idnits 2.17.1 draft-jarvinen-core-fasor-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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: If the server supports (implements) the Retransmission Count option and the option is present in a request, the server MUST echo the option value in its Piggybacked Response unmodified. If the server replies with an Empty Acknowledgement the server MUST silently ignore the option and MUST not include it in a later separate response to that request. -- The document date (July 3, 2018) is 2123 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- -- Obsolete informational reference (is this intentional?): RFC 896 (Obsoleted by RFC 7805) Summary: 0 errors (**), 0 flaws (~~), 2 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 CoRE Working Group I. Jarvinen 3 Internet-Draft M. Kojo 4 Intended status: Experimental I. Raitahila 5 Expires: January 4, 2019 University of Helsinki 6 Z. Cao 7 Huawei 8 July 3, 2018 10 Fast-Slow Retransmission Timeout and Congestion Control Algorithm for 11 CoAP 12 draft-jarvinen-core-fasor-00 14 Abstract 16 This document specifies an alternative retransmission timeout and 17 congestion control back off algorithm for the CoAP protocol, called 18 Fast-Slow RTO (FASOR). 20 The algorithm specified in this document employs an appropriate and 21 large enough back off of Retransmission Timeout (RTO) as the major 22 congestion control mechanism to allow acquiring unambiguous RTT 23 samples with high probability and to prevent building a persistent 24 queue when retransmitting. The algorithm also aims to retransmit 25 quickly using an accurately managed retransmission timeout when link- 26 errors are occuring, basing RTO calculation on unambiguous round-trip 27 time (RTT) samples. 29 Status of This Memo 31 This Internet-Draft is submitted in full conformance with the 32 provisions of BCP 78 and BCP 79. 34 Internet-Drafts are working documents of the Internet Engineering 35 Task Force (IETF). Note that other groups may also distribute 36 working documents as Internet-Drafts. The list of current Internet- 37 Drafts is at https://datatracker.ietf.org/drafts/current/. 39 Internet-Drafts are draft documents valid for a maximum of six months 40 and may be updated, replaced, or obsoleted by other documents at any 41 time. It is inappropriate to use Internet-Drafts as reference 42 material or to cite them other than as "work in progress." 44 This Internet-Draft will expire on January 4, 2019. 46 Copyright Notice 48 Copyright (c) 2018 IETF Trust and the persons identified as the 49 document authors. All rights reserved. 51 This document is subject to BCP 78 and the IETF Trust's Legal 52 Provisions Relating to IETF Documents 53 (https://trustee.ietf.org/license-info) in effect on the date of 54 publication of this document. Please review these documents 55 carefully, as they describe your rights and restrictions with respect 56 to this document. Code Components extracted from this document must 57 include Simplified BSD License text as described in Section 4.e of 58 the Trust Legal Provisions and are provided without warranty as 59 described in the Simplified BSD License. 61 Table of Contents 63 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 64 2. Conventions . . . . . . . . . . . . . . . . . . . . . . . . . 3 65 3. Problems with Existing CoAP Congestion Control Algorithms . . 3 66 4. FASOR Algorithm . . . . . . . . . . . . . . . . . . . . . . . 4 67 4.1. Computing Normal RTO (FastRTO) . . . . . . . . . . . . . 4 68 4.2. Slow RTO . . . . . . . . . . . . . . . . . . . . . . . . 5 69 4.3. Retransmission Timeout Back Off Logic . . . . . . . . . . 6 70 4.3.1. Overview . . . . . . . . . . . . . . . . . . . . . . 6 71 4.3.2. Retransmission State Machine . . . . . . . . . . . . 7 72 4.4. Retransmission Count Option . . . . . . . . . . . . . . . 9 73 4.5. Alternatives for Exchanging Retransmission Count 74 Information . . . . . . . . . . . . . . . . . . . . . . . 11 75 5. Security Considerations . . . . . . . . . . . . . . . . . . . 11 76 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 11 77 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 11 78 7.1. Normative References . . . . . . . . . . . . . . . . . . 11 79 7.2. Informative References . . . . . . . . . . . . . . . . . 12 80 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 12 82 1. Introduction 84 CoAP senders use retransmission timeout (RTO) to infer losses that 85 have occurred in the network. For such a heuristic to be correct, 86 the RTT estimate used for calculating the retransmission timeout must 87 match to the real end-to-end path characteristics. Otherwise, 88 unnecessary retransmission may occur. Both default RTO mechanism for 89 CoAP [RFC7252] and CoCoA [I-D.ietf-core-cocoa] have issues in dealing 90 with unnecessary retransmissions and in the worst-case the situation 91 can persist causing congestion collapse [JRCK18]. 93 This document specifies FASOR retransmission timeout and congestion 94 control algorithm. FASOR algorithm ensures unnecessary 95 retransmissions that a sender may have sent due to an inaccurate RTT 96 estimate will not persist avoiding the threat of congestion collapse. 97 FASOR also aims to quickly restore the accuracy of the RTT estimate. 98 Armed with an accurate RTT estimate, FASOR not only handles 99 congestion robustly but also can quickly infer losses due to link 100 errors. 102 2. Conventions 104 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 105 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 106 document are to be interpreted as described in BCP 14, RFC 2119 107 [RFC2119]. 109 3. Problems with Existing CoAP Congestion Control Algorithms 111 Correctly inferring losses requires the retransmission timeout (RTO) 112 to be longer than the real RTT in the network. Under certain 113 circumstances the RTO may be incorrectly small. If the real end-to- 114 end RTT is larger than the retransmission timeout, it is impossible 115 for the sender to avoid making unnecessary retransmissions that 116 duplicate data still existing in the network because the sender 117 cannot receive any feedback in time. Unnecessary retransmissions 118 cause two basic problems. First, they increase the perceived end-to- 119 end RTT if the bottleneck has buffering capacity, and second, they 120 prevent getting unambiguous RTT samples. Making unnecessary 121 retransmissions is also a pre-condition for the congestion collapse 122 [RFC0896], which may occur in the worst case if retransmissions are 123 not well controlled [JRCK18]. Therefore, the sender retransmission 124 timeout algorithm should actively attempt to prevent unnecessary 125 retransmissions from persisting under any circumstance. 127 Karn's algorithm [KP87] has prevented unnecessary retransmission from 128 turning into congestion collapse for decades due to robust RTT 129 estimation and retransmission timeout backoff handling. The recent 130 CoAP congestion control algorithms, however, diverge from the 131 principles of Karn's algorithm in significant ways and may pose a 132 threat to the stability of the Internet due to those differences. 134 The default RTO mechanism for CoAP [RFC7252] uses only an initial RTO 135 dithered between 2 and 3 seconds, while CoCoA [I-D.ietf-core-cocoa] 136 measures RTT both from unambiguous and ambiguous RTT samples and 137 applies a modified version of the TCP RTO algorithm [RFC6298]. The 138 algorithm in RFC 7252 lacks solution to persistent congestion. The 139 binary exponential back off used for the retransmission timeout does 140 not properly address unnecessary retransmissions when RTT is larger 141 than the default RTO (ACK_TIMEOUT). If the CoAP sender performs 142 exchanges over an end-to-end path with such a high RTT, it 143 persistently keeps making unnecessary retransmissions for every 144 exchange wasting some fraction of the used resources (network 145 capacity, battery power). 147 CoCoA [I-D.ietf-core-cocoa] attempts to improve scenarios with link- 148 error related losses and solve persistent congestion by basing its 149 RTO value on an estimated RTT. However, there are couple of 150 exceptions when the RTT estimation is not available: 152 - At the beginning of a flow where initial RTO of 2 seconds is 153 used. 155 - When RTT suddenly jumps high enough to trigger the rule in CoCoA 156 that prevents taking RTT samples when more than two 157 retransmissions are needed. This may also occur when the packet 158 drop rate on the path is high enough. 160 When RTT estimate is too small, unnecessary retransmission will occur 161 also with CoCoA. CoCoA being unable to take RTT samples at all is a 162 particularly problematic phenomenon as it is similarly persisting 163 state as with the algorithm outlined in RFC 7252 and the network 164 remains in a congestion collapsed state due to persisting unnecessary 165 retransmissions. 167 4. FASOR Algorithm 169 FASOR is composed of three key components: RTO computation, Slow RTO, 170 and novel retransmission timeout back off logic. 172 4.1. Computing Normal RTO (FastRTO) 174 The FASOR algorithm measures the RTT for an CoAP message exchange 175 over an end-to-end path and computes the RTO value using the TCP RTO 176 algorithm specified in [RFC6298]. We call this normal RTO or 177 FastRTO. In contrast to the TCP RTO mechanism, FASOR SHOULD NOT use 178 1 second lower-bound when setting the RTO because RTO is only a 179 backup mechanisms for loss detection with TCP, whereas with CoAP RTO 180 is the primary and only loss detection mechanism. A lower-bound of 1 181 second would impact timeliness of the loss detection in low RTT 182 environments. The RTO value MAY be upper-bounded by at least 60 183 seconds. A CoAP sender using the FASOR algorithm SHOULD set initial 184 RTO to 2 seconds. The computed RTO value as well as the initial RTO 185 value is subject to dithering; they are dithered between RTO + 1/4 x 186 SRTT and RTO + SRTT. For dithering initial RTO, SRTT is unset; 187 therefore, SRTT is replaced with initial RTO / 3 which is derived 188 from the RTO formula and equals to a hypothetical initial RTT that 189 would yield the initial RTO using the SRTT and RTTVAR initialization 190 rule of RFC 6298. That is, for initial RTO of 2 seconds we use SRTT 191 value of 2/3 seconds. 193 RTO is updated only with unambiguous RTT samples. Therefore, it 194 closely tracks the actual RTT of the network and can quickly trigger 195 a retransmission when the network state is not dubious. 196 Retransmitting without extra delay is very useful when the end-to-end 197 path is subject to losses that are unrelated to congestion. When the 198 first unambiguous RTT sample is received, the RTT estimator is 199 initialized with that sample as specified in [RFC6298] except RTTVAR 200 that is set to R/2K. 202 4.2. Slow RTO 204 We introduce Slow RTO as a safe way to ensure that only a unique copy 205 of message is sent before at least one RTT has elapsed. To achieve 206 this the sender must ensure that its retransmission timeout is set to 207 a value that is larger than the path end-to-end RTT that may be 208 inflated by the unnecessary retransmission themselves. Therefore, 209 whenever a message needs to be retransmitted, we measure Slow RTO as 210 the elapsed time required for getting an acknowledgement. That is, 211 Slow RTO is measured starting from the original transmission of the 212 request message until the receipt of the acknowledgement, regardless 213 of the number of retransmissions. In this way, Slow RTO always 214 covers the worst-case RTT during which a number of unnecessary 215 retransmissions were made but the acknowledgement is received for the 216 original transmission. In contrast to computing normal RTO, Slow RTO 217 is not smoothed because it is derived from the sending pattern of the 218 retransmissions (that may turn out unnecessary). In order to drain 219 the potential unnecessary retransmissions successfully from the 220 network, it makes sense to wait for the time used for sending them 221 rather than some smoothed value. However, Slow RTO is multiplied by 222 a factor to allow some growth in load without making Slow RTO too 223 aggressive (by default the factor of 1.5 is used). FASOR then 224 applies Slow RTO as one of the backed off timer values used with the 225 next request message. 227 Slow RTO allows rapidly converging towards stable operating point 228 because 1) it lets the duplicate copies sent earlier to drain from 229 the network reducing the perceived end-to-end RTT, and 2) allows 230 enough time to acquire an unambiguous RTT sample for the RTO 231 computation. Robustly acquiring the RTT sample ensures that the next 232 RTO is set according to the recent measurement and further 233 unnecessary retransmissions are avoided. Slow RTO itself is a form 234 of back off because it includes the accumulated time from the 235 retransmission timeout back off of the previous exchange. FASOR uses 236 this for its advantage as the time included into Slow RTO is what is 237 needed to drain all unnecessary retransmissions possibly made during 238 the previous exchange. Assuming a stable RTT and that all of the 239 retransmissions were unnecessary, the time to drain them is the time 240 elapsed from the original transmission to the sending time of the 241 last retransmission plus one RTT. When the acknowledgement for the 242 original transmission arrives, one RTT has already elapsed, leaving 243 only the sending time difference still unaccounted for which is at 244 minimum the value for Slow RTO (when an RTT sample arrives 245 immediately after the last retransmission). Even if RTT would be 246 increasing, the draining still occurs rapidly due to exponentially 247 backed off frequency in sending the unnecessary retransmissions. 249 4.3. Retransmission Timeout Back Off Logic 251 4.3.1. Overview 253 FASOR uses normal RTO as the base for binary exponential back off 254 when no retransmission were needed for the previous CoAP message 255 exchange. When retransmission were needed for the previous CoAP 256 message exchange, the algorithm rules, however, are more complicated 257 than with the traditional RTO back off because Slow RTO is injected 258 into the back off series to reduce high impact of using Slow RTO. 259 FASOR logic chooses from three possible back off series alternatives: 261 FAST back off: Perform traditional RTO back off with the normal 262 RTO as the base. Applied when the previous message was not 263 retransmitted. 265 FAST_SLOW_FAST back off: First perform a probe using the normal 266 RTO for the original transmission of the request message to 267 improve cases with losses unrelated to congestion. If the probe 268 for the original transmission of the request message is successful 269 without retransmissions, continue with FAST back off for the next 270 message exchange. If the request message needs to be 271 retransmitted, continue by using Slow RTO for the first 272 retransmission in order to respond to congestion and drain the 273 network from the unnecessary retransmissions that were potentially 274 sent for the previous exchange. If still further RTOs are needed, 275 continue by backing off the normal RTO further on each timeout. 276 FAST_SLOW_FAST back off is applied just once when the previous 277 request message using FAST back off required one or more 278 retransmissions. 280 SLOW_FAST back off: Perform Slow RTO first for the original 281 transmisssion to respond to congestion and to acquire an 282 unambiguous RTT sample with high probability. Then, if the 283 original request needs to be retransmitted, continue with the 284 normal RTO-based RTO back off serie by backing off the normal RTO 285 on each timeout. SLOW_FAST back off is applied when the previous 286 request message using FAST_SLOW_FAST or SLOW_FAST back off 287 required one or more retransmissions. Once an acknowledgement for 288 the original transmission with unambigous RTT sample is received, 289 continue with FAST back off for the next message exchange. 291 For the initial message, FAST is used with INITIAL_RTO as the FastRTO 292 value. From there on, state is updated when an acknowledgement 293 arrives. Following unambiguous RTT samples, FASOR always uses FAST. 294 Whenever retransmissions are needed, the back off series selection is 295 first downgraded to FAST_SLOW_FAST back off and then to SLOW_FAST 296 back off if further retransmission are needed in FAST_SLOW_FAST. 298 When Slow RTO is used as the first RTO value, the sender is likely to 299 acquire unambiguous RTT sample even when the network has high delay 300 due to congestion because Slow RTO is based on a very recent 301 measurement of the worst-case RTT. However, using Slow RTO may 302 negatively impact the performance when losses unrelated to congestion 303 are occurring. Due to its potential high cost, FASOR algorithm 304 attempts to avoid using Slow RTO unnecessarily. 306 The CoAP protocol is often used by devices that are connected through 307 a wireless network where non-congestion related losses are much more 308 frequent than in their wired counterparts. This has implications for 309 the retransmission timeout algorithm. While it would be possible to 310 implement FASOR such that it immediately uses Slow RTO when a dubious 311 network state is detected, which would handle congestion very well, 312 it would do significant harm for performance when RTOs occur due to 313 non-congestion related losses. Instead, FASOR uses first normal RTO 314 for one transmission and only responds using Slow RTO if RTO expires 315 also for that request message. Such a pattern quickly probes if the 316 losses were unrelated to congestion and only slightly delays response 317 if real congestion event is taking place. To ensure that an 318 unambiguous RTT sample is also acquired on a congested network path, 319 FASOR then needs to use Slow RTO for the original transmission of the 320 subsequent packet if the probe was not successful. 322 4.3.2. Retransmission State Machine 324 FASOR consists of the three states discussed above while making 325 retranmission decisions, FAST, FAST_SLOW_FAST and SLOW_FAST. The 326 state machine of the FASOR algorithm is depicted in Figure 1. 328 +-------------------b----------------+ 329 | | 330 v | 331 +--FAST--a-->FAST_SLOW_FAST-----a----->SLOW_FAST--+ 332 | ^ ^ | ^ | 333 | | | | | | 334 +-b-+ +------b------+ +-a-+ 336 a: retransmission acknowledged, ambiguous RTT sample acquired; 337 b: no retransmission, umambiguous RTT sample acquired; 339 Figure 1: State Machine of FASOR 341 In the FAST state, if the original transmission of the message has 342 not been acknowledged by the receiver within the time defined by 343 FastRTO, the sender will retransmit it. If there is still no 344 acknowledgement of the retransmitted packet within 2*FastRTO, the 345 sender performs the second retransmission and if necessary, each 346 further retransmission applying binary exponential back off of 347 FastRTO. The retransmission interval in this state is defined as 348 FastRTO, 2^1 * FastRTO, ..., 2^i * FastRTO. 350 When there is an acknowledgement after any retransmission, the sender 351 will calculate SlowRTO value based on the algorithm defined in 352 Section 4.2. 354 When these is an acknowledgement after any retransmission, the sender 355 will also switch to the second state, FAST_FLOW_FAST. In this state, 356 the retransmission interval is defined as FastRTO, Max(SlowRTO, 357 2*FastRTO), FastRTO * 2^1, ..., 2^i * FastRTO. The state will be 358 switched back to the FAST state once an acknowledgement is returned 359 within FastRTO, i.e., no retransmission happens for a message. This 360 is reasonable because it shows the network has recovered from 361 congestion or bloated queue. 363 If some retransmission has been made before the acknowledged arrives 364 in the FAST_SLOW_FAST state, the sender updates the SlowRTO value, 365 and moves to the third state, SLOW_FAST. The retransmission interval 366 in the SLOW_FAST state is defined as SlowRTO, FastRTO, FastRTO * 2^1, 367 ..., 2^i * FastRTO. 369 In SLOW_FAST state, the sender switches back to the FAST state if an 370 unambiguous acknowledgement arrives. Otherwise, the sender stays in 371 the SLOW_FAST state if retransmission happens again. 373 4.4. Retransmission Count Option 375 When retransmissions are needed to deliver a CoAP message, it is not 376 possible to measure RTT for the RTO computation as the RTT sample 377 becomes ambiguous. Therefore, it would be beneficial to be able to 378 distinguish whether an acknowledgement arrives for the original 379 transmission of the message or for a retransmission of it. This 380 would allow reliably acquiring an RTT sample for every CoAP message 381 exchange and thereby compute a more accurate RTO even during periods 382 of congestion and loss. 384 The Retransmission Count Option is used to distinguish whether an 385 Acknowledgement message arrives for the original transmission or one 386 of the retransmissions of a Confirmable message. However, the 387 Retransmission Count Option cannot be used with an Empty 388 Acknowledgement (or Reset) message because the CoAP protocol 389 specification [RFC7252] does not allow adding options to an Empty 390 message. Therefore, Retransmission Count Option is useful only for 391 the common case of Piggybacked Response. In case of Empty 392 Acknowledgements the operation of FASOR is the same as without the 393 option. 395 +-----+---+---+---+---+------------+--------+--------+---------+ 396 | No. | C | U | N | R | Name | Format | Length | Default | 397 +-----+---+---+---+---+------------+--------+--------+---------+ 398 | TBD | | | X | | Rexmit-Cnt | uint | 0-1 | 0 | 399 +-----+---+---+---+---+------------+--------+--------+---------+ 401 C=Critical, U=Unsafe, N=NoCacheKey, R=Repeatable 403 Table 1: Retransmission Count Option 405 Implementation of the Retransmission Count option is optional and it 406 is identified as elective. However, when it is present in a CoAP 407 message and a CoAP endpoint processes it, it MUST be processed as 408 described in this document. The Retransmission Count option MUST NOT 409 occur more than once in a single message. 411 The value of the Retransmission Count option is a variable-size (0 to 412 1 byte) unsigned integer. The default value for the option is the 413 number 0 and it is represented with an empty option value (a zero- 414 length sequence of bytes). However, when a client intents to use 415 Retransmit Count option, it MUST reserve space for it by limiting the 416 request message size also when the value is empty in order to fit the 417 full-sized option into retransmissions. 419 The Retransmission Count option can be present in both the request 420 and response message. When the option is present in a request it 421 indicates the ordinal number of the transmission for the request 422 message. 424 If the server supports (implements) the Retransmission Count option 425 and the option is present in a request, the server MUST echo the 426 option value in its Piggybacked Response unmodified. If the server 427 replies with an Empty Acknowledgement the server MUST silently ignore 428 the option and MUST not include it in a later separate response to 429 that request. 431 When Piggybacked Response carrying the Retransmission Count option 432 arrives, the client uses the option to match the response message to 433 the corresponding transmission of the request. In order to measure a 434 correct RTT, the client must store the timestamp for the original 435 transmission of the request as well as the timestamp for each 436 retransmission, if any, of the request. The resulting RTT sample is 437 used for the RTO computation. If the client retransmitted the 438 request without the option but the response includes the option, the 439 client MUST silently ignore the option. 441 The original transmission of a request is indicated with the number 442 0, except when sending the first request to a new destination 443 endpoint. The first original transmission of the request to a new 444 endpoint carries the number 255 (0xFF) and is interpreted the same as 445 an original transmission carrying the number 0. Retransmissions, if 446 any, carry the ordinal number of the retransmission. Once the first 447 Piggybacked Response from the new endpoint arrives the client learns 448 whether or not the other endpoint implements the option. If the 449 first response includes the echoed option, the client learns that the 450 other endpoint supports the option and may continue including the 451 option to each retransmitted request. From this point on the 452 original transmissions of requests implicitly include the option 453 number 0 and a zero-byte integer will be sent according to the CoAP 454 uint-encoding rules. If the first Piggybacked Response does not 455 include the option, the client SHOULD stop including the option into 456 the requests to that endpoint. 458 When the Retransmission Count option is in use, the client bases the 459 retransmission timeout for the normal RTO in the back off series as 460 follows: 462 max(RTO, Previous-RTT-Sample) 464 Previous-RTT-Sample is the RTT sample acquired from the previous 465 message exchange. If no RTT sample was available with the previous 466 message exchange (e.g., the server replied with an Empty 467 Acknowledgement), RTO computed earlier is used like in case the 468 Retransmission Count option is not in use. 470 4.5. Alternatives for Exchanging Retransmission Count Information 472 An alternative way of exchanging the retransmission count information 473 between a client and server is to encode it in the Token. The Token 474 is a client-local identifier and a client solely decides how it 475 generates the Token. Therefore, including a varying Token value to 476 retransmissions of the same request is all possible as long as the 477 client can use the Token to differentiate between requests and match 478 a response to the corresponding request. The server is required to 479 make no assumptions about the content or structure of a Token and 480 always echo the Token unmodified in its response. 482 How exactly a client encodes the retransmission count into a Token is 483 an implementation issue. Note that the original transmission of a 484 request may carry a zero-length Token given that the rules for 485 generating a Token as specified in RFC 7252 [RFC7252] are followed. 486 This allows reducing the overhead of including the Token into the 487 reguests in such cases where Token could otherwise be omitted. 488 However, similar to Retransmit Count option the maximum request 489 message size MUST be limited to accommodate the Token with retransmit 490 count into the retransmissions of the request. 492 5. Security Considerations 494 6. IANA Considerations 496 This memo includes no request to IANA. 498 7. References 500 7.1. Normative References 502 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 503 Requirement Levels", BCP 14, RFC 2119, 504 DOI 10.17487/RFC2119, March 1997, 505 . 507 [RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, 508 "Computing TCP's Retransmission Timer", RFC 6298, 509 DOI 10.17487/RFC6298, June 2011, 510 . 512 [RFC7252] Shelby, Z., Hartke, K., and C. Bormann, "The Constrained 513 Application Protocol (CoAP)", RFC 7252, 514 DOI 10.17487/RFC7252, June 2014, 515 . 517 7.2. Informative References 519 [I-D.ietf-core-cocoa] 520 Bormann, C., Betzler, A., Gomez, C., and I. Demirkol, 521 "CoAP Simple Congestion Control/Advanced", draft-ietf- 522 core-cocoa-03 (work in progress), February 2018. 524 [JRCK18] Jarvinen, I., Raitahila, I., Cao, Z., and M. Kojo, "Is 525 CoAP Congestion Safe?", Applied Networking Research 526 Workshop (ANRW'18), July 2018. 528 [KP87] Karn, P. and C. Partridge, "Improving Round-trip Time 529 Estimates in Reliable Transport Protocols", SIGCOMM'87 530 Proceedings of the ACM Workshop on Frontiers in Computer 531 Communications Technology, August 1987. 533 [RFC0896] Nagle, J., "Congestion Control in IP/TCP Internetworks", 534 RFC 896, DOI 10.17487/RFC0896, January 1984, 535 . 537 Authors' Addresses 539 Ilpo Jarvinen 540 University of Helsinki 541 P.O. Box 68 542 FI-00014 UNIVERSITY OF HELSINKI 543 Finland 545 EMail: ilpo.jarvinen@helsinki.fi 547 Markku Kojo 548 University of Helsinki 549 P.O. Box 68 550 FI-00014 UNIVERSITY OF HELSINKI 551 Finland 553 EMail: markku.kojo@cs.helsinki.fi 555 Iivo Raitahila 556 University of Helsinki 557 P.O. Box 68 558 FI-00014 UNIVERSITY OF HELSINKI 559 Finland 561 EMail: iivo.raitahila@helsinki.fi 562 Zhen Cao 563 Huawei 564 Beijing 565 China 567 EMail: zhencao.ietf@gmail.com