thus for a wave propagation in core, =kn 1, and in cladding, =kn 2. Considering that fiber modes propagates partly in core and in cladding, their propagation constants must satisfy kn 2 < <kn 1. A mode ceases to be guided when its mode index or effective index (ñ) is n 2. The condition at ñ= n 2 is referred as mode cutoff. As mentioned before, single-mode fiber conditions are the objective (only HE 11 mode it is propagated in fiber, and this is the fundamental mode) and to achieve it the core radius a, the core refractive index n 1, and the cladding refractive index n 2, must satisfy the single-mode condition [1, 2] V= k 0 a (n 1 2 - n 2 2 ) 1/2 < 2,405, (4) where V could be viewed as a normalized wave number, since for a given fiber (fixed a, n 1 and n 2 ) it s proportional to the wave number (). It s also referred as normalized frequency. The smallest wavelength for which a given fiber is single mode is called the cutoff wavelength. Considering the linear regime, in the transmission process of a pulse over optical fiber some phenomenon occurs that affects the signal quality, such as attenuation, whose consequence is to reduce the optical power of the signal, and dispersion, which impact is the spread of the optical transmitted pulse. While attenuation determines the maximum distance between transmitter and receiver, the dispersion has a double influence, as it determines the maximum rate that the signal can be transmitted and in consequence also the distance between the transmitter and receiver. A. Attenuation Under normal conditions and considering only the linear r y Fiber core Fiber axis z

Qualification of fiber optic networks 4 effects, the power of an optical pulse propagating in a fiber is governed by the Beer equation [1, 2] Tab. 1. Fiber bandwidth ranges /, (5) where is the attenuation coefficient, dependent on the wavelength of the transmitted pulse, and it includes effects such as material absorption, Rayleigh scattering and waveguide imperfections. Fig. 7 depicts an experimental spectral attenuation fiber profile, measured in 1999 and installed 12 years before. B. Dispersion in single-mode fibers In the discussion of dispersion in single-mode fibers, chromatic dispersion (CD) and polarization mode dispersion (PMD), it will be considered. Fig. 7. Fiber spectral attenuation Material absorption can be divided into two types, namely losses due only to fiber material (silica) considered it as pure, and losses related with impurities in silica. The first ones are designated as intrinsic losses, and results from electronic (ultraviolet absorption) and vibrational (infrared absorption) resonances of silica molecules; and the other ones as extrinsic losses, where the main contribution comes from the presence of water vapors in silica. A vibrational resonance of the OH ion occurs near 2,73 m [1]. Its harmonic and combination tones with silica produces absorption effects at 0,95-, 1,24- and 1,39- m. The absorption effects at 1,24- and 1,39- m, could be seen at Fig. 7. Nowadays fibers where OH absorption effect it s notably reduced or even canceled are being produced. This type of fiber is named as dry fiber (low water peak-lwp or zero water peak fibers-zwp). Rayleigh scattering is a fundamental loss mechanism resulting from microscopic fluctuations in silica density, and it s proportional to -4. Waveguide imperfections results from the destruction, by any reason, of the cylindrical structure of fiber geometry, and it s divided into two categories: macro bends (occurs during fiber manufacturing) which ones are, in practice negligible, and micro bend (occurs during cabling or fiber installation). According to its spectral attenuation profile (or attenuation coefficient at specific wavelengths), fiber types were classified, in international standards, in four categories: Type A, specified at 1,31 m and 1,55 m; type B, specified at 1,55 m; type C (LWP), specified at 1,31 m, 1,383 m, 1,55 m and 1,625 m; and D (ZWP) as C type but zero attenuation around 1,383 m. It was considered useful by telecom industry the definition of fiber bandwidth ranges, which is illustrated in Tab. 1 1) Chromatic Dispersion (CD) The response of a linear and isotropic medium, to an applied electric field is represented, in frequency domain [2], by,,,, (6) where, represents the dielectric polarization, originated from an applied electric field, to a medium with linear susceptibility,,. The vacuum permittivity is represented by. As linear susceptibility () of silica has a frequency dependence behavior, different spectral components travel at different speeds, and arrive, to fiber end, at different moments, resulting in pulse broadening. This characteristic is called chromatic dispersion (also referred as group velocity dispersion or intramodal dispersion). The group velocity v g associated with the fundamental mode is (dβ/dω) -1. In a fiber with length L, where a pulse with spectral width was applied, the pulse spread T at the fiber end is [1]. (7) The parameter 2 = d 2 /d 2 is known as the Group velocity dispersion (GDV), and determines how much an optical pulse would broaden on propagation inside the fiber. Rewriting equation (7) in terms of wavelength (), by using =2c/ and = (-2c/ results in where, 1 2. D is called the dispersion parameter and is expressed in ps/(km-nm). In order to evaluate an estimate of the effect of dispersion (8) (9)

Qualification of fiber optic networks 5 on the bit rate B D, the criterion B D T<1 could be used, resulting in B D L D <1. Dispersion (D) could be split in two components: the dispersion in the material (DM) and waveguide dispersion (DW). DM occurs because the refractive index of silica, n(), changes with frequency, and increases as wavelength increases, from negative values to positive ones. The refractive index n() could be approximated by the Sellmeier series equation and the number of terms to include (3 or 5) is a parameter of CD measurement. DW depends on fiber parameters, such as the core radius a and the index difference n, and it is always negative, changing a little bit with wavelength (decrease as wavelength increases). D=DM+DW, and the wavelength at it is zero, is called zero dispersion wavelength ( ZD ), and its standard values is around 1,31 m. The fibers with this ZD value are called, standard fibers. DW manipulation trough core radius, index difference and other fiber parameters, makes it possible to design fibers such that ZD is shifted into the vicinity of 1,55 m, and the fibers have the designation of dispersion shifted fibers. It is also possible to tailor D in order to keep it at low values in a specific wavelength range (1,3 to 1,6 m), and these fibers are designated by dispersion-flattened fibers. The pulse broadening persists for higher order dispersion effects, which are governed by the dispersion slope (S) [1] curve, expressed by 2 4 (10) where 3 = d 2 /d d 3 /d 3 is the third order dispersion parameter of. At wavelength = ZD, 2 is zero and S is proportional to 3. It s possible to estimate the impact on the transmission rate (B D ) relative to the parameter S, for an optical source of spectral width (), were dispersion (D) becomes D = S, resulting in B D L S <1. In order to evaluate practical propagate pulse conditions in optical fiber there is a needed to consider spectral bandwidth of optical source (). So, if for a pure monochromatic wave frequency ( 0 ), the magnitude of the (real component) electric field vector associated with the wave are expressed by E(r,t) = J (x,y) cos [ 0t ( 0)z]. (11) Then for = 0 ±, E(r,t) = J (x,y) [cos (( )t ( )z) + cos (( )t ( )z)] (12) For pulses which << 0, it s useful to expand in a Taylor series around 0, so ~! (13) Considering the first order term of (13),, and applying it in (12), results after some mathematical and trigonometric work in, E(r,t) 2 J(x,y) cos(t z) cos( t z) (14) From (14) it s possible conclude that pulses in optical fiber, can be viewed in time t and space z, as the product of a rapidly varying sinusoid, cos( t z), by slow varying and nonsinusoid signal (envelope), cos(t- 1 z). The pulse travels at a velocity of 0 / 0, called phase velocity, whereas the envelope travels at 1/ 1, called group velocity. Equation (14) could be rewritten as E(r,t) = J(x,y) e [A(z,t) exp i( t z)] (15) The amplitude of A(z,t), the envelope, considering () represented by terms up to 3 rd order of (13), is done by [2], 1 2 0, exp 2 6 (16) Calculating / and noting that is replace by in time domain, it results or 2 1 6 0, (17) which represents the basic propagation equation that governs pulse evolution inside a lossless single-mode fiber. The expression of () till 3 rd order terms, as before, provide useful information about the impact of dispersion in transmission process Δ 2 Δ 6 Δ (18) where, = - 0 and m = (d m /d m ) in = 0 ; = v -1 g ; 2 is the DGV; 3 is the S value. 2) Polarization mode dispersion (PMD) As mentioned earlier, in the dispersion discussion in fiber optic propagation we must also consider PMD, which arises from birefringent characteristic of fiber material. A medium whose refractive indices are not equal along different directions (e.g. x and y) is said to be birefringent, and can it arise from geometry perturbations of the medium or due intrinsic property of the material. An optical fiber that shows small departures from its perfect cylindrical symmetry is said to be geometrical birefringent, and lead to a birefringent condition because of different mode indices associated with the othogonality polarized components of the fundamental mode. As the pulse excites both polarization components, and each have different propagation constants, or velocity, it

Qualification of fiber optic networks 6 becomes broader, as the components disperse along the fiber. Assuming the difference in constants propagation of modes () as invariant, then the time difference, known as differential group delay (DGD) due to PMD effect after propagation in the fiber, considered the unit length is given by Δτ= Δβ / ω [2]. However is not constant, both in time, due environmental change conditions, and along fiber length, because the effects could happens at different fiber location with different length. The above mentioned characteristics results in a randomly PMD effect, where some are canceled by others. To consider this behavior, DGD inverse dependence is not function on link length, but on the square root of it. III. STANDARD TEST METHODS AND FIBER SPECIFICATIONS The characterization of optical networks aims to identify any anomalies, capabilities and operational state of a network, or certify it, by applying standard test methods and specifications. Characterization of a fiber optic network should include the following tests: Characterization of optical splices (loss, reflectance, and location); Fiber attenuation coefficient; Link insertion (IL) loss and return loss (RL); Spectral attenuation (SA); Chromatic Dispersion (CD); Polarization mode dispersion (PMD). This set of tests and who will do the work, is a decision of the networks owner. In some cases, tests for characterization of optical splices, attenuation coefficient of the fiber and insertion loss, are contracted with the network installer, while the remaining tests, namely spectral attenuation, CD and PMD, are contracted to entities with capabilities accredited for this purpose. The fiber tests were performed accordingly the following standard test methods: IEC 60793-1-40 Optical fibres - Measuring methods and Test procedures: Attenuation (method B- insertion loss and method C- backscattering); IEC 60793-1-42 Optical fibres - Measuring methods and Test procedures: Chromatic dispersion (method A Phase shift ); IEC 60793-1-42 Optical fibres - Measuring methods and Test procedures: Polarization mode dispersion (method A Fixed analyzer, option 2 Fourier Transform, and method C Interferometry). IEC 61300-3-6 Fiber optic interconnecting devices and passive components- Basic test and measurement procedures Part 3-6 Examinations and measurements return loss (method 2B OTDR/ Optical time domain reflectometer by backscatter analysis). The applicable fiber specifications to verify, belongs to the group ITU-T Series G: Transmission systems and media, digital systems and networks Transmission media characteristics Optical fibre cables, namely: ITU-T G652 - Characteristics of a single-mode optical fibre and cable; ITU-T G653 Characteristics of a dispersion-shifted single-mode optical fibre and cable; ITU-T G655 - Characteristics of a non-zero dispersionshifted single-mode optical fibre and cable. In addition to these specifications, customer also refers: Allowable maximum splice loss 0,1 db/ splice; Allowable maximum connector loss 0,5 db/ connector. IV. TEST RESULTS The test results presented are a sample of the overall work done, and were chosen in order to illustrate specification compliance (brand new networks) or link limitations (old network under operation). In the examples showed are covered a network supported in G 653, G 652, and G 652+ G 655 fiber cable. A. G-653 fiber new network The whole network, installed in conduits, has about 830 km of fiber optic cable Dispersion Shifted G653 (ITU-T G653), with 32 or 24 fibers, corresponding in total to more or less 24,000 km of optical fiber. The specifications to confirm are depicted in Tab. 2 Tab. 2. G 653 Fiber and link specifications The attenuation test results, represented as average attenuation coefficient of all fibers in the cable link and respective standard deviation, in function of wavelength are depicted in Fig. 8. Fig. 8. Average spectral attenuation coeff. From the results we to conclude that specifications are met

Qualification of fiber optic networks 7 with tolerance (Link Max. Atn. Coef. @ 1550 nm 0,3 db/km). The fibers also show good performance at water peak wavelength (1383 nm). The CD test results, represented as average CD coefficient of all fibers in the cable link and respective standard deviation, as function of wavelength are depicted in Fig. 9. with eight fibers each. As environmental conditions are quite stable, its contribution to different behavior of fibers is negligible. Any damage of tubes 1 and 3 with impact in fibers it is possible, but it is quite strange why only the first fibers of mentioned tubes, have problems. Going deeper with this evaluation, it is possible that problems arise inside fiber organizer, due to bad fiber allocation, probably by submission of fiber to stress and short angles (each tube has its organizer). If this were the case, maybe these fibers attenuation coefficient were higher than others. The test results of fiber attenuation coefficient of this link are depicted in Fig. 12. Fig. 9. Average CD coeff. From Fig. 9 it is possible to conclude that 0 and CD maximum value, in range 1525 nm to 1575 nm are met. Fig. 12. Fibers attenuation coeff. Fig. 10. Slope @ 0 (S) From Fig. 10 it s also possible to conclude that S specification are fulfilled. In other network link (13,4 km length) were requested PMD tests. According PMD specifications and link length, the maximum allowable delay (ps) are 1,83 ps (0,5 13,4 = 1,83 ps). The test results are showed in Fig. 11. As it can be seen all fibers have similar attenuation coefficient, which does not confirm the previous assumption. As test values are not problematic, it is possible discharge it, and leave as it is. If this was not the case, then would be a need to ask for factory tests, if they exist, and analyze it. Splitting the link in individual sections, destroying the installation work, and repeat fiber tests for each section separately, is another approach. Let s assume that the section length are equal (6,7 km each) and the PMD delay of one section is similar to fiber 3/4, the best one (0,046 ps). The other section must have a PMD delay of 0,454 ps, and in consequence, a PMD coefficient of 0,18 ps/km 1/2. This value is better than specified (<0,5 ps/km 1/2 ), and in these conditions there is no problem in leaving the situation as it is, with no extra work (retest and redoing installation work). As showed in this example, test results must be evaluated individually and all together in order to outwit any trouble or noncompliance values, with negative impact on network performance, in present or future time. This is what customers want from 3 rd party accredited laboratories. Fig. 11. PMD/DGD @ 1550 nm As it can be seen all values are according specifications. However, fiber 1/2 and 17/18 have values that are much larger than other fibers. The cable is installed in underground conduits. The fibers in the cable are organized in three tubes B. G-652 fiber on operation network The cases reported here refer to sections of aerial or duct networks, with several years of installation, and eventually with several changes since its beginning. In addition, the cables used were probably manufactured at different ocasions (different years). The specifications are indicated in Tab. 3. The objective of the tests was to qualify the network to

Qualification of fiber optic networks 8 multigigabit transmission speed (>2Gb/s) and identify constraints to WDM using. Tab. 3 G652 Fiber and link specifications S, whose results are depicted in Fig. 15, we conclude that, besides 0 and S being specifications fulfilled, the fibers of section five are different from other ones, whic confirms our initial statement about link evolution in time. The link has an approximated length of 71 km and consists of five sequential sections ranging between 4.9 km and 22.8 km (L1-4.9 km; L2-15.2 km; L3-15.7 km; L4-12.5 km, L5-22.8 km). Each link section starts and ends at a telecom intermediate station. The connectors used are FC/PC. According to the customer request, five pairs of fibers in each section were tested, in insertion loss, spectral attenuation, CD and PMD. The resulting spectral attenuation results are shown in Fig. 13, indicates that specifications are met (Att. Coef. @ 1310 nm < 0,5 db/km; Att. Coef. @ 1550 nm < 0,4 db/km). The attenuation fiber profile, indicates the we are in presence of old fibers (maybe 80 s), because relevant loss at OH + ion wavelength influence (1,24-m and 1,38-m). Fig. 15. Link section 0 and S The PMD test results are depicted in Fig. 16. As we can see this specification is also met. Fig. 16. Link section PMD Coeff. and DGD Fig. 13. Link section spectral attenuation coeff. Regarding CD test results, shown in Fig. 14, it is possible to conclude that CD coefficient profile are similar to all link sections. The results of these tests reflect the fact that this connection has been built over time and certainly using optic fiber cables of various manufactures. Since the goal of the tests was to identify the spectral regions for use of WDM and qualification of these leads to binary debts (B D ) greater than 2 Gb/s, it is concluded that are available the following bands on spectral terms: Original (1260 to 1360 nm ), S (1460 to 1530 nm ) and C (1530 to 1565 nm ). To evaluate B D limitations, the rule which stipulates that the maximum delay (DGD) shall not exceed 1/10 of the bit width is applied. So we conclude that there are no objections regarding to B D till 40 Gb/s (bit width of 25 ps, 2.5 ps maximum delay. Fig. 14. Link section CD coef. Going deeper in CD measurements analysis, namely 0 and C. G-652/ G655 fiber new network This network has a ring configuration with length of approximately 600 km and was designed to withstand traffic with 10 Gb/s at its core. The access of the various LANs points (11 in total) to the core was designed to withstand traffic rate of 1 Gb/s. The cable core network uses fiber type G652-C (36 fibers) and LAN s, fiber type G655-C (12 fibers). The core is aerial type, while LAN is duct type.

Qualification of fiber optic networks 9 The objective of certification activity was network evaluation regarding International and Customer specifications, the audit of the installation process, perform network optical tests (IL, SA, CD and PMD) and B D transmission test (10 Gb/s on core; 1Gb/s on LAN s). The fiber and link specifications are depicted in Tab. 4. The CD and PMD results are showed in Tab.6 and Tab. 7 respectively, and we also conclude that specifications are fulfilled. Tab. 6 CD measurement results Tab. 4 G652/G655 Fiber and link specifications Tab. 7 PMD measurement results From overall tests performed, we chose a link with a length of 114 km, composed of two sections of equal length interconnected by E2000/APC connectors, where each of sections have 26 subsections interconnected by splice fusion. The tests consisted of insertion losses, spectral attenuation, CD and PMD. The results of insertion loss and attenuation coefficient are presented in Tab. 5. Tab. 5 Insertion Loss (1310, 1383, 1550 and 1625 nm In this link, 24 h PMD tests over one cable fiber (Fb 47) were also performed, which results (before and after) are depicted if Fig. 16. The results are displayed in Tab. 8. The results displayed shows conformity to specifications, with a significant margin. The spectral attenuation test results are depicted in Fig. 17, and also show good performance for each fiber type. Fig. 17. G655 Fb47 PMD test results. Tab. 8 G655 Fb47 PMD measurement results Fig. 16. Average link Atten. Coeff. As we can see PMD changes over 24h were not significant,

Qualification of fiber optic networks 10 as expected, because the weather conditions were stable (spring season) and not windy. The average global test results are depicted in Tab. 8, from which we conclude that specifications were met. Tab. 9 Global average test results and specifications V. CONCLUSIONS The test of a network of fiber optic communications is an activity of great importance because it allows: identification of any problems; guiding and supervision of the installation process; provision of network capability data and status; certify network capability. The resources involved in the tests are very relevant. The personnel involved must have background, knowledge, skills and experience in fiber optic networks installation and test. Relative to equipment, a quick estimate of the costs of test setup, points to a value in the order of 100,000.00, only for the tests referred in this work. If needed, for instance, to take a measurement of localized PMD, a figure around 200,000.00, it will be needed; and if we wished to perform trials of transmission traffic injection at 10 Gb/s or higher, we will need to consider, at least, an identical amount. This context justifies that price/cost of testing be relevant. The experience and information we collect through these years of work, shows that without 3rd party test control, performed continuously and consistently, the probability that there are errors and installation problems is not negligible. With regard to future work to develop, we refer the study of aging in telecommunications fiber optic networks, which could focus on the G653 network installed in 1994. Nearly 20 years have elapsed since the construction of this network; its time guarantee it is not exhausted; it is supported on a stable conduit infrastructure, and its core layout remains as it was in the beginning. This study would involve the need to carry out further testing, in a sample basis, and the network owner must assume the cost. For old networks, we consider the testing activity about PMD constrains to high speed signal transmission (above 40 GHz). In these networks, the PMD test should be performed in the shortest possible sections, because these sections may have each different age, cable manufacturers and fibers. It is important to identify as accurately as possible the location of any problems. Also in these networks, especially in aerial sections, it should be make the measurement of PMD for 24 hours period, and at different times of the year, covering hot, cold and windy seasons. REFERENCES [1] G. P. Agrawal, Optical fibers, in Fiber-optic communication system, 4ª ed., New York: John Wiley & Sons Inc., 2000. [2] R. Ramaswami et al, Propagation of signals in optical fiber, in Optical Networks A pratical prespective, 3ª ed., New York: Elsevier, 2010 [3] D. Anderson, et al, Fundamentals of fiber optics, in Troubleshooting Optical-Fiber Networks, 2ª ed., New York: Elsevier 2004 [4] H. Murata, Handbook of Optical Fibers and Cables, 2ª ed., New York: Marcel Dekker 1996, [5] IEC (International Electrotechnical Commission), Optical fibres - Measuring methods and Test procedures: Attenuation, IEC 60793-1-40, 1ª ed., 2001 [6] IEC (International Electrotechnical Commission), Fiber optic interconnecting devices and passive components- Basic test and measurement procedures Part 3-6 Examinations and measurements return loss, IEC 61300-3-6, 2ª ed., 2003 [7] IEC (International Electrotechnical Commission), Optical fibres Measurement methods and test procedures Chromatic Dispersion, IEC 60793-1-42, 2º ed., 2007 [8] IEC (International Electrotechnical Commission), Optical fibres - Measuring methods and Test procedures: Polarization mode dispersion, IEC 60793-1-48, 2ª ed., 2007 [9] ITU-T (Telecommunication Standardization Sector of ITU), Definitions and test methods for linear, deterministic attributes of single-mode fibre an cable, Rec. G 650-1, 07/2010 [10] ITU-T (Telecommunication Standardization Sector of ITU), Definitions and test methods for statistical and non-linear related attributes of single-mode fibre an cable, Rec. G 650-2, 07/2007 [11] ITU-T (Telecommunication Standardization Sector of ITU), Characteristics of a single-mode optical fibre and cable, Rec. G 652, 11/2009 [12] ITU-T (Telecommunication Standardization Sector of ITU), Characteristics of a dispersion-shifted single-mode optical fibre and cable, Rec. G.653, 7/2010 [13] ITU-T (Telecommunication Standardization Sector of ITU), Characteristics of a non-zero dispersion-shifted single-mode optical fibre and cable, Rec. G.655, 11/2009