idnits 2.17.1 draft-papadimitriou-ipo-non-linear-routing-impairm-01.txt: ** The Abstract section seems to be numbered -(618): Line appears to be too long, but this could be caused by non-ascii characters in UTF-8 encoding -(637): Line appears to be too long, but this could be caused by non-ascii characters in UTF-8 encoding -(646): Line appears to be too long, but this could be caused by non-ascii characters in UTF-8 encoding Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** Looks like you're using RFC 2026 boilerplate. This must be updated to follow RFC 3978/3979, as updated by RFC 4748. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- ** Missing document type: Expected "INTERNET-DRAFT" in the upper left hand corner of the first page ** The document seems to lack a 1id_guidelines paragraph about 6 months document validity -- however, there's a paragraph with a matching beginning. Boilerplate error? == There are 42 instances of lines with non-ascii characters in the document. == The page length should not exceed 58 lines per page, but there was 13 longer pages, the longest (page 14) being 60 lines Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) ** The abstract seems to contain references ([IPO-ORI], [IPO-IMP]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == Line 552 has weird spacing: '... 0.3 pi at 40...' == The document doesn't use any RFC 2119 keywords, yet seems to have RFC 2119 boilerplate text. == Unrecognized Status in 'Category: Informational Draft', assuming Proposed Standard (Expected one of 'Standards Track', 'Full Standard', 'Draft Standard', 'Proposed Standard', 'Best Current Practice', 'Informational', 'Experimental', 'Informational', 'Historic'.) -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (November 2001) is 8169 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) -- Missing reference section? '1' on line 21 looks like a reference -- Missing reference section? 'IPO-IMP' on line 633 looks like a reference -- Missing reference section? 'IPO-ORI' on line 637 looks like a reference -- Missing reference section? '2' on line 68 looks like a reference -- Missing reference section? 'AGR-FOCS' on line 621 looks like a reference -- Missing reference section? 'OFC00-NLP' on line 645 looks like a reference -- Missing reference section? 'ELEC-ODS' on line 625 looks like a reference -- Missing reference section? 'OFC02-NLP' on line 641 looks like a reference -- Missing reference section? 'AGR-NFO' on line 618 looks like a reference -- Missing reference section? 'GYS-XT' on line 629 looks like a reference Summary: 6 errors (**), 0 flaws (~~), 5 warnings (==), 12 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 IPO Working Group Dimitri Papadimitriou 3 Category: Informational Draft Jean-Paul Faure 4 Expiration Date: May 2002 Olivier Audouin 5 Alcatel 7 Roy Appelman 8 Civcom 10 November 2001 12 Non-linear Routing Impairments 13 in Wavelength Switched Optical Networks 15 draft-papadimitriou-ipo-non-linear-routing-impairm-01.txt 17 Status of this Memo 19 This document is an Internet-Draft and is in full conformance with 20 all provisions of Section 10 of RFC2026 [1]. 22 Internet-Drafts are working documents of the Internet Engineering 23 Task Force (IETF), its areas, and its working groups. Note that 24 other groups may also distribute working documents as Internet- 25 Drafts. Internet-Drafts are draft documents valid for a maximum of 26 six months and may be updated, replaced, or obsoleted by other 27 documents at any time. It is inappropriate to use Internet- Drafts 28 as reference material or to cite them other than as "work in 29 progress." 31 The list of current Internet-Drafts can be accessed at 32 http://www.ietf.org/ietf/1id-abstracts.txt 33 The list of Internet-Draft Shadow Directories can be accessed at 34 http://www.ietf.org/shadow.html. 36 1. Abstract 38 Today, in transparent optical networks, the increasing bit-rate (10 39 Gbit/s and up to 40 Gbit/s in the future), combined with the 40 increasing number of wavelengths (16 and higher up to 320) and a 41 narrowing of the channels spacing, enhance the impact of non-linear 42 effects on optical signal quality. 44 Thus, non-linear effects like Self-Phase Modulation (SPM), Cross- 45 Phase Modulation (XPM), Four-Wave Mixing (FWM) as well as Stimulated 46 Raman Scattering (SRS) and Brillouin scattering have to be examined 47 in order to evaluate their impacts on the transmission quality. If 48 these effects appear to be significant, they have to be taken into 49 account in the routing of a wavelength throughout a transparent 50 optical network. 52 D.Papadimitriou et al. � Expires May 2002 1 53 The aim of this draft is to extend the previous works dedicated to 54 routing impairments ([IPO-IMP] and [IPO-ORI]) in order to determine 55 which are the non-linear effects that must be considered and which 56 kind of engineering rules may be used to take these effects into 57 account in constraint-based optical routing. 59 Moreover, we propose to introduce IGP routing protocol extensions to 60 transport information related to non-linear impairments relevant for 61 wavelength routing decisions. 63 2. Conventions used in this document 65 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 66 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in 67 this document are to be interpreted as described in RFC-2119 [2]. 69 3. Introduction 71 Non-linear effects are due to the fact that the optical properties 72 of the medium (refractive index, loss, etc.) become dependent of the 73 signal power of the optical channels present in this medium. As a 74 consequence, this power dependency tends to modify the propagation 75 of the optical waves and also lead to interactions between these 76 waves. Non-linear interactions depend on the transmission length 77 (distance between the transmitter and the receiver), the type of 78 fiber, the cross-sectional area of the fiber, the wavelength and the 79 power level. Basically, these effects become more intensive when the 80 optical power or the transmission length increase or when the 81 channel spacing becomes narrower (the different wavelengths tend to 82 interact more each other). As a consequence, non-linearities can 83 impose significant limitations on high bit-rates (10 Gbit/s and 84 higher), Long Haul (LH) and Ultra-Long Haul (ULH) systems, or high 85 capacity DWDM systems. 87 Linear impairments are extensively addressed in [IPO-IMP] and 88 corresponding IGP routing protocol extensions in [IPO-ORI]. In these 89 previous works, the approach for non-linear impairments was to 90 consider that: �One could assume that non-linear impairments are 91 bounded and increase the required OSNR level by X dB, where X for 92 performance reasons would be limited to 1 or 2 dB, consequently 93 setting a limit on the maximum number of spans. For the approach 94 described here to be useful, it is desirable for this span limit to 95 be longer than that imposed by the constraints which can be treated 96 explicitly.� 98 However, this approximation may lead to both an over or an under 99 estimation of the real impact of non-linear effects. If the actual 100 impact is less than 2 dB on the OSNR, then, the margin taken will 101 forbid some feasible path(s) (as it limits the maximum number of 102 spans). On the contrary, if the real impact is over 2 dB, the 103 corresponding route(s) may be chosen despite they are not feasible 104 from the optical transmission point of view. 106 D.Papadimitriou et al. � Expires May 2002 2 107 Therefore, the objective of this document is first to determine 108 which kind of non-linear effects must be taken into account, and to 109 give some simple engineering rules to determine their maximum 110 tolerable value, second, to propose IGP routing protocol extensions 111 in order to cover non-linear optical routing impairments. 113 Additional complexity may arise from the fact that for instance when 114 minimizing degradations through Self Phase Modulation (SPM) after 115 setting a distinct Lambda LSP (L-LSP) or optical channel in the 116 network, this L-LSP will suffer a changing degradation by Cross- 117 Phase Modulation (XPM) through the changing number of concurrent 118 optical channels on the fiber links. Thus, as we will point out, 119 cross channels effects should be minimized at the system design in 120 order to be compatible with SPM and other impairments. 122 4. Non-Linear Impairments 124 Non-linear optical impairments can be classified into two 125 categories. The first category consists of effects occurring due to 126 the dependence of the refractive index on the optical signal power 127 (generally called Kerr effect). This category includes Self-Phase 128 Modulation (SPM), Cross-Phase Modulation (XPM) and Four-Wave Mixing 129 (FWM). The second category of effects consists of inelastic 130 scattering effects in the fiber medium and are due to the 131 interaction of the light waves with the optical phonons of the fiber 132 medium leading to Stimulated Raman Scattering (SRS) or with the 133 acoustic phonons (sound waves) of the medium leading to Stimulated 134 Brillouin Scattering (SBS). 136 SPM and XPM essentially affect the phase of the signals and cause 137 its spectral broadening which lead to temporal distorsions because 138 of dispersion. FWM lead to energy exchange between signals that 139 induces in-band crosstalk, whereas SBS and SRS provide gain or loss 140 to the light waves. Nevertheless, the actual impact of these non- 141 linear effects on transmission quality depends strongly on 142 dispersion management. 144 4.1 Refractive Index 146 The general equation for the refractive index of the core in an 147 optical fiber is given by: 149 n = n(0) + [n(2) x P / A(eff)] 151 where: 152 - n(0) = the refractive index of the fiber core at low optical 153 power level (no unit) 154 - n(2) = the non-linear refractive index coefficient (for 155 instance, 2.35 x 10^(-20) m^2/W for silica) 156 - P = optical signal power in Watts (W) 157 - A(eff) = the effective area of the core in square meters (m^2) 159 Clearly, this equation indicates that two strategies for minimizing 161 D.Papadimitriou et al. � Expires May 2002 3 162 non-linearities due to refractive index power dependence are to 163 minimize the launched optical power P and/or to maximize the 164 effective area of the fiber A(eff). 166 Minimizing P is limited by the fact that during network design, 167 there is a strong trade-off between non-linear effects and optical 168 signal to noise ratio (decreasing P will decrease non-linear effects 169 but also the OSNR). On the other hand, augmenting A(eff) is being 170 targeted by some fiber vendors while keeping other non-linear 171 effects unchanged. 173 This optical power dependence of the refractive index introduces the 174 following non-linear effects: SPM, XPM, FWM, SBS and SRS as 175 explained here below. 177 4.1.1 Self-Phase Modulation (SPM) 179 Self-Phase Modulation (SPM) arises from the power dependency of the 180 refractive index of the fiber core. Fluctuations in the optical 181 signal power cause changes in the phase of the signal referred to as 182 a non-linear phase shift. This induces an additional frequency chirp 183 on the spectrum of the optical pulse which interacts with the 184 fiber�s dispersion to broaden the pulse and lead to intensity 185 fluctuations. Therefore, this effect leads to higher penalties due 186 to Inter-Symbol Interference (ISI). This chirping effect affects 187 each channel independently of the other and is proportional to the 188 optical channel power. Therefore SPM effects are more pronounced in 189 systems using higher transmitted signal power. Moreover because the 190 SPM effect leads to extra ISI, higher bit-rate systems will be more 191 affected. 193 It is important to point out that in a DWDM transmission system at 194 10 Gbit/s with 100 GHz channel spacing, SPM is generally considered 195 as a significant non-linear effect (except may be when low 196 dispersion fibers are used). 198 4.1.2 Cross-Phase Modulation (XPM) 200 For a given optical signal, Cross-Phase Modulation (XPM) is a 201 consequence of a modification of the refractive index of the medium 202 due to the optical power of the closest neighboring channels present 203 in the fiber. As for SMP, the induced phase shift lead to intensity 204 fluctuations after interaction with dispersion. XPM increases when 205 optical channel spacing becomes narrower as long as the adjacent 206 channels are closer in the spectral domain, so that they travel 207 roughly at the same velocity and interact over a longer time period. 208 As XPM effect depends on channel spacing, it can be a significant 209 problem for high capacity DWDM system with channel spacing of 50 GHz 210 or lower. 212 On the other hand, when the bit rate increases (from 10 Gbit/s to 213 40 Gbit/s), the impact of XPM decrease as long as the time during 214 which the interacting channels temporally overlap is considerably 216 D.Papadimitriou et al. � Expires May 2002 4 217 reduced. 219 For moderate bit rate systems (10 Gbit/s), XPM effect generally 220 becomes significant compared to SPM when channel spacing is lower 221 than 100 GHz, and when the local dispersion is low. This means that 222 XPM should be taken into account for 50 GHz spacing DWDM systems. 223 Here, we want to point out that solutions demonstrated in laboratory 224 environments may be implemented to decrease the effect of XPM by 225 introducing a phase-mismatch between optical channels at the link 226 input by using interleaved polarization or using a suitable 227 dispersion management. 229 4.1.3 Four-Wave Mixing (FWM) 231 In a (high capacity) DWDM system, based on different optical 232 channels at different wavelengths (i.e. frequencies), the power 233 dependence of the refractive index of the fiber core also gives rise 234 to the generation of new frequencies (i.e. new optical signal). 236 Practically, there is an interaction between the different channels, 237 leading to energy transfer between these channels. This effect is 238 called Four-Wave Mixing (FWM) because if three optical channels with 239 frequencies f1, f2 and f3 propagates simultaneously within the same 240 fiber, a fourth optical channel is generated and frequency f4 which 241 is related to the other frequencies by the following relation: f4 = 242 f1 (+ or -) f2 (+ or -) f3. In theory, several frequencies 243 corresponding to different combinations are possible. However, in 244 practice only the frequency combinations of the form f4 = f1 + f2 � 245 f3 are the most troublesome for (high capacity) DWDM systems. These 246 fourth optical channels can become even nearly phase-matched when 247 optical channel wavelengths are close to the zero-dispersion point. 249 As a consequence, significant optical power can be transferred 250 between neighboring optical channels through the FWM effect. Though, 251 in contrast to SPM and XPM, which are bit-rate dependent, the FWM 252 effect is not really dependent of the bit-rate. Nevertheless, like 253 XPM, it depends strongly on the optical channels spacing and the 254 fiber dispersion. Clearly, FWM becomes significant only at narrow 255 channel spacing (50 GHz or lower) or when the local dispersion is 256 low. 258 As a consequence, significant optical power can be transferred 259 between neighboring optical channels through the FWM effect. Though, 260 in contrast to SPM and XPM, which are bit-rate dependent, the FWM 261 effect is not really dependent of the bit-rate. Nevertheless, like 262 XPM, it depends strongly on the optical channels spacing and the 263 fiber dispersion. Clearly, FWM becomes significant and should be 264 taken into account for DWDM systems with narrow channel spacing (50 265 GHz or lower) or when the local dispersion is low. 267 4.2 Scattering Effects 269 Scattering effects, the second set of mechanisms generating non- 271 D.Papadimitriou et al. � Expires May 2002 5 272 linearities give rise to SRS and SBS. 274 4.2.1 Stimulated Raman Scattering (SRS) 276 In DWDM systems, the fiber acts as a Raman amplifier such that the 277 longer wavelengths channel are amplified by the shorter wavelengths 278 channels as long as the wavelength difference is within the Raman 279 Gain spectrum. 281 Therefore, if two or more signals at different wavelengths are 282 injected into a fiber, the SRS effect causes optical signal power to 283 be transferred from the lower wavelength optical channels to the 284 higher wavelength optical channels. The gain coefficient increases 285 with increasing channel spacing up to 125 nm. This amplification 286 leads to increase power fluctuations, which add to receiver and 287 degrade receiver performance. This phenomenon known as Raman inter- 288 channel crosstalk can be avoided if channel powers are made so small 289 that Raman amplification is negligible over the fiber length. 291 However, the coupling between wavelengths occurs only if both 292 optical channels are launched simultaneously so that the impact of 293 the SRS is reduced by the dispersion introduced by the silica 294 medium. Basically, SRS induces a gain tilt over the whole bandwidth 295 of the fiber, which is proportional to the total power of all 296 channels present in the fiber. 298 Moreover, periodic amplification of the DWDM signal in ULH fiber 299 links can also increase the impact of the SRS-induced degradation. 300 This phenomenon occurs because in-line amplifiers add noise which 301 experiences less Raman loss than the signal itself, resulting in 302 degradation of the SNR. In brief, the total capacity of DWDM systems 303 is then limited to below 100 Gbps for a transmission distance of 304 5000 km or more. 306 As a matter of fact, SRS should not be considered for impairment 307 based optical routing, as long as its induced Raman tilt will be 308 managed link by link during the network design. 310 4.2.2 Stimulated Brillouin Scattering (SBS) 312 Scattering effects in the optical fiber occur due to the interaction 313 of the optical channels with the sound waves (acoustic phonons) 314 present in the silica medium. In SBS, the scattering process is 315 stimulated by photons with a wavelength higher than the wavelength 316 of the incident signal. This interaction takes place over a very 317 narrow band of 20 MHz at 1550nm. The scattered waves and the 318 incident optical light waves propagate in opposite directions. Thus, 319 SBS produces an additional loss in the propagating signal but does 320 not induce any interaction between different optical channels. 322 In practical implemented systems, the Brillouin inter-channel 323 crosstalk phenomenon can be easily avoided by always keeping the SBS 324 threshold power higher than the optical signal power. Moreover, the 326 D.Papadimitriou et al. � Expires May 2002 6 327 probability for SRS to occur is much higher than that for SBS 328 because the gain bandwidth for SRS is ~5 THz, while the gain 329 bandwidth for SBS is ~0.05 GHz. Consequently, the losses induced by 330 the SBS effect are not a real problem when considering impairment- 331 based optical routing. 333 5. Fiber and Optical Amplifiers 335 In the course of moving from pure optical centrally managed 336 transmission to flexible wavelength switched networks many problems 337 have to be solved. One of these problems is the fiber diversity 338 among the different links, and the impact of non-linear effects 339 inside the different transmission fibers. 341 5.1 Influence of the fiber medium 343 An optical transparent network is composed of many nodes (optical 344 LSR) connected by links (a link is a transmission system between two 345 nodes). Each link is composed of transmission spans of identical 346 fiber, whereas in the most general case, different types of fibers 347 may be deployed among the different links. The main types of fibers 348 that are generally deployed in today optical networks are mainly 349 based on G.652 as well as G.655 and G.653 ITU-T Recommendations. 350 Also and Dispersion Compensating Fibers (DCF) used to compensate the 351 dispersion is present inside the network element. 353 The intensity of non-linear effects is also dependent on the 354 intrinsic optical properties of each fiber, thus, fiber diversity in 355 optical networks must be taken into account when regarding the non- 356 linear impairments. In particular, non-linear effects arise not only 357 in the transmission fiber but also inside the Dispersion 358 Compensating Fiber (DCF) present in the network elements. 360 An additional important point is that in today optical networks, the 361 dispersion of the transmission fiber is compensated for each link by 362 a specific dispersion map. The residual dispersion after cascading a 363 certain number of links must be compatible with the cumulated impact 364 of non-linear effect such as SPM or XPM. 366 In most common optical networks, of moderate bit-rate (10 Gbit/s) 367 and channel spacing (100 GHz), Self-phase modulation (SPM) is the 368 strongest non-linear degradation effect. Changing the path length in 369 the network will then increase the SPM contribution when it 370 increases the total path length, or when the optical path crosses 371 highly non-linear fiber links. 373 Only for highly dense DWDM systems (channel spacing of 50 GHz or 374 less), XPM at its turn will take more and more importance, as well 375 as FWM (even if this latter effect can be reduced with specific 376 design in the link). In that case, after setting a distinct path in 377 the network, this path may suffer a changing degradation by XPM and 378 eventually FWM through the changing number of concurrent channels on 379 a same fiber link. Thus, XPM and FWM must be minimized at system 381 D.Papadimitriou et al. � Expires May 2002 7 382 design but even then their impact must be taken into account for 383 impairment-based optical routing. 385 5.2 Optical Amplifiers 387 Potentially, non-linear effects may also occur in the fiber 388 amplifiers, but they can be considerably reduced provided a specific 389 design will be done. Then, it is not worth to take them into 390 account, as long as they are kept under control at system design. 392 6. Non-linear Phase 394 The intensity of the SPM, XPM and FWM non-linear effects can be 395 quantified through the Non-Linear Phase shift NLP (see [AGR-FOCS]) 396 induced in the fiber by the Kerr effect. As a matter of fact, the 397 NLP was shown to be a robust empiric parameter able to evaluate the 398 impact of non-linear effects as described in [OFC00-NLP] and [OFC02- 399 NLP] while related considerations can be found in [ELEC-ODS]. 401 In this section we propose to use the Non-Linear Phase (NLP) as an 402 empiric criterion to correlate the cumulated effects of SPM, XPM and 403 FWM with a given penalty. This penalty corresponds to an upper bound 404 value of the NLP (NLPmax) which depends on the bit-rate, the channel 405 spacing and the fiber type. 407 Since the NLP is additive along an optical path (including several 408 links and spans), the cumulated NLP value (NLPcum) can be compared 409 to the maximum tolerated value of the NLP (NLPmax). Consequently, 410 this method ensures that an optical channel is not affected by non- 411 linear effects when NLPcum < NLPmax. 413 6.1 Definition 415 The Non-Linear Phase (NLP) for a given transmission span NLP(span) 416 is given by the following formula (see [AGR-NFO]): 418 NLP(span) = P(in) x F(span) 420 where: 421 - P(in) = optical power in Watts (W) at the span input 422 - F(span) = function assumed to be constant for a given span. 424 The F(span) function is directly proportional to: 426 F(span) ~ [n(2) x L(eff)] / [w x A(eff)] 428 where: 429 - n(2) = non-linear refractive index coefficient (m^2/W) 430 - L(eff) = effective interaction length (m) 431 - w = wavelength of the optical channel (m) 432 - A(eff) = effective area of the fiber core in square meters (m^2) 434 While the effective interaction length L(eff) is defined as: 436 D.Papadimitriou et al. � Expires May 2002 8 437 L(eff) = [1 � exp(-aL)] / a 439 where: 440 - a = linear absorption coefficient of the fiber (m^-1) 441 - L = fiber length (m) 443 It is important to point out that the function F is defined as an 444 integral of a rational function, which can be easily calculated and 445 leads to an analytical formula. Then, for each span, the NLP can be 446 easily computed using the above coefficients: w, n(2), A(eff) and 447 L(eff). 449 For a transmission span, one must take into account the NLP due to 450 the transmission fiber and Dispersion Compensating Fiber (DCF). 451 Therefore, the total NLP for a whole span (NLP(span)) is then given 452 by the following formula: 454 NLP(span) = NLP(fiber) + NLP(DCF) = P(in) x F(span) 456 where: 457 - P(in) = optical power at the span input 458 - F(span) = global function containing the parameters of the span 459 including transmission fiber and inline DCF 461 Notice that the only varying parameter in the formula described in 462 this section is the optical power at the span input P(in). 464 6.1.1 Calculation of the NLP for a link 466 A link between two optical LSR is constituted by N transmission 467 spans. Then, the cumulated NLP for a given link (NLP(link)) is equal 468 to the sum of the different NLP due to each span and is given by: 470 NLP(link) = Sum(NLP[i]) = Sum(P[i] x F[i]) 472 where: 473 - NLP[i] = NLP due to span �i� 474 - P[i] = power at the input of span �i� 475 - F[i] = F function of span �i� 477 This formula allows the calculation of the cumulated NLP for any 478 link, provided one knows the optical power at each span input, and 479 the F function for each span. 481 An interesting case is when the N spans (including the same fiber 482 type) can be considered as roughly identical, so that we can 483 simplify the above formula to: 485 NLP(link) = N x NLP(span) = N x P x F(span) 487 where: 488 - NLP(span) = NLP of the spans 490 D.Papadimitriou et al. � Expires May 2002 9 491 - P = power at the input of the link 492 - F(span) = F function of the spans 494 It is important to note that the NLP(link) is a static information 495 which characterizes the link, and is calculated locally. Therefore, 496 it is the only required parameter to be flooded by IGP protocols. 498 6.1.2 Calculation of the NLP for an optical path 500 Considering the establishment of an optical path within a network, 501 this path will be a succession of �j� different links, each link 502 being composed of a specific fiber type. For instance, consider an 503 optical path going from ingress node S to egress node D, via node A 504 and B where the link between S and A is includes SMF fiber, the link 505 between A and B, E-Leaf fiber and the link between B and D, True- 506 Wave Fiber. 508 Then, the total cumulated NLP over the whole optical path NLP(path) 509 is given by: 511 NLP(path) = Sum(NLP(link)[j]) 513 where NLP(link)[j] is the NLP due to link �j�. 515 With this simple formula, it is possible to calculate the total 516 cumulated NLP (referred to as NLPcum) for any optical path, using 517 the previously calculated NLP of the different links. 519 6.2 NLP Constraint 521 For a given optical path, the total cumulated NLP due to SPM, XPM 522 and FWM non-linear effects can be computed according to the previous 523 formula. 525 In Section 6.1, we have demonstrated that the NLP is an additive 526 variable along an optical path (including several links and spans) 527 which depends on the bit-rate, the channel spacing and the fiber 528 type. Therefore, the cumulated NLP value (NLP(path)) for a given 529 optical path can be compared to the maximum tolerated value of the 530 NLP (NLPmax). The latter is used as empiric criterion to correlate 531 the SPM, XPM and FWM non-linear effects leading to the NLPmax upper 532 bound value of the NLP. 534 Consequently, the non-linear optical routing cumulative constraint 535 including the SPM, XPM and FWM effects can be expressed as follow: 536 for a given residual dispersion after crossing an optical path, the 537 total cumulated dispersion NLP(path) must be lower than NLP(max), 538 the maximum tolerable value for the NLP: 540 NLP(path) < NLPmax 542 D.Papadimitriou et al. � Expires May 2002 10 543 When the NLP(path) fulfills this constraint, the corresponding 544 optical channel is not limited by the SPM, XPM and FWM non-linear 545 effects do not limit a given optical system. 547 For example, simulations have shown that the non-linear constraint 548 NLPmax can be expressed as follows (assuming an accurate dispersion 549 compensation management): 551 - NLPmax < 0.45 pi at 10Gbit/s 552 - NLPmax < 0.3 pi at 40Gbit/s 554 It is important to point out that we assume in this approach that 555 the dispersion is managed at the link level using available 556 technology being developed, so that for each link, the residual 557 dispersion is compatible with the NLP of the link. 559 Moreover, in a high density DWDM system, the NLP shift per span for 560 a given optical channel does not only depend on the optical power of 561 that channel and the fiber type. The NLP depends also on the power 562 of the closest neighboring optical channels typically (8 in 563 practical applications), as well as on the channel spacing. This 564 implies that one have to consider the NLPmax constraint with respect 565 to the channel spacing. Consequently, the NLP value per link 566 (NLP(link)) must be flooded by the IGP routing protocol to take the 567 channel spacing effect into account. 569 7. Traffic-Engineering Routing Protocol Extension 571 As mentioned here above, the NLP parameter must be flooded per 572 optical channel spacing (i.e. 100 GHz, 50 GHz and 25 GHz) using a 573 dedicated extension to the IGP TE-Routing protocol. 575 In OSPF, these NLP parameters are included in a common sub-TLV of 576 the Link TLV in the Traffic Engineering LSA. The Type value of this 577 sub-TLV is to be attributed (TBA). The length of this sub-TLV is 12 578 octets and the corresponding value specifies the NLP value (in IEEE 579 floating point format) per channel spacing. The format of the NLP 580 sub-TLV is as shown: 582 0 1 2 3 583 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 584 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 585 | Type = TBA | Length = 12 | 586 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 587 | NLP at 100GHz | 588 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 589 | NLP at 50GHz | 590 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 591 | NLP at 25GHz | 592 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 594 D.Papadimitriou et al. � Expires May 2002 11 595 In IS-IS, we propose to enhance the sub-TLVs for the extended IS-IS 596 reachability TLV. The length of the NLP sub-TLV is 12 octets and 597 specifies the NLP value (in IEEE floating point format) per channel 598 spacing (in IEEE floating point format). Specifically, we add the 599 following sub-TLV: 601 - Sub-TLV type: TBA 602 - Length(in bytes): 12 603 - Name: NLP 605 8. Security Considerations 607 There are no additional security considerations than the ones 608 already covered in OSPF and IS-IS. 610 9. Reference 612 1. Bradner, S., "The Internet Standards Process -- Revision 3", BCP 613 9, RFC 2026, October 1996. 615 2. Bradner, S., "Key words for use in RFCs to Indicate Requirement 616 Levels", BCP 14, RFC 2119, March 1997 618 3. [AGR-NFO] Govind P. Agrawal, �Nonlinear Fiber Optics�, (section 619 2.6.2: �Nonlinear refraction�), Academic Press, 1995. 621 4. [AGR-FOCS] Govind P. Agrawal, �Fiber-Optic Communication 622 Systems�, Second Edition, Wiley Series in Microwave and Optical 623 Engineering, March 1997. 625 5. [ELEC-ODS] A. F�rbet et al., �Optimised dispersion scheme for 626 long-haul optical communication systems�, Electronic Letters 14, 627 October 1999, Vol.35, No.21. 629 6. [GYS-XT] T. Gyselings, �Investigation and Reduction of CrossTalk 630 in Wavelength Division Multiplexed All-Optical Cross-Connects�, 631 PhD Thesis, INTEC, Universiteit Gent. 633 7. [IPO-IMP] A. Chiu et al., �Impairments And Other Constraints On 634 Optical Layer Routing�, Internet Draft, Work in progress, draft- 635 ietf-ipo-impairments-00.txt, May 2001. 637 8. [IPO-ORI] A. Banerjee et al., �Impairment Constraints for Routing 638 in All-Optical Networks�, Internet Draft, Work in progress, 639 draft-banerjee-routing-impairments-00.txt, May 2001. 641 9. [OFC02-NLP] J.-C. Antona et al. �Nonlinear cumulated phase as a 642 criterion to assess performance of terrestrial WDM systems�, 643 Technical paper submitted to OFC�02. 645 10. [OFC00-NLP] Y. Frignac and S. Bigo, �Numerical optimization of 646 residual dispersion in dispersion-managed systems at 40 Gbit/s�, 647 Paper TuD3, OFC�00, Baltimore. 649 D.Papadimitriou et al. � Expires May 2002 12 650 10. Acknowledgments 652 The authors would like to thank B. Sales, E. Desmet, J.C. Antona, S. 653 Bigo and A. Jourdan for their constructive comments and inputs. 655 11. Author's Addresses 657 Dimitri Papadimitriou 658 Alcatel 659 Francis Wellesplein 1, 660 B-2018 Antwerpen, Belgium 661 Phone: +32 3 240-8491 662 Email: dimitri.papadimitriou@alcatel.be 664 Jean-Paul Faure 665 Alcatel 666 Route de Nozay 667 91461 Marcoussis Cedex, France 668 Phone: +33 1 6963-1307 669 Email: jean-paul.faure@ms.alcatel.fr 671 Olivier Audouin 672 Alcatel 673 Route de Nozay 674 91461 Marcoussis Cedex, France 675 Phone: +33 1 6963-2365 676 Email: olivier.audouin@ms.alcatel.fr 678 Roy Appelman 679 Civcom 680 Phone: +1 972 3 922-9229 681 Email: roy.a@civcom.com 683 D.Papadimitriou et al. � Expires May 2002 13 684 Full Copyright Statement 686 "Copyright (C) The Internet Society (date). All Rights Reserved. 687 This document and translations of it may be copied and furnished to 688 others, and derivative works that comment on or otherwise explain it 689 or assist in its implementation may be prepared, copied, published 690 and distributed, in whole or in part, without restriction of any 691 kind, provided that the above copyright notice and this paragraph 692 are included on all such copies and derivative works. However, this 693 document itself may not be modified in any way, such as by removing 694 the copyright notice or references to the Internet Society or other 695 Internet organizations, except as needed for the purpose of 696 developing Internet standards in which case the procedures for 697 copyrights defined in the Internet Standards process must be 698 followed, or as required to translate it into languages other than 699 English. 701 The limited permissions granted above are perpetual and will not be 702 revoked by the Internet Society or its successors or assigns. 704 This document and the information contained herein is provided on an 705 "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING 706 TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING 707 BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION 708 HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF 709 MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE." 711 D.Papadimitriou et al. � Expires May 2002 14