idnits 2.17.1 draft-multiple-layer-resource-optimization-00.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack a Security Considerations section. ** 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.) Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document doesn't use any RFC 2119 keywords, yet seems to have RFC 2119 boilerplate text. -- The document date (November 5, 2018) is 1999 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- -- Looks like a reference, but probably isn't: '1' on line 148 -- Looks like a reference, but probably isn't: '2' on line 91 -- Looks like a reference, but probably isn't: '3' on line 103 -- Looks like a reference, but probably isn't: '0' on line 167 == Missing Reference: 'Step' is mentioned on line 167, but not defined -- Looks like a reference, but probably isn't: '4' on line 279 == Unused Reference: 'Ref1' is defined on line 405, but no explicit reference was found in the text == Unused Reference: 'Ref2' is defined on line 409, but no explicit reference was found in the text == Unused Reference: 'Ref3' is defined on line 413, but no explicit reference was found in the text == Unused Reference: 'Ref4' is defined on line 417, but no explicit reference was found in the text Summary: 2 errors (**), 0 flaws (~~), 7 warnings (==), 6 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Cross Stratum Optimization Research Group H. Yang 3 Internet-Draft K. Zhan 4 Intended status: Informational A. Yu 5 Expires: May 9, 2019 Q. Yao 6 J. Zhang 7 Beijing University of Posts and Telecommunications 8 November 5, 2018 10 Multiple Layer Resource Optimization for Optical as a Service 11 draft-multiple-layer-resource-optimization-00 13 Abstract 15 Human activities are highly predictable, so are the network traffic. 16 It shows a tidal phenomenon in traffic. Current resource allocation 17 methods cannot adapt to tidal traffic, resulting in low resource 18 utilization. To predicted tidal traffic, we have established a 19 neural network model optimized by adaptive artificial fish swarm 20 algorithm. Then we propose a novel multi-path pre-reserved resource 21 allocation strategy to increase resource utilization. The results 22 prove the effectiveness of our method. 24 Status of This Memo 26 This Internet-Draft is submitted in full conformance with the 27 provisions of BCP 78 and BCP 79. 29 Internet-Drafts are working documents of the Internet Engineering 30 Task Force (IETF). Note that other groups may also distribute 31 working documents as Internet-Drafts. The list of current Internet- 32 Drafts is at https://datatracker.ietf.org/drafts/current/. 34 Internet-Drafts are draft documents valid for a maximum of six months 35 and may be updated, replaced, or obsoleted by other documents at any 36 time. It is inappropriate to use Internet-Drafts as reference 37 material or to cite them other than as "work in progress." 39 This Internet-Draft will expire on May 9, 2019. 41 Copyright Notice 43 Copyright (c) 2018 IETF Trust and the persons identified as the 44 document authors. All rights reserved. 46 This document is subject to BCP 78 and the IETF Trust's Legal 47 Provisions Relating to IETF Documents 48 (https://trustee.ietf.org/license-info) in effect on the date of 49 publication of this document. Please review these documents 50 carefully, as they describe your rights and restrictions with respect 51 to this document. Code Components extracted from this document must 52 include Simplified BSD License text as described in Section 4.e of 53 the Trust Legal Provisions and are provided without warranty as 54 described in the Simplified BSD License. 56 Table of Contents 58 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 59 1.1. Conventions Used in This Document . . . . . . . . . . . . 3 60 2. PREDICTION STRATEGY . . . . . . . . . . . . . . . . . . . . . 3 61 2.1. Artificial neural network model . . . . . . . . . . . . . 3 62 2.2. Adaptive artificial fish swarm artificial neural networks 63 (AAFS-ANN ) . . . . . . . . . . . . . . . . . . . . . . . 4 64 3. MULTI-PATH PRE-RESERVED RESOURCE ALLOCATION . . . . . . . . . 5 65 3.1. Reconfiguration time calculation . . . . . . . . . . . . 6 66 3.2. Multi-path pre-reserved resource allocation(MP-RA) . . . 6 67 4. Experimental evaluation and results analysis . . . . . . . . 7 68 5. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . 9 69 6. ACKNOWLEDGMENT . . . . . . . . . . . . . . . . . . . . . . . 9 70 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 9 71 7.1. Normative References . . . . . . . . . . . . . . . . . . 9 72 7.2. Informative References . . . . . . . . . . . . . . . . . 9 73 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 10 75 1. Introduction 77 With the rapid growth of cloud computing, 5G services, and the 78 periodicity of people's activities, traffic load has exhibited 79 periodicity in both time and space domains, namely tidal traffic [1]. 80 The number of people using optical metropolitan networks is enormous 81 and unevenly distributed. In addiction, the separation of work areas 82 and residential areas is an important cause of tidal traffic. 83 Generally, tidal traffic will reduce the performance of networks 84 during to following two reasons: firstly, the network traffic will be 85 blocked due to the sharp increase in traffic in the high-traffic 86 area; secondly, network nodes may be idle and waste resources in the 87 low-traffic areas. The static configuration resources will intensify 88 both network and service congestion during traffic peak hours, as 89 well as low resource utilization during low-traffic times and 90 regions. In the future, global mobile Internet traffic will increase 91 by 10 times [2], urbanization is rapidly advancing, the scope and 92 severity of space and time domains affected by tidal traffic are 93 increasing as communication need and network technologies developing. 94 Tidal traffic will further affect the optical access network and the 95 optical core network, making it essential issue for network 96 operators. Therefore, a more reasonable and efficient resource 97 allocation scheme is urgently needed to solve the congestion and 98 resource waste caused by the tidal traffic. 100 Known from the above, tidal traffic prediction becomes the core 101 process of network optimization decision-making. Currently, there 102 are several prediction methods, like support vector machine (SVM) and 103 multi-layer perceptron (MLP). Literature [3] proposes a deep- 104 learning-based prediction strategy to implement traffic assessment of 105 data center optical networks. At the same time, a deep-learning- 106 based global evaluation factor resource allocation algorithm is 107 suggested to achieve lower blocking rate of the network. Compared 108 with the traditional algorithm, deep learning can improve the 109 accuracy of prediction, but it cannot identify the tidal traffic in 110 specific festivals. In addition, the lower priority service will be 111 discarded to reduce the network blocking rate. This method does not 112 make good use of idle resources of other nodes, and some traffic 113 requests cannot be executed normally. So we propose multi-path pre- 114 reserved resource allocation based on traffic prediction. 116 In this paper, we establish an adaptive artificial fish-group neural 117 network model to predict traffic, then use the predicted traffic 118 demand to optimize the network at different times. Meanwhile, we 119 propose multi-path pre-reserved resource allocation to adapt to the 120 resource requirements of different nodes. Simulation results 121 demonstrate that our strategy achieves a lower network blocking rate 122 and higher resource utilization. 124 1.1. Conventions Used in This Document 126 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 127 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 128 document are to be interpreted as described in [RFC2119]. 130 2. PREDICTION STRATEGY 132 Before presenting the resource allocation algorithm, we provide an 133 introduction to traffic prediction model. We establish a neural 134 network of adaptive artificial fish algorithm to predict traffic 135 request. The key resides in the construction of the artificial fish 136 individual model. The optimal variables of the neural network are 137 two weight matrices and two threshold variables _io,v_o . 139 2.1. Artificial neural network model 141 We build the neural network structure as shown in figure 1. The 142 input is composed of six entries, i_(s,1) is the hour of the day, 143 i_(s,2) is the day of the week, i_(s,3) is a flag for holiday/ 144 weekend, i_(s,4) is the previous days average load, i_(s,5) is the 145 load from the same hour of the previous day, and i_(s,6) is the load 146 from the same hour and same day from the previous week. The result 147 of the output node Y_(s,1) represents the traffic request that we 148 want to predict[1]. Training sample setA={(X^i,Y^i )|i=1,2,,n}X^i is 149 the i_th group training data input, and Y^i is the i_th group input 150 corresponding expected output. We define the error function as 151 follows: 153 where O^i is the actual output of the i_th. 155 2.2. Adaptive artificial fish swarm artificial neural networks (AAFS- 156 ANN ) 158 In the artificial fish swarm algorithm, we introduce adaptive step 159 size and visible range to improve convergence accuracy and speed. 160 Generate initial artificial fish population N, namely N group 161 {omega_ij,nu_io,omega_io,nu_o}. Every artificial fish is a neural 162 network. The food concentration is defined as FC=1/E. X_i is the 163 state of current location state,X_j is random state of the 164 search,d_ij is the distance between X_i and X_j, omega_ij 165 (i),omega_ij (j) and omega_ij (i+1) respectively are X_i,X_jnext 166 state X_(i+1) matrix omega_ij} element of i_th row j_th column, 167 "Rand(Step") represents a random number between [0, Step]. 169 Let X_0 be the current artificial fish, its position is C, X_1 is the 170 current optimal fish, X_2 is the nearest fish, Then we set two 171 visible fields viusual_1=d_01,viusual_2=d_02. Two target positions 172 A, B are randomly determined in the range of viusual_1 and viusual_2 173 respectively, then compare FC_A,FC_B,FC_C, 175 If FC_A,FC_B 177 If FC_A,FC_B 179 omega(i+1)=omega(i)+Rand(step) 181 If one or both of them are better than C, Then advance to the best 182 point, and execute formula (3) 184 omega(i+1)=omega(i)+Rand(step)(omega(j)-omega(i))/d_ij 186 Go for A with viusual_1Rand() as the step size, to B with 187 viusual_2alphaRand(), where a ,which equal to 1 or slightly less than 188 1,is the visual factor.The other three optimization variables are 189 similarly. 191 3. MULTI-PATH PRE-RESERVED RESOURCE ALLOCATION 193 The resource allocation method bases on the AAFS-ANN described above, 194 and we propose a multi-path pre-reserved resource allocation way to 195 optimize optical network. We uses the predicted result to perform 196 configuration time calculation and estimate the future network 197 resource demand to pre-reserve resource for traffic request. 199 ------------------------------------- 200 | --- --- | 201 | | A |--------------| B | | 202 | --- \ --- \ | 203 | \ \ | 204 | \ \ | 205 | \ \ | 206 | --- --- | 207 | | C |---------------| D | | 208 | --- --- | 209 ------------------------------------- 211 Fig.1(a) Sample network 213 ^ 214 | 215 |------------------------------------ 216 T4 | | | | | | | | | | 217 |------------------------------------ 218 T3 | | $ | $ | | | | | | | 219 |------------------------------------ 220 T2 | | $ | $ | | | | | | | 221 |------------------------------------ 222 T1 | | | | | | | | | | 223 |------------------------------------ 224 T0 | | | | | | | | | | 225 ----------------------------------------> 226 S0 S1 S2 S3 S4 228 Fig.1(b) Requested resources 230 ^ 231 | 232 |------------------------------------ 233 T4 | | | | | | | | | | 234 |------------------------------------ 235 T3 | # | $ | # | # | # | | | | | 236 |------------------------------------ 237 T2 | # | $ | # | # | # | | | | | 238 |------------------------------------ 239 T1 | # | | | # | # | | | | | 240 |------------------------------------ 241 T0 | # | | | # | # | | | | | 242 ----------------------------------------> 243 S0 S1 S2 S3 S4 245 Fig.1(c) Requested resources 247 ^ 248 | 249 |------------------------------------ 250 T4 | | | | | | | | | | 251 |------------------------------------ 252 T3 | # | # | $ | | # | # | | | | 253 |------------------------------------ 254 T2 | # | # | $ | # | # | # | | | | 255 |------------------------------------ 256 T1 | # | | | # | | | | | | 257 |------------------------------------ 258 T0 | # | | | | | | | | | 259 ----------------------------------------> 260 S0 S1 S2 S3 S4 262 Fig.1(d) Requested resources 264 3.1. Reconfiguration time calculation 266 Frequent reconfiguration can result in service interruption and 267 unstable of distributed routing algorithm, so we need to predict the 268 next 24-hour traffic demand D^24 for the next configuration time 269 point calculation. Algorithm 1 is the calculation process of the 270 reconfiguration time point. 272 3.2. Multi-path pre-reserved resource allocation(MP-RA) 274 We reserve network resources for the predicted traffic. This type of 275 service request is called Advance Reservation Service (AR). As the 276 optical link is continuously established and removed, fragments are 277 easily generated in both the time domain and the spectrum domain. 278 The application of the Sliceable Bandwidth Variable Transceiver 279 (S-BVT) [4] further enhances the flexibility of EON. The S-BVT has a 280 slicing capability, i.e. it can provide multiple optical carriers for 281 carrying optical links to different destinations. In order to reduce 282 time and spectral fragmentation (referred to as two-dimensional 283 fragmentation) and to solve the problem of insufficient resources, we 284 propose cutting the request into multiple parts, and transfer on 285 multiple paths. 287 The underlying optical network can be modeled as 288 G_s=(L_s,N_s,R_st,D_s){ L_s: link set, N_s: optical node set, R_st: 289 resource status of optical nodes and optical links at time t, D_s: 290 distance of each pair of nodes in the set of nodes N in the network 291 topology}. R_A=(s,d,w,b,h) denotes a predicted service request, where 292 s and d represent the source and destination nodes of the service, b 293 is the time of service starts, h is the duration of the AR service, 294 and w is the service start time b, and the duration h period required 295 link rate. P_((s,d)) represents the path set of the source node to 296 the destination node. 298 If there are not enough spectrum resources available in the link for 299 the incoming request, we will attempt to cut the request into 300 different parts and assign those parts to different frequency bands. 301 For a simple example, as shown in Firgue 3(a), in order to reflect 302 the state of the spectral resources in the time domain, we use a two- 303 dimensional time spectrum resource model and assume that each time 304 slot has the same time period. The network diagram is illustrated in 305 the figure 3(a). Now there was an AR request, from node A to node D. 306 The request requires two spectrum slots, lasting from T2 to T3, as 307 shown in figure 2(b). Figure 2(c) and Figure 2(d) show the spectrum 308 states of path A-C-D and path A-B-D, respectively. The black slot 309 represents the occupied spectrum slot, the white slot represents the 310 spectrum slot available for the spectrum resource, and the blue slot 311 represents the spectrum slot occupied by the AR request. Before 312 splitting the AR request, the two paths do not have enough resources 313 to allocate. However, after we split the request into two parts, we 314 can distribute them to two spectrum segments to implement AR- 315 requested service provision. The MP-RA is as shown in Algorithm 2. 317 4. Experimental evaluation and results analysis 319 In this paper, we present the results of the AAFS-ANN prediction. 320 Our goal is to demonstrate the accuracy and network performance of 321 AAFS-ANN in different network environments. To fully reflect the 322 changes in the network environment, we use WIDE data from 96h traffic 323 data from April 6th to 9th, 2017, to train and verify. Figure 3(a) 324 is the comparison between the actual traffic and the prediction 325 results, which verify the effectiveness of our method. As shown in 326 Figure 3(a), the prediction results of AAFS-ANN are significantly 327 better than the traditional predictions. This is because the 328 introduction of the adaptive step size and the visible field, making 329 the artificial fish compares the FC in the large field of view. Our 330 method enhances the global convergence and the optimization 331 precision. The prediction error occurs because the traffic is 332 directly affected by many non-linear sudden factors such as hot 333 events, user movement patterns. Therefore, many traffic cannot be 334 accurately predicted. 336 We also compare MP-RA with several state-of-the-art resource 337 allocation techniques including evolutionary algorithms(EA) and 338 artificial neural networks (ANN). From firgue 3(b), we can see that 339 MP-RA performs well among the three optimization resource allocation 340 method, MP-RA greatly improves resource utilization. 342 According to the prediction results, the MP-RA can allocate resources 343 to traffic more reasonably. This is because the algorithm considers 344 the traffic that will be reached at each point in time and the 345 resources it needs. Then re-plans the resources at the configuration 346 time. As can be observed in the results shown in Figure 3(c), MP-RA 347 can greatly reduce the probability of traffic blocking. 349 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 350 | | Resource utilization rate | 351 | Traffic load +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 352 | | MP-RA | ANN | EA | 353 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 354 | 40 | 0.254 | 0.242 | 0.251 | 355 | 70 | 0.263 | 0.253 | 0.272 | 356 | 95 | 0.273 | 0.275 | 0.300 | 357 | 120 | 0.332 | 0.29 | 0.420 | 358 | 145 | 0.389 | 0.325 | 0.504 | 359 | 170 | 0.457 | 0.356 | 0.583 | 360 | 200 | 0.52 | 0.403 | 0.723 | 361 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 363 Tab.1 Network blocking probability of four strategies 364 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 365 | | Network blocking probability | 366 | Traffic load +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 367 | | MP-RA | ANN | EA | 368 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 369 | 50 | 0.008 | 0.0075 | 0.0078 | 370 | 70 | 0.009 | 0.010 | 0.012 | 371 | 100 | 0.0095 | 0.025 | 0.029 | 372 | 125 | 0.01 | 0.06 | 0.074 | 373 | 150 | 0.0108 | 0.08 | 0.10 | 374 | 175 | 0.025 | 0.115 | 0.129 | 375 | 200 | 0.06 | 0.15 | 0.20 | 376 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 378 Tab.2 Average hop of four strategies 380 5. CONCLUSION 382 In the tidal traffic scenario, we propose AAFS-ANN model and MP-RA 383 strategy. We use AAFS-ANN model to predict traffic and MP-RA to 384 optimize metropolitan optical network. Results demonstrate that 385 AAFS-ANN and MP-RA successfully increase prediction accuracy and 386 resource utilization, as well as reduce the traffic blocking rate. 388 6. ACKNOWLEDGMENT 390 This work has been supported in part by NSFC project (61871056), 391 National Postdoctoral Program for Innovative Talents (BX201600021), 392 Fundamental Research Funds for the Central Universities (2018XKJC06) 393 and State Key Laboratory of Information Photonics and Optical 394 Communications (BUPT), P. R. China (No. IPOC2017ZT11). 396 7. References 398 7.1. Normative References 400 [RFC2119] Bradner, S., "Key words for use in RFC's to Indicate 401 Requirement Levels", RFC 2119, March 1997. 403 7.2. Informative References 405 [Ref1] Alvizu, R., Troia, S., and G. Maier, "Matheuristic with 406 machine-learning-based prediction for software-defined 407 mobile metro-core networks", May 2017. 409 [Ref2] Zhong, Z., Hua, N., and H. Liu, "Considerations of 410 effective tidal traffic dispatching in software-defined 411 metro IP over optical networks", July 2015. 413 [Ref3] Yu, A., Yang, H., and W. Bai, "Leveraging deep learning to 414 achieve efficient resource allocation with traffic 415 evaluation in datacenter optical networks", March 2018. 417 [Ref4] Zhong, Z., Hua, N., and M. Tornatore, "Energy efficiency 418 and blocking reduction for tidal traffic via stateful 419 grooming in IP-over-optical networks", September 2016. 421 Authors' Addresses 423 Hui Yang 424 Beijing University of Posts and Telecommunications 425 No.10,Xitucheng Road,Haidian District 426 Beijing 100876 427 P.R.China 429 Phone: +8613466774108 430 Email: yang.hui.y@126.com 431 URI: http://www.bupt.edu.cn/ 433 Kaixuan Zhan 434 Beijing University of Posts and Telecommunications 435 No.10,Xitucheng Road,Haidian District 436 Beijing 100876 437 P.R.China 439 Phone: +8618401695826 440 Email: zhankai@bupt.edu.cn 441 URI: http://www.bupt.edu.cn/ 443 Ao Yu 444 Beijing University of Posts and Telecommunications 445 No.10,Xitucheng Road,Haidian District 446 Beijing 100876 447 P.R.China 449 Email: yuao@bupt.edu.cn 450 URI: http://www.bupt.edu.cn/ 451 Qiuyan Yao 452 Beijing University of Posts and Telecommunications 453 No.10,Xitucheng Road,Haidian District 454 Beijing 100876 455 P.R.China 457 Email: yqy89716@bupt.edu.cn 458 URI: http://www.bupt.edu.cn/ 460 Jie Zhang 461 Beijing University of Posts and Telecommunications 462 No.10,Xitucheng Road,Haidian District 463 Beijing 100876 464 P.R.China 466 Phone: +8613911060930 467 Email: lgr24@bupt.edu.cn 468 URI: http://www.bupt.edu.cn/