61 CHAPTER 5 SPECTRAL EFFICIENCY IN DWDM 5.1 SPECTRAL EFFICIENCY IN DWDM Due to the ever-expanding Internet data traffic, telecommunication networks are witnessing a demand for high-speed data transfer. The increase in data transfer rate could be achieved either by increasing the base band transmission rate, and /or increasing the number of wavelengths in Ultra Dense Wavelength Division Multiplexing (UDWDM) and/or increasing the Spectral Efficiency (i.e Capacity per unit bandwidth) in a fiber-optic backbone. By increasing the number of wavelengths to be transmitted through a single mode fiber (pair), the nonlinearities like FWM and SRS would also be increased. Jau Tang (2001) has demonstrated that the spectral efficiency decreases beyond certain input optical power. The Spectral efficiency Vs Input optical power for various number of channels is shown in Figure 5.1. Also for the increase in the number of wavelengths, the reduction in spectral efficiency starts at much lower input powers. Hence the input optical power has to be optimum or minimum so as the increase the spectral efficiency. It would be better if the Spectral efficiency rather than the number of wavelengths in a DWDM fiber optic communication system is increased. This is because the cost of the end equipments in a DWDM system is proportional to the number of wavelengths. Hence in order to minimize the cost of communication, the capacity per unit bandwidth (i.e Spectral Efficiency) has to be increased.
62 Stimulated Brillouin Scattering (SBS) sets the threshold maximum power for a single channel fiber-optic communication system, whereas SRS sets the limit of the transmitted power in a WDM signal. SBS produces the stokes signal due to the generation of acoustic phonons which will transfer the power from the incoming optical signal. The Stokes signal travels in the backward direction and hence there is a reduction in the forward propagating optical signal, which will come out of the fiber. The power threshold for SBS depends on the fiber core effective area and inversely proportional to the effective length of the fiber. Hence if the fiber is a LEAF (Large Effective Area Fiber), then it can withstand large optical input power and hence it has high spectral efficiency. However, the large core area, requires other constraints in the refractive index of the core and cladding in order to make the fiber to be in a single mode operation. Also every wavelengths need certain minimum optical input power for maintaining its own OSNR for easy detection at the optical receiver side. Figure 5.1 Spectral Efficiency Vs Input Power for different values of N [Courtesy :Jau Tang, Journal of Lightwave Technology, Vol.19,No. 8, Aug 2001]
63 But, allowable power per channel is less for high value of N, which otherwise needs high total power. Hence we would have to use only optimum number of wavelengths. As the number of wavelengths increased within a particular range [ min to max ], which may be C band (1525 nm to 1565 nm) or L band (1565 nm to 1625 nm), then allowable P max is decreased. It puts the restriction on N. Line widths (spectral widths) of the individual optical sources in each transmitter has its influence on SBS, which limits the optical power that can be transmitted through a single-mode fiber in long distance optical communication systems. 5.2 NEED FOR INCREASING THE SPECTRAL EFFICIENCY Even though, fiber has bandwidth ocean, the number of wavelengths in a DWDM can not be increased as the cost of the DWDM Multiplexing and Demultiplexing equipments are increased for the high wavelength count. If Spectral efficiency i.e the capacity per unit bandwidth is increased, more traffic could be accommodated into a single wavelength carrier and hence less no. of wavelengths are needed. Different modulation techniques have different spectral efficiencies. Improved spectral efficiency will increase the total transmission capacity even without the increase in OA gain bandwidth. It can thus relax the required flatness for each EDFA. Everybody, in all sectors of life are using Internet, which needs high bandwidth transmission media. In the dispersive fiber, by introducing the dispersion in the controlled manner, fiber channel capacity can be improved. Shannon derived the celebrated formula for the capacity SC of the communication system as SC = B. log 1 + bits/sec (5.1)
64 where B is the channel bandwidth. P o is the average optical signal power and P N is average noise power. Spectral Efficiency (SE) is given by Equation (5.2) SE = SC /B bits /sec /Hz (5.2) Equation (5.1) is valid only for the linear channel. However, when the field intensity inside the fiber core region gets increased due to increase in the number of wavelengths then the fiber behaves in a nonlinear manner. SC would become a function of dispersion of the fiber as shown in Equation 5.3. Jau Tang (2001) has given the exact expression in the nonlinear regime involving the parameters like N c - the number of channels, c the bandwidth of each channel for the dispersion free nonlinear optical fiber and a fiber with dispersion. For an S/N ratio of 40 db for a typical optical transmission and receiver/amplifier system, it corresponds to about 13 bit/s/hz if Kerr nonlinearity is not present. However, in addition to the amplifier noise, one needs to include the effects from Kerr nonlinearity, which also degrades transmission quality. To investigate the effects of Kerr nonlinearity (K) on channel capacity in optical fiber transmission, Conventional Shannon s treatment for a linear system is generalized to a nonlinear system where the nonlinear noises are induced by inter- and intra channel interference from input signals.
65 The calculation of the Shannon channel capacity for optical transmission in a nonlinear fiber involves analytical solution of the nonlinear Schroedinger equation (NSE). Although it is known in practical dense wavelength division multiplexing (DWDM) applications, fibers without dispersion are inferior to fibers with dispersion. Figure 5.2 Dependence of Spectral Efficiency on Input Power for different K Figure 5.2 shows the dependence of Spectral efficiency on input optical power for various Kerr coefficient K (km -1 W -1 ). For the Kerr coefficient of 0 per watt per kilometer, the spectral efficiency linearly increases without any turning point. With the Kerr coefficient of 1.22 /W/km, there is a turning point of linearly increasing spectral efficiency at 0.1 mw per channel.
66 5.3 Multi Span System Figure 5.3 shows the block diagram of a Multispan system. A(t) is the input of M Multiplexed DWDM channels. A(L,t) is the output after one fiber span of length L. A(t) Fiber A(L,t) A Coherent Detector N Spans A Optical Amplifier Figure 5.3 Block Diagram of a Multi Span System Transmission system with ideal coherent detection is considered here. At the end of each span, the optical signal is amplified to compensate for the fiber attenuation and white Gaussian amplifier noise is added. written as follows Shannon channel capacity formula for nonlinear dispersive fiber is P/ c SE = log 2 1+ 1+ 2 2 L N s T 2 (P/2 c) 2 N s L 2 (P/ c) 3 /( ) ln N c 2 c 2 + N s P w [1+ L 2 N s 2 T 2 (P/2 c) 2 ] 3 ---- (5.3)
67 where Ns Number of spans Pw - White noise power density (W/THz) - Attenuation coefficient( db / km) - Kerr nonlinearity coefficient(km -1 W -1 ) L = (1-exp(- L))/ Nc - No. of channels T = 2 x x f T f T - total bandwidth(thz) P - Input power per channel (Watts) c = 2 x x f c f c - Channel width (THz) - Coefficient of dispersion (ps 2 /Km) In the above formula, the left hand side is to be maximized by considering as a variable and the condition on which will maximize the left hand side i.e the Shannon Capacity should be obtained. For that, this Shannon Capacity equation could be differentiated with respect to the dispersion and the value of the dispersion at which the Shannon Capacity per unit bandwidth is maximum is obtained. The optimum coefficient of dispersion opt is obtained in Equation (5.4).
68 opt = (4 ) / (Nc 2. c 2 ) (5.4) where is the attenuation of the fiber Nc is the no. of channels c is the bandwidth of one channel For the particular set of parameters, Attenuation = 0.2 db/km, Fiber length L = 80 km the numerical value of opt is obtained as opt = 5.2825x10-5 ps 2 /km Figure 5.4 shows that for the dispersive fiber, the spectral efficiency reaches the maximum value at an optimal. Figure 5.4 Spectral Efficiency Vs Dispersion for Dispersive Fiber Thus the optimum dispersion could be set to have more channel capacity per unit bandwidth in the fiber.
69 In Long-Haul DWDM optical networks, higher bit rate, increased number of wavelengths (channels) and /or higher optical signal power limits the capacity per unit bandwidth or the spectral efficiency. This is because the refractive index of the core becomes light intensity dependent. The possibilities of improving the spectral efficiency by various means is thoroughly analysed in Jau Tang (2001). Consider a case where there are two choices for getting the overall capacity of 100 Gb/s, as follows: Choice 1: 20 channels of 5 Gb/s each. Choice 2 : 10 channels of 10 Gb/s each. From the nonlinearity point of view, Choice 1 is better, because 5 Gb/s needs 5 mw / Channel, whereas 10 Gb/s channel needs 10 mw/ channel in order to maintain the same energy density inside the fiber core. Based on the above fact, the number of channels should not be decreased too much, which would otherwise demands more optical power per channel as the data rate gets increased. However, the spectral efficiency decreases when the number of wavelengths are increased. Hence the number of wavelengths should be the optimum one. In order to meet today s data traffic demands, the overall capacity has to be maximized by all means. In that process, the number of wavelengths should be optimum (That is it should not be too high as viewed from economical point of view or it should not be too low as viewed from nonlinearity point of view). The paper titled Capacity Optimization in DWDM Optical Communication System has been published in an International Journal on
70 Information and Communication Technologies [IJCT], Vol. 2, No. 1-2, 2009. The paper titled Analysis of Optical Modulation formats for DWDM System has been published in an International Journal on Information and Communication Technologies [IJCT], Vol.4, No. 1-2, 2011. The paper titled Capacity Improvement in Dispersive, Non-Linear Optical Fiber has been published in an International Conference on Computational Intelligence and Multimedia Applications [ICCIMA-2007] conducted by IEEE Computer Society in December 13-15, 2007.