Multi-layer Fiber for Dispersion Compensating And Wide Band Amplification

Multi-layer Fiber for Dispersion Compensating And Wide Band Amplification Asso. Prof. A. S. Samra, Eng. H. A. M. Harb Department of Electronics and Communications Faculty of Engineering, Mansoura University, Egypt (e-mail: ahsamra@yahoo.co.uk, harbhani@yahoo.com) ABSTRACT This paper investigates the dispersion compensating performance in multi-layer fiber. We show that very large negative dispersion value can be obtained, depending on the geometrical parameters. The splice loss between the standard G.65 fiber and the multi-layer fiber is calculated. Raman amplifier using multi-layer fiber as a gain medium is investigated using one pump as well as pumps, ASE is calculated. Keywords: Dispersion compensation, multi clad fiber, Raman amplification 1 INTRODUCTION In recent years DWDM optical communication is seeing a steady mitigate from.5 to 4 Gbps over each wavelength achieving higher spectral efficiency, which is defined as the ratio of average transmission rate to channel spacing. Amplification and dispersion compensation/management have assumed great importance as there are the main impairing factors for achieving repeater less transmission distance in excess of 1km over standard single mode fibers. One of the earliest techniques suggested to reduce the dispersion at 155nm band was to tailor the refractive index profile of a single mode fiber in such a way that its zero dispersion wavelength is shifted from the conventional 11nm window to a round 155nm[1]. These fibers, called dispersion shifted fibers (DSF) through appeared promising for a while, but, were found to be unusable in DWDM link due to the fact that operating a fiber with near zero dispersion is known to introduce nonlinear effects like FWM[]. It is known that FWM effect can be greatly reduced by allowing a small but finite local dispersion all along a DWDM link. This task could be fulfilled either through dispersion management (i.e. by combing alternate lengths of positive and negative dispersion fibers []) or by employing so called nonzero dispersion shifted fibers. Which is designed to leave a small residual average dispersion of.6ps/km.nm to omit nonlinear propagation effects in the single mode fiber. Chromatic dispersion is a linear effect and inserting a component with opposite sign could greatly reduce its detrimental effect in G.65 fibers at the 155nm band. Out of the several different technique that have been proposed in the literature, the ones which seem to hold immediate promise could be classified as dispersion compensating fiber (DCF)[4], chirped fiber Bragg grating (FBG) [5],[6], high order mode (HOM) fibers[7]. In chirped grating the optical pitch (product between the grating period and the mode effective index) varies along length of the FBG. As a result, resonant reflection frequency of the FBG becomes a function of position along length of FBG. Thus, each frequency component of a propagating pulse is reflected from a different point along length of chirped FBG. This is depending on the sign of the chirp; a chirped FBG could impart either a positive or negative dispersion to a propagating pulse [8]. Since, dispersion compensation is achieved or reflection to access the dispersion corrected pulse. And optical circulator or a fiber coupler is required as an additional component with associated insertion loss. Further more, errors in the chirped phase mask periodicity could lead to ripples in group delay with wavelength. The HOM technique exploits large negative dispersion slop, which are characteristics of higher order modes of a fiber relative to the fundamental mode. Thus one requires a fiber, which supports more than one mode at the operating wavelengths. Further, conversion of power from the fundamental to a higher order mode and reconverting the same back to fundamental mode have not been as easy task through lately there have been a number of promising studies reported on the technique[9],[1].out of all these, by far the DCF technique has been the most widely used technique. One of the main advantages of this technical solution is that when appropriately designed it can provide a passive system, in principle, with negative chromatic dispersion coefficient D as high as -5ps/km.nm. Such a scheme should compensate the positive chromatic dispersion over a relatively short length of the DCF and having low sensitivity to environmental influence (temperature, UbiCC Journal, Volume 4, Number, August 9 87

vibertional, etc) like that of the signal carrying fiber [11]. After a brief presentation of many possible solutions usable to compensate dispersion, we will discuses the multi-layer fiber design in the next section. Different variations in refractive index profiles with several layers and different shapes had been simulated in previous work [11],[1], but, step index profiles composed of two concentric spatially separated cores appears to give the best performances according to the fabrication constraints. The third section focuses on the splicing losses between the multi-layer fiber and the standard G.65 fiber. The use of multi-layer fiber with Raman amplifier will be discussed in the fourth section. Finally, conclusions are presented in section 5. Multi-layer Fiber Fig. 1 shows the refractive index profile of the multi-layer fiber which has a dual core design. It consists of two concentric cores: the inner core with a large and the outer core with a small. Here is defined as = ( n n )/n.where 1=. and i i =., n is calculated with the well known Sellmeier equation: n A. λ (1) ( ) i = 1+ j= 1 λ Bj with A 1 =.698188, A =.4865177, A =.89749, B 1 =.755551, B =.1176566, B =9.87549 are the Sellmeier constants [1]. As seen from Fig. 1 multi-layer fiber has four distinct regions: rod (<r< r 1 ), gap (r 1 <r<r ), barrier (r <r<r ), and clad (r>r ), and can be thought as composed of two substructures namely rod and tube, as indicated in the fig. (1). It is seen from Fig. that at wavelengths shorter than 155nm the field is essentially confined to the inner core and for which the guide essentially functions like a step index single mode fiber, the effect of the outer core being negligible. Around 155nm optical coupling takes place between the inner and the outer core modes. At wavelengths longer than 155nm however, most of the power of the fundamental mode spreads to the outer core and is effectively guided in the outer core. The fractional power in the second supermode of the fiber, which is orthogonal to the first supermode, is maximal in the outer core for wavelengths longer than 155nm. This phenomenon induces a rapid change in the slope of the effective index (ne) versus wavelength around 155nm [11],[1]. Fig. depicts the variation of n e as a function of the wavelength for the fundamental supermode. The resulting chromatic dispersion coefficient of the fiber is then computed through the following formula: λ dn D = e () c dλ A sample result corresponding to the fiber parameters in Fig.1 shown in Fig.4. Since such a profile is easily attainable with common perform fabrication systems. n 1 n Multi-layer fiber n Rod fiber r 1 1 1.8.6.4. Amplitude 1.5 5 5 Tube fiber r r 1.55 1 λ(µm) 1.6 15 15-1 Radius (µm) Fig.(1) Refractive index profile of the multi-layer Mfiber. Fig.() Evolution of the mode amplitude of the fundamental supermode versus wavelength. UbiCC Journal, Volume 4, Number, August 9 88

Effective index The parameters of the two cores are chosen that each of these supports a single azimuthally symmetric mode in the operating range of wavelengths. The fiber parameters are so optimized that the two individual modes (corresponding to the inner and the outer core) are nearly phase matched at 155nm. In such a case, because the non-supermode, this mode is expected to have a large dispersion, the magnitude and spectral variation of which can be optimized by varying the separation (r -r 1 ).the behavior of the dispersion curve is sensitive to the variation of the rod radius as see from Fig. 5. Splice Losses To calculate the splice loss between the standard G.65 fiber and the designed multi-layer fiber, we have used the analysis given in [14]. According to this analysis the fractional power coupled from a G.65 fiber to a multi-layer fiber is given by the equation (4), where subscripts in ψ (which is the mode field shape) correspond to the fiber type. Thus, the total splice loss, including both input and output splices, is given by Total splice loss= log1 T () where, T = dx ( ) 1 π ψ ψ rdrdϕ G.65 Coaxial π π ψ ψ.65.65. G G rdrdϕ ψ ψ Coaxial Coaxialrdrdϕ 1.45 1.445 1. 1.6 1. 1.8 1.44 1.5 1.56 1.6 1.68 1.74 1.8 Wavelength x(µm) (4) The spectral variation in the operating range of wavelengths, assuming 4 % tapering of the G.65 fiber, including the effect of these wavelength dependant losses in our analysis, we have iteratively tuned the fiber parameters such that the net output gain spectrum is flat. Fig.7 shows the spectral variation of the total splice loss at the input and the output splice Dispersion ps/km.nm fx ( ) 4 Amplification 5 1 1.4 1.5 1.6 x Wavelength (µm) Fig. 4. Evolution of the chromatic dispersion versus wavelength. Fiber Raman amplifier (FRA) is considered to be a key component to realize a next generation photonic networks because of its features of the noise reduction, flexible gain bandwidth, and simple configuration. The Raman amplifier configuration with forward pumping is shown in Fig.8, the pumping signals are launched into fiber through an optical coupler and propagate a long with the information signals that are fed at the fiber input [15]. A typical Raman gain spectrum for pure silica fiber is shown in Fig. 9, for the pump wavelength of 145nm. The optical signal gain strongly depends on the Raman gain coefficient, which is a function of the wavelength. The total amplified power over all signal band of the optical fiber Raman amplifier with one pumping source is Psig ( z) = Psig ( υ, z) dυ, where υ is the stokes frequency, and the signal spectral power density P sig (υ,z) is the stokes power P s (υ,z) in z point along the fiber per unit frequency range. Fig.. Evolution of effective index of the fundamental supermode with wavelength. To reduce the splice loss and its spectral variation, we have considered the G.65 fiber to be tapered by 4%. The effect of tapering has been modeled as scaling of the fiber dimensions. This results in the spreading of the model field of the G.65 fiber, leading to a better overlap with the fundamental supermode field of the multi-layer fiber in addition. Multi-layer fiber shows high Raman amplification. Multi-layer fiber can be designed in a manner that in the wavelength range where Raman gain coefficient (g R ) decreases, the effective area of interaction A eff also, decreases in almost the same manner. Such that the effective Raman gain (g R /A eff ) is reasonably flat on a large wavelength range. Using multi-layer fiber with only one pump will achieve flat gain Raman amplification over any band [16]. UbiCC Journal, Volume 4, Number, August 9 89

-Dispersion ps/km.nm 1 5 75 15 75 5.9 1 14 5 6 7 8 9 1 11 1 1.1 r 1 (µm) 1. 1.5 4.5 6 1.5 7.5 9 1.54 1.5 1 1.55 λ (µm) 1.5 151.56 Standard SMF G.65 Fig. 5. Wavelength dependent dispersion curves for different r 1 M Multi-layer fiber Standard SMF G.65 Splice Splice Fig. 6. Input and output splices between SMF(G.65) and multi-layer fiber. Input signal Coupler PUMP Forward pumping Output signal Figure 8. Configuration of Raman amplifier with forward pumping. 1 Total splice loss (db) fx ( ) 1.9 1.8 1.7 1.6 15. 15.4 15.6 15.8 16 x Wavelength (µm) Fig. 7. Variation of the total spices loss at the input and output splices. This fiber has a unique property that the effective Raman gain spectrum is inherently flat over a large wavelength range and the effective Raman gain spectrum can be tuned by the fiber parameters.the parameters of the two cores are chosen so that each of them supports a single azimuthally symmetric mode in the operating range of wavelength. These parameters are optimized such that the two individual modes (corresponding to the inner and the outer core) are nearly phase matched at 155 nm [17], [18]. The signal wavelengths below PWM will be tightly confined to the inner core, leading to a high pump-signal overlap, and thus a small A eff. However, as the signal wavelength approaches the phase matching wavelength, the fractional power of the fundamental mode will gradually increase in the outer core. Raman Gain coefficient (m/watt) g( υ ) 1.8.6.4. 1.45 1.5 1.55 1.6 1.65 1.7 υ 1.4 Wavelength 1 (µm) Figure 9. Raman gain spectrum in a pure silica optical fiber for pump wavelength 145nm. Hence, the overlap between the pump and the signal fields starts to decrease, increasing the effective area. Thus, by suitably choosing the fiber parameters, phase matching wavelength and the pump wavelength, one can ensure that the decrease in A eff almost compensates for the decrease in the Raman gain coefficient, such that a flat effective gain spectrum is achieved. Hence, the mode field at the pump wavelength 145nm and the signal wavelengths will be tightly confined to the inner core and thus the pump and the signal overlap will be high, giving a small A eff. However, as the signal wavelength approaches and crosses the phase wavelength, the fractional power of the fundamental mode will gradually increase in the outer core. UbiCC Journal, Volume 4, Number, August 9 81

c1( υ ) Assuming the same parameters for all refractive indices do not affect the general trend of the results presented here. The effective index n e is shown in Fig., for a multi-layer fiber with r 1 =1µm, r =15µm, r =µm. Fig.1, shows the effective Raman gain spectrum using multi-layer fiber for the band model, it is obvious that a flat effective gain spectrum with - db is achievable over the range (145-155 nm), corresponding ASE is shown in Fig. 1. Fig. 11 shows the effective Raman gain spectrum using inherently flat gain with two pumps, such Raman amplifier can serve as broadband amplifier (for C and L bands) as well as dispersion compensating module. It is also cost-effective since it uses only two pumps (146, 1486nm). The ASE curve is shown in Fig. 1, and by compare it with the multipumping scheme, we can conclude that decreasing the number of pumps, also decreases the ASE. Effective Raman gain 1. log( s1( υ 1) ) 1 c( υ ) 5 15 1.45 1.5 1.55 1.6 1.65 υ 1.45. Wavelength 1 (µm) Figure 1. Effective Raman gain with inherently flat fiber with single pump. Effective Raman gain 1. log( s1( υ 1) ) 1 5 15 5 CONCLUSION Band model pump with IFGF Band model With IFGF 1.45 1.5 1.55 1.6 1.65 υ Wavelength 1.45. (µm) Figure 11. Effective Raman gain with inherently flat fiber with two pumps. We briefly review the most usable solutions to compensate dispersion. We also, study the effect of chromatic dispersion of the multi-layer fiber, showing the effect of varying the geometric parameters (rod radius). The splice loss between the multi-layer fiber and SMF(G.65) is also presented. e( υ ) ASE Power (db) e1( υ ) 5 1.6 e( υ ). 4.8 6.4 8 Figure 1. ASE power of the inherently flat gain fiber with single pump and two pumps. We show that using multi-layer fiber as gain medium in Raman amplifier show that multi-layer fiber is inherently flat gain (IFGF). The Raman amplifier with single pump using multi-layer fiber as well as with two pumps is introduced; also, the ASE is estimated. It is obvious that, decreasing the number of pumps also decreases the ASE. Here, we conclude that Raman amplifier with two pumps using inherently flat gain fiber can cover the C and L bands for DWDM communication systems. MATHCAD is used as an analytical programming tool. REFERENCES Single pump with IFGF pumps with IFGF Multipumping 1.5 1.55 1.6 1.65 υ Wavelength 1.7.. (µm) 1 [1] M.A. Saifi, S. J. Lang, L. G. Cohen, and J. Stone, Triangular profile single mode fiber, Opt. Lett., vol.7, No. 4 (198). [] G. P. Agrawal, Nonlinear fiber optics Third ed. Academic, San Diego, Ca. 1. ISBN: -1-4514-. [] I.P. Kaminow, Optical fiber telecommunications, Elsevier Academic Press IV. ISBN: -1-9517-. [4]A. Huttunen,"Optimization of dual-core and microstructure fiber geometries for dispersion compensation and large mode area", OPTICS EXPRESS, Vol.1, No., Jan. 5. [5] Ruchti, Randy, Performance of multiclad scintillating and waveguide optical fibers readout with visible light photon counters Proc. SPIE Vol. 7, p. 78-94,7. [6] B.J. Eggleton et al., Recompression of pulse broadened by transmission through 1 km of non dispersion shifted fiber at 1.55m using 4 mm long optical fiber Bragg grating with tunable chirp and central wavelength, IEEE Photon. Technol. Lett. Vol.7, no. 5, 1995. [7] G. P. Agrawal, Raman Amplification in fiber optical communication systems 1 st ed. Elsevier Academic Press, 5. ISBN: -1-4456-9. [8] B.P. Pal, All fiber components,, in Electromagnetic field unconventional structures and material A. Lakhtakia and O. N. Singh, Eds. Wiley, New York,. [9] S. Ramachandran et al., All fiber grating based higher order dispersion compensator for broadband compensation and 1km transmission at 4Gbps, In Proc. ECOC, Paper PD-5,. [1] A.H.Gnauck, L. D. Garret, Y. Danziger, L. Levy and M.Shur, Dispersion and dispersion slop compensation of NZ-DSF for 4 Gbps operation over the entire C band. In Proc. OFC, Paper PD-8,. [11] K. Thyagarajan, R. K. Varshney, and P. Palai, A novel design of a dispersion compensating fiber, IEEE. Photon. Techno. Lett, Vol.8, no.11 1996. [1] P. Palai, R. K. Varshney, and K. Thyagarajan, A dispersion flattening dispersion compensating fiber design for broadband dispersion compensation, Fiber Integr. Opt., Vol., 1, pp.1-7. UbiCC Journal, Volume 4, Number, August 9 811

[1] M. J. Adams, An Introduction to optical waveguides John Willy & Sons, pp.1, 1981 [14] A. Ghatak and K. Thyagarajan, Introduction to fiber optics. Cambridge Univ. Press, 8. [15] M.N. Islam, Raman Amplifiers for Telecommunications IEEE J. Sel. Top. Quant. Elect., Vol.8, No.,, pp.548-559. [16] S. P. Singh and N. Singh, "Nonlinear Effects in Optical Fiber: Origin, Management and Applications", Progress In Electromagnetic Research, PIER 7, 7, pp.49 75. [17] I.P. Kaminow, Optical fiber telecommunications, Elsevier Academic Press IV. ISBN: -1-9517. Ahmed Shaban Samra was born in Mansoura,Egypt 1954. He received the B.Sc. and the M.Sc degree in communications engineering from Menoufia University 1977, 198 respectively, and the Ph.D. degree in optical communications and integrated optics from ENSEG, Greroble, France in 1988. He is now an associate professor at the faculty of engineering, Mansoura University. His research interests are in the field of optical communications and optical measurement technique. Hani Ali Mahmoud Harb was born in Mansoura, Egypt 1976. He received the B.Sc. in electronics engineering and the M.Sc degree in communications engineering, both from Mansoura University, Egypt, in 1999 and, respectively. He is currently working toward the Ph.D. degree in communications at Mansoura University. His research activities have been devoted to optical communication systems, optical CDMA, DWDM, and Raman fiber amplifiers. UbiCC Journal, Volume 4, Number, August 9 81