idnits 2.17.1 draft-thubert-6man-ipv6-over-wireless-03.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 -- The document date (May 2, 2019) is 1820 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-23) exists of draft-ietf-6lo-ap-nd-12 == Outdated reference: A later version (-20) exists of draft-ietf-6lo-backbone-router-11 == Outdated reference: A later version (-24) exists of draft-bi-savi-wlan-16 == Outdated reference: A later version (-30) exists of draft-ietf-6tisch-architecture-20 == Outdated reference: A later version (-15) exists of draft-ietf-mboned-ieee802-mcast-problems-05 == Outdated reference: A later version (-21) exists of draft-ietf-rift-rift-05 == Outdated reference: A later version (-02) exists of draft-thubert-6lo-unicast-lookup-00 Summary: 0 errors (**), 0 flaws (~~), 8 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 6MAN P. Thubert, Ed. 3 Internet-Draft Cisco Systems 4 Intended status: Informational May 2, 2019 5 Expires: November 3, 2019 7 IPv6 Neighbor Discovery on Wireless Networks 8 draft-thubert-6man-ipv6-over-wireless-03 10 Abstract 12 This document describes how the original IPv6 Neighbor Discovery and 13 Wireless ND (WiND) can be applied on various abstractions of wireless 14 media. 16 Status of This Memo 18 This Internet-Draft is submitted in full conformance with the 19 provisions of BCP 78 and BCP 79. 21 Internet-Drafts are working documents of the Internet Engineering 22 Task Force (IETF). Note that other groups may also distribute 23 working documents as Internet-Drafts. The list of current Internet- 24 Drafts is at https://datatracker.ietf.org/drafts/current/. 26 Internet-Drafts are draft documents valid for a maximum of six months 27 and may be updated, replaced, or obsoleted by other documents at any 28 time. It is inappropriate to use Internet-Drafts as reference 29 material or to cite them other than as "work in progress." 31 This Internet-Draft will expire on November 3, 2019. 33 Copyright Notice 35 Copyright (c) 2019 IETF Trust and the persons identified as the 36 document authors. All rights reserved. 38 This document is subject to BCP 78 and the IETF Trust's Legal 39 Provisions Relating to IETF Documents 40 (https://trustee.ietf.org/license-info) in effect on the date of 41 publication of this document. Please review these documents 42 carefully, as they describe your rights and restrictions with respect 43 to this document. Code Components extracted from this document must 44 include Simplified BSD License text as described in Section 4.e of 45 the Trust Legal Provisions and are provided without warranty as 46 described in the Simplified BSD License. 48 Table of Contents 50 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 51 2. Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . 4 52 3. IP Models . . . . . . . . . . . . . . . . . . . . . . . . . . 6 53 3.1. Physical Broadcast Domain . . . . . . . . . . . . . . . . 6 54 3.2. MAC-Layer Broadcast Emulations . . . . . . . . . . . . . 7 55 3.3. Mapping the IPv6 Link Abstraction . . . . . . . . . . . . 8 56 3.4. Mapping the IPv6 Subnet Abstraction . . . . . . . . . . . 9 57 4. Wireless ND . . . . . . . . . . . . . . . . . . . . . . . . . 10 58 4.1. Introduction to WiND . . . . . . . . . . . . . . . . . . 10 59 4.2. Links and Link-Local Addresses . . . . . . . . . . . . . 11 60 4.3. Subnets and Global Addresses . . . . . . . . . . . . . . 11 61 5. WiND Applicability . . . . . . . . . . . . . . . . . . . . . 12 62 5.1. Case of LPWANs . . . . . . . . . . . . . . . . . . . . . 13 63 5.2. Case of Infrastructure BSS and ESS . . . . . . . . . . . 13 64 5.3. Case of Mesh Under Technologies . . . . . . . . . . . . . 14 65 5.4. Case of DMC radios . . . . . . . . . . . . . . . . . . . 14 66 5.4.1. Using IPv6 ND only . . . . . . . . . . . . . . . . . 15 67 5.4.2. Using Wireless ND . . . . . . . . . . . . . . . . . . 15 68 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 17 69 7. Security Considerations . . . . . . . . . . . . . . . . . . . 18 70 8. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 18 71 9. References . . . . . . . . . . . . . . . . . . . . . . . . . 18 72 9.1. Normative References . . . . . . . . . . . . . . . . . . 18 73 9.2. Informative References . . . . . . . . . . . . . . . . . 19 74 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 21 76 1. Introduction 78 IEEE STD. 802.1 [IEEEstd8021] Ethernet Bridging provides an efficient 79 and reliable broadcast service for wired networks; applications and 80 protocols have been built that heavily depend on that feature for 81 their core operation. Unfortunately, Low-Power Lossy Networks (LLNs) 82 and local wireless networks generally do not provide the broadcast 83 capabilities of Ethernet Bridging in an economical fashion. 85 As a result, protocols designed for bridged networks that rely on 86 multicast and broadcast often exhibit disappointing behaviours when 87 employed unmodified on a local wireless medium (see 88 [I-D.ietf-mboned-ieee802-mcast-problems]). 90 Wi-Fi [IEEEstd80211] Access Points (APs) deployed in an Extended 91 Service Set (ESS) act as Ethernet Bridges [IEEEstd8021], with the 92 property that the bridging state is established at the time of 93 association. This ensures connectivity to the node (STA) and 94 protects the wireless medium against broadcast-intensive Transparent 95 Bridging reactive Lookups. 97 In other words, the association process is used to register the MAC 98 Address of the STA to the AP. The AP subsequently proxies the 99 bridging operation and does not need to forward the broadcast Lookups 100 over the radio. 102 Like Transparent Bridging, IPv6 [RFC8200] Neighbor Discovery 103 [RFC4861] [RFC4862] Protocol (IPv6 ND) is a reactive protocol, based 104 on multicast transmissions to locate an on-link correspondent and 105 ensure the uniqueness of an IPv6 address. The mechanism for 106 Duplicate Address Detection (DAD) [RFC4862] was designed for the 107 efficient broadcast operation of Ethernet Bridging. Since broadcast 108 can be unreliable over wireless media, DAD often fails to discover 109 duplications [I-D.yourtchenko-6man-dad-issues]. In practice, IPv6 110 addresses very rarely conflict because of the entropy of the 64-bit 111 Interface IDs, not because address duplications are detected and 112 resolved. 114 The IPv6 ND Neighbor Solicitation (NS) [RFC4861] message is used for 115 DAD and address Lookup when a node moves, or wakes up and reconnects 116 to the wireless network. The NS message is targeted to a Solicited- 117 Node Multicast Address (SNMA) [RFC4291] and should in theory only 118 reach a very small group of nodes. But in reality, IPv6 multicast 119 messages are typically broadcast on the wireless medium, and so they 120 are processed by most of the wireless nodes over the subnet (e.g., 121 the ESS fabric) regardless of how few of the nodes are subscribed to 122 the SNMA. As a result, IPv6 ND address Lookups and DADs over a large 123 wireless and/or a LowPower Lossy Network (LLN) can consume enough 124 bandwidth to cause a substantial degradation to the unicast traffic 125 service [I-D.vyncke-6man-mcast-not-efficient]. 127 Because IPv6 ND messages sent to the SNMA group are broadcasted at 128 the radio MAC Layer, wireless nodes that do not belong to the SNMA 129 group still have to keep their radio turned on to listen to multicast 130 NS messages, which is a total waste of energy for them. In order to 131 reduce their power consumption, certain battery-operated devices such 132 as IoT sensors and smartphones ignore some of the broadcasts, making 133 IPv6 ND operations even less reliable. 135 These problems can be alleviated by reducing the IPv6 ND broadcasts 136 over wireless access links. This has been done by splitting the 137 broadcast domains and by routing between subnets, at the extreme by 138 assigning a /64 prefix to each wireless node (see [RFC8273]). 140 Another way is to proxy at the boundary of the wired and wireless 141 domains the Layer-3 protocols that rely on MAC Layer broadcast 142 operations. For instance, IEEE 802.11 [IEEEstd80211] situates proxy- 143 ARP (IPv4) and proxy-ND (IPv6) functions at the Access Points (APs). 145 But proxying ND requires a perfect knowledge of the peer IPv6 146 addresses for which proxying is provided. In a generic fashion, 147 radio connectivity changes with movements and variations in the 148 environment, which makes forming and maintaining that knowledge a 149 hard problem in the general case. 151 Discovering peer addresses by snooping the IPV6 ND protocol as 152 proposed for SAVI [I-D.bi-savi-wlan] was found to be unreliable. An 153 IPv6 address may not be discovered immediately due to a packet loss, 154 or if a "silent" node is not currently using one of its addresses, 155 e.g., a node that waits in wake-on-lan state. A change of state, 156 e.g. due to a movement, may be missed or misordered, leading to 157 unreliable connectivity and an incomplete knowledge of the set of 158 peers. 160 Wireless ND (WiND) introduces a new approach to IPv6 ND that is 161 designed to apply to the WLANs and WPANs types of networks. On the 162 one hand, WiND avoids the use of broadcast operation for Address 163 Resolution and Duplicate Address Detection, and on the other hand, 164 WiND supports use cases where Subnet and MAC-level domains are not 165 congruent, which is common in those types of networks unless a 166 specific MAC-Level emulation is provided. 168 To achieve this, WiND applies routing inside the Subnets, which 169 enables MultiLink Subnets. Hosts register their addresses to their 170 serving routers with [RFC8505]. With the registration, routers have 171 a complete knowledge of the hosts they serve and in return, hosts 172 obtain routing services for their registered addresses. The 173 registration is abstract to the routing protocol, and it can be 174 protected to prevent impersonation attacks with [I-D.ietf-6lo-ap-nd]. 176 The routing service can be a simple reflexion in a Hub-and-Spoke 177 Subnet that emulates an IEEE Std 802.11 Infrastructure BSS at Layer 178 3. It can also be a full-fledge routing protocol, in particular RPL 179 [RFC6550] that was designed to adapt to various LLNs such as WLAN and 180 WPAN radio meshes with the concept of Objective Function. Finally, 181 the routing service can also be ND proxy that emulates an IEEE Std 182 802.11 Infrastructure ESS at Layer 3. WiND specifies the IPv6 183 Backbone Router for that purpose in [I-D.ietf-6lo-backbone-router]. 185 More details on WiND can be found in Section 4.1. 187 2. Acronyms 189 This document uses the following abbreviations: 191 6BBR: 6LoWPAN Backbone Router 192 6LBR: 6LoWPAN Border Router 194 6LN: 6LoWPAN Node 196 6LR: 6LoWPAN Router 198 ARO: Address Registration Option 200 DAC: Duplicate Address Confirmation 202 DAD: Duplicate Address Detection 204 DAR: Duplicate Address Request 206 EDAC: Extended Duplicate Address Confirmation 208 EDAR: Extended Duplicate Address Request 210 MLSN: Multi-Link Subnet 212 LLN: Low-Power and Lossy Network 214 NA: Neighbor Advertisement 216 NBMA: Non-Broadcast Multi-Access 218 NCE: Neighbor Cache Entry 220 ND: Neighbor Discovery 222 NDP: Neighbor Discovery Protocol 224 NS: Neighbor Solicitation 226 RPL: IPv6 Routing Protocol for LLNs 228 RA: Router Advertisement 230 RS: Router Solicitation 232 WiND: Wireless Neighbor Discovery 234 WLAN: Wireless Local Area Network 236 WPAN: Wireless Personal Area Network 238 3. IP Models 240 3.1. Physical Broadcast Domain 242 At the physical (PHY) Layer, a broadcast domain is the set of nodes 243 that may receive a datagram that one sends over an interface, in 244 other words the set of nodes in range of radio transmission. This 245 set can comprise a single peer on a serial cable used as point-to- 246 point (P2P) link. It may also comprise multiple peer nodes on a 247 broadcast radio or a shared physical resource such as the legacy 248 Ethernet shared wire. 250 On WLAN and WPAN radios, the physical broadcast domain is defined by 251 a particular transmitter, as the set of nodes that can receive what 252 this transmitter is sending. Literally every datagram defines its 253 own broadcast domain since the chances of reception of a given 254 datagram are statistical. In average and in stable conditions, the 255 broadcast domain of a particular node can be still be seen as mostly 256 constant and can be used to define a closure of nodes on which an 257 upper-layer abstraction can be built. 259 A PHY-layer communication can be established between 2 nodes if their 260 physical broadcast domains overlap. 262 On WLAN and WPAN radios, this property is usually reflexive, meaning 263 that if B can receive a datagram from A, then A can receive a 264 datagram from B. But there can be asymmetries due to power levels, 265 interferers near one of the receivers, or differences in the quality 266 of the hardware (e.g., crystals, PAs and antennas) that may affect 267 the balance to the point that the connectivity becomes mostly uni- 268 directional, e.g., A to B but practically not B to A. It takes a 269 particular effort to place a set of devices in a fashion that all 270 their physical broadcast domains fully overlap, and it can not be 271 assumed in the general case. In other words, the property of radio 272 connectivity is generally not transitive, meaning that A may be in 273 range with B and B may be in range with C does not necessarily imply 274 that A is in range with C. 276 We define MAC-Layer Direct Broadcast (DMC) a transmission mode where 277 the broadcast domain that is usable at the MAC layer is directly the 278 physical broadcast domain. IEEE 802.15.4 [IEEE802154] and IEEE 279 802.11 [IEEEstd80211] OCB (for Out of the Context of a BSS) are 280 examples of DMC radios. This constrasts with a number of MAC-layer 281 Broadcast Emulation schemes that are described in the next section. 283 3.2. MAC-Layer Broadcast Emulations 285 While a physical broadcast domain is constrained to a single shared 286 wire, Ethernet Bridging emulates the broadcast properties of that 287 wire over a whole physical mesh of Ethernet links. For the upper 288 layer, the qualities of the shared wire are essentially conserved, 289 with a reliable and cheap broadcast operation over a closure of nodes 290 defined by their connectivity to the emulated wire. 292 In large switched fabrics, overlay techniques enable a limited 293 connectivity between nodes that are known to a mapping server. The 294 emulated broadcast domain is configured to the system, e.g., with a 295 VXLAN network identifier (VNID). Broadcast operations on the overlay 296 can be emulated but can become very expensive, and it makes sense to 297 proactively install the relevant state in the mapping server as 298 opposed to rely on reactive broadcast lookups. 300 An IEEE Std 802.11 Infrastructure Basic Service Set (BSS) also 301 provides a closure of nodes as defined by the broadcast domain of a 302 central Access Point (AP). The AP relays both unicast and broadcast 303 packets and ensures a reflexive and transitive emulation of the 304 shared wire between the associated nodes, with the capability to 305 signal link-up/link-down to the upper layer. Within an 306 Infrastructure BSS, the physical broadcast domain of the AP serves as 307 emulated broadcast domain for all the nodes that are associated to 308 the AP. Broadcast packets are relayed by the AP and are not 309 acknowledged. To ensure that all nodes in the BSS receive the 310 broadcast transmission, AP transmits at the slowest PHY speed. This 311 translates into maximum co-channel interferences for others and 312 longest occupancy of the medium, for a duration that can be 100 times 313 that of a unicast. For that reason, upper layer protocols should 314 tend to avoid the use of broadcast when operating over Wi-Fi. 316 In an IEEE Std 802.11 Infrastructure Extended Service Set (ESS), 317 infrastructure BSSes are interconnected by a bridged network, 318 typically running Transparent Bridging and Spanning tree Protocol. 319 In the original model, the state in the Transparent Bridge is set by 320 observing the source MAC address of the frames. When a state is 321 missing for a destination MAC address, the frame is broadcasted with 322 the expectation that the response will populate the state. This is a 323 reactive operation, meaning that the state is populated reactively to 324 a need for forwarding. It is also possible to send a gratuitous 325 frame to advertise self throughout the bridged network, and that is 326 also a broadcast. The process of the association prepares a bridging 327 state proactively at the AP, so as to avoid the reactive broadcast 328 lookup. It may also generates a gratuitous broadcast sourced at the 329 MAC address of the STA to prepare or update the state in the 330 Transparent Bridges. This model avoids the need of multicast over 331 the wireless access, and it is only logical that IPv6 ND evolved 332 towards proposes similar methods at Layer-3 for its operation. 334 In some cases of WLAN and WPAN radios, a mesh-under technology (e.g., 335 a IEEE 802.11s or IEEE 802.15.10) provides meshing services that are 336 similar to bridging, and the broadcast domain is well defined by the 337 membership of the mesh. Mesh-Under emulates a broadcast domain by 338 flooding the broadcast packets at Layer-2. When operating on a 339 single frequency, this operation is known to interfere with itself, 340 forcing deployment to introduce delays that dampen the collisions. 341 All in all, the mechanism is slow, inefficient and expensive. 343 Going down the list of cases above, the cost of a broadcast 344 transmissions becomes increasingly expensive, and there is a push to 345 rethink the upper-layer protocols so as to reduce the depency on 346 broadcast operations. 348 There again, a MAC-layer communication can be established between 2 349 nodes if their MAC-layer broadcast domains overlap. In the absence 350 of a MAC-layer emulation such as a mesh-under or an Infrastructure 351 BSS, the MAC-layer broadcast domain is congruent with that of the 352 PHY-layer and inherits its properties for reflexivity and 353 transitivity. IEEE 802.11p, which operates Out of the Context of a 354 BSS (DMC radios) is an example of a network that does not have a MAC- 355 Layer broadcast domain emulation, which means that it will exhibit 356 mostly reflexive and mostly non-transitive transmission properties. 358 3.3. Mapping the IPv6 Link Abstraction 360 IPv6 defines a concept of Link, Link Scope and Link-Local Addresses 361 (LLA), an LLA being unique and usable only within the Scope of a 362 Link. The IPv6 Neighbor Discovery (ND) [RFC4861][RFC4862] Duplicate 363 Adress Detection (DAD) process leverages a multicast transmission to 364 ensure that an IPv6 address is unique as long as the owner of the 365 address is connected to the broadcast domain. It must be noted that 366 in all the cases in this specification, the Layer-3 multicast 367 operation is always a MAC_Layer broadcast for the lack of a Layer-2 368 multicast operation that could handle a possibly very large number of 369 groups in order to make the unicast efficient. This means that for 370 every multicast packet regardless of the destination group, all nodes 371 will receive the packet and process it all the way to Layer-3. 373 On wired media, the Link is often confused with the physical 374 broadcast domain because both are determined by the serial cable or 375 the Ethernet shared wire. Ethernet Bridging reinforces that illusion 376 by provising a MAC-Layer broadcast domain that emulates a physical 377 broadcast domain over the mesh of wires. But the difference shows on 378 legacy Non-Broadcast Multi-Access (NBMA) such as ATM and Frame-Relay, 379 on shared links and on newer types of NBMA networks such as radio and 380 composite radio-wires networks. It also shows when private VLANs or 381 Layer-2 cryptography restrict the capability to read a frame to a 382 subset of the connected nodes. 384 In mesh-under and Infrastructure BSS, the IP Link extends beyond the 385 physical broadcast domain to the emulated MAC-Layer broadcast domain. 386 Relying on Multicast for the ND operation remains feasible but 387 becomes detrimental to unicast traffic, energy-inefficient and 388 unreliable, and its use is discouraged. 390 On DMC radios, IP Links between peers come and go as the individual 391 physical broadcast domains of the transmitters meet and overlap. The 392 DAD operation cannot provide once and for all guarantees on the 393 broadcast domain defined by one radio transmitter if that transmitter 394 keeps meeting new peers on the go. The nodes may need to form new 395 LLAs to talk to one another and the scope where LLA uniqueness can be 396 dynamically checked is that pair of nodes. As long as there's no 397 conflict a node may use the same LLA with multiple peers but it has 398 to revalidate DAD with every new peer node. In practice, each pair 399 of nodes defines a temporary P2P link, which can be modeled as a sub- 400 interface of the radio interface. 402 3.4. Mapping the IPv6 Subnet Abstraction 404 IPv6 also defines a concept of Subnet for Glocal and Unique Local 405 Addresses. Addresses in a same Subnet share a same prefix and by 406 extension, a node belongs to a Subnet if it has an interface with an 407 address on that Subnet. A Subnet prefix is Globally Unique so it is 408 sufficient to validate that an address that is formed from a Subnet 409 prefix is unique within that Subnet to guarantee that it is globally 410 unique. IPv6 aggregation relies on the property that a packet from 411 the outside of a Subnet can be routed to any router that belongs to 412 the Subnet, and that this router will be able to either resolve the 413 destination MAC address and deliver the packet, or route the packet 414 to the destination within the Subnet. If the Subnet is known as 415 onlink, then any node may also resolve the destination MAC address 416 and deliver the packet, but if the Subnet is not onlink, then a host 417 that does not have an NCE for the destination will need to pass the 418 packet to a router. 420 On IEEE Std. 802.3, a Subnet is often congruent with an IP Link 421 because both are determined by the physical attachment to an Ethernet 422 shared wire or an IEEE Std. 802.1 bridged broadcast domain. In that 423 case, the connectivity over the Link is transitive, the Subnet can 424 appear as onlink, and any node can resolve a destination MAC address 425 of any other node directly using IPv6 Neighbor Discovery. 427 But an IP Link and an IP Subnet are not always congruent. In a 428 shared Link situation, a Subnet may encompass only a subset of the 429 nodes connected to the Link. In Route-Over Multi-Link Subnets (MLSN) 430 [RFC4903], routers federate the Links between nodes that belong to 431 the Subnet, the Subnet is not onlink and it extends beyond any of the 432 federated Links. 434 The DAD and lookup procedures in IPv6 ND expects that a node in a 435 Subnet is reachable within the broadcast domain of any other node in 436 the Subnet when that other node attempts to form an address that 437 would be a duplicate or attempts to resolve the MAC address of this 438 node. This is why ND is only applicable for P2P and transit links, 439 and requires extensions for other topologies. 441 4. Wireless ND 443 4.1. Introduction to WiND 445 Wireless Neighbor Discovery (WiND) 446 [RFC6775][RFC8505][I-D.ietf-6lo-backbone-router][I-D.ietf-6lo-ap-nd] 447 defines a new ND operation that is based on 2 major paradigm changes, 448 proactive address registration by hosts to their attachment routers 449 and routing to host routes (/128) within the subnet. This allows 450 WiND to avoid the classical ND expectations of transit links and 451 Subnet-wide broadcast domains. 453 The proactive address registration is performed with a new option in 454 NS/NA messages, the Extended Address Registration Option (EARO) 455 defiend in [RFC8505]. This method allows to prepare and maintain the 456 host routes in the routers and avoids the reactive NS(Lookup) found 457 in IPv6 ND. This is a direct benefit for wireless Links since it 458 avoids the MAC level broadcasts that are associated to NS(Lookup). 460 The EARO provides information to the router that is independent to 461 the routing protocol and routing can take multiple forms, from a 462 traditional IGP to a collapsed ub-and-Spoke model where only one 463 router owns and advertises the prefix. [RFC8505] is already 464 referenced for RIFT [I-D.ietf-rift-rift], RPL [RFC6550] with 465 [I-D.thubert-roll-unaware-leaves] and IPv6 ND proxy 466 [I-D.ietf-6lo-backbone-router]. 468 WiND does not change IPv6 addressing [RFC4291] or the current 469 practices of assigning prefixes to subnets. It is still typical to 470 assign a /64 to a subnet and to use interface IDs of 64 bits. 471 Duplicate Address detection within the Subnet is performed with a 472 central registrar, using new ND Extended Duplicate Address messages 473 (EDAR and EDAC) [RFC8505]. This operation modernizes ND for 474 application in overlays with Map Resolvers and enables unicast 475 lookups [I-D.thubert-6lo-unicast-lookup] for addresses registered to 476 the resolver. 478 WiND also enables to extend a legacy /64 on Ethernet with ND proxy 479 over the wireless. This way nodes can form any address the want and 480 move freely from an L3-AP (that is really a backbone router in 481 bridging mode, more in [I-D.ietf-6lo-backbone-router]) to another, 482 without renumbering. Backbone Routers federate multiple LLNs over a 483 Backbone Link to form a MultiLink Subnet (MLSN). Backbone Routers 484 placed along the LLN edge of the Backbone handle IPv6 Neighbor 485 Discovery, and forward packets on behalf of registered nodes. 487 An LLN node (6LN) registers all its IPv6 Addresses using an NS(EARO) 488 as specified in [RFC8505] to the 6BBR. The 6BBR is also a Border 489 Router that performs IPv6 Neighbor Discovery (IPv6 ND) operations on 490 its Backbone interface on behalf of the 6LNs that have registered 491 addresses on its LLN interfaces without the need of a broadcast over 492 the wireless medium. 494 WiND is also compatible with DHCPv6 and other forms of address 495 assignment in which case it can still be used for DAD. 497 4.2. Links and Link-Local Addresses 499 For Link-Local Addresses, DAD is performed between communicating 500 pairs of nodes. It is carried out as part of a registration process 501 that is based on a NS/NA exchange that transports an EARO. During 502 that process, the DAD is validated and a Neighbor Cache Entry (NCE) 503 is populated with a single unicast exchange. 505 Ior instance, in the case of a Bluetooth Low Energy (BLE) 506 [RFC7668][IEEEstd802151] Hub-and Spoke configuration, Uniqueness of 507 Link local Addresses need only to be verified between the pairs of 508 communicating nodes, a central router and a peripheral host. In that 509 example, 2 peripheral hosts connected to the same central router can 510 not have the same Link Local Address because the Binding Cache 511 Entries (BCEs) would collide at the central router which could not 512 talk to both over the same interface. The WiND operation is 513 appropriate for that DAD operation, but the one from ND is not, 514 because peripheral hosts are not on the same broadcast domain. On 515 the other hand, Global and ULA DAD is validated at the Subnet Level, 516 using a registrar hosted by the central router. 518 4.3. Subnets and Global Addresses 520 WiND extends IPv6 ND for Hub-and-Spoke (e.g., BLE) and Route-Over 521 (e.g., RPL) Multi-Link Subnets (MLSNs). 523 In the Hub-and-Spoke case, each Hub-Spoke pair is a distinct IP Link, 524 and a Subnet can be mapped on a collection of Links that are 525 connected to the Hub. The Subnet prefix is associated to the Hub. 526 Acting as 6LR, the Hub advertises the prefix as not-onlink to the 527 spokes in RA messages Prefix Information Options (PIO). Acting as 528 6LNs, the Spokes autoconfigure addresses from that prefix and 529 register them to the Hub with a corresponding lifetime. Acting as a 530 6LBR, the Hub maintains a binding table of all the registered IP 531 addresses and rejects duplicate registrations, thus ensuring a DAD 532 protection for a registered address even if the registering node is 533 sleeping. Acting as 6LR, the Hub also maintains an NCE for the 534 registered addresses and can deliver a packet to any of them for 535 their respective lifetimes. It can be observed that this design 536 builds a form of Layer-3 Infrastructure BSS. 538 A Route-Over MLSN is considered as a collection of Hub-and-Spoke 539 where the Hubs form a connected dominating set of the member nodes of 540 the Subnet, and IPv6 routing takes place between the Hubs within the 541 Subnet. A single logical 6LBR is deployed to serve the whole mesh. 542 The registration in [RFC8505] is abstract to the routing protocol and 543 provides enough information to feed a routing protocol such as RPL as 544 specified in [I-D.thubert-roll-unaware-leaves]. In a degraded mode, 545 all the Hubs are connected to a same high speed backbone such as an 546 Ethernet bridging domain where IPv6 ND is operated. In that case, it 547 is possible to federate the Hub, Spoke and Backbone nodes as a single 548 Subnet, operating IPv6 ND proxy operations 549 [I-D.ietf-6lo-backbone-router] at the Hubs, acting as 6BBRs. It can 550 be observed that this latter design builds a form of Layer-3 551 Infrastructure ESS. 553 5. WiND Applicability 555 WiND allows P2P, P2MP hub-and spoke, MAC-level broadcast domain 556 emulation such as mesh-under and Wi-Fi BSS, and Route-Over meshes. 558 There is an intersection where Link and Subnet are congruent and 559 where both ND and WiND could apply. These includes P2P, the MAC 560 emulation of a PHY broadcast domain, and the particular case of 561 always on, fully overlapping physical radio broadcast domain. But 562 even in those cases where both are possible, WiND is preferable vs. 563 ND because it reduces the need of broadcast ( this is discussed in 564 the introduction of [I-D.ietf-6lo-backbone-router]). 566 There are also numerous practical use cases in the wireless world 567 where Links and Subnets are not P2P and not congruent: 569 o IEEE std 802.11 infrastructure BSS enables one subnet per AP, and 570 emulates a broadcast domain at L2. Infra ESS extends that and 571 recommends to use an IPv6 ND proxy [IEEEstd80211] to coexist with 572 Ethernet connected nodes. WiND incorporates an ND proxy to serve 573 that need and that was missing so far. 575 o BlueTooth is Hub-and-Spoke at the MAC layer. It would make little 576 sense to configure a different subnet between the central and each 577 individual peripheral node (e.g., sensor). Rather, [RFC7668] 578 allocates a prefix to the central node acting as router (6LR), and 579 each peripheral host (acting as a host (6LR) forms one or more 580 address(es) from that same prefix and registers it. 582 o A typical Smartgrid networks puts together Route-Over MLSNs that 583 comprise thousands of IPv6 nodes. The 6TiSCH architecture 584 [I-D.ietf-6tisch-architecture] presents the Route-Over model over 585 a [IEEEstd802154] Time-Slotted Channel-Hopping mesh, and 586 generalizes it for multiple other applications. Each node in a 587 Smartgrid network may have tens to a hundred others nodes in 588 range. A key problem for the routing protocol is which other 589 node(s) should this node peer with, because most of the possible 590 peers do not provide added routing value. When both energy and 591 bandwidth are constrained,talking to them is a bad idea and most 592 of the possible P2P links are not even used. Peerings that are 593 actually used come and go with the dynamics of radio signal 594 propagation. It results that allocating prefixes to all the 595 possible P2P Links and maintain as many addresses in all nodes is 596 not even considered. 598 5.1. Case of LPWANs 600 LPWANs are by nature so constrained that the addresses and Subnets 601 are fully pre-configured and operate as P2P or Hub-and-Spoke. This 602 saves the steps of neighbor Discovery and enables a very efficient 603 stateful compression of the IPv6 header. 605 5.2. Case of Infrastructure BSS and ESS 607 In contrast to IPv4, IPv6 enables a node to form multiple addresses, 608 some of them temporary to elusive, and with a particular attention 609 paid to privacy. Addresses may be formed and deprecated 610 asynchronously to the association. Even if the knowledge of IPv6 611 addresses used by a STA can be obtained by snooping protocols such as 612 IPv6 ND and DHCPv6, or by observing data traffic sourced at the STA, 613 such methods provide only an imperfect knowledge of the state of the 614 STA at the AP. This may result in a loss of connectivity for some 615 IPv6 addresses, in particular for addresses rarely used and in a 616 situation of mobility. This may also result in undesirable remanent 617 state in the AP when a STA ceases to use an IPv6 address. It results 618 that snooping protocols is not a recommended technique and that it 619 should only be used as last resort. 621 The recommended alternate is to use the IPv6 Registration method 622 speficied in p. By that method, the AP exposes its capability to 623 proxy ND to the STA in Router Advertisement messages. In turn, the 624 STA may request proxy ND services from the AP for one or more IPv6 625 addresses, using an Address Registration Option. The Registration 626 state has a lifetime that limits unwanted state remanence in the 627 network. The registration is optionally secured using 628 [I-D.ietf-6lo-ap-nd] to prevent address theft and impersonation. The 629 registration carries a sequence number, which enables a fast mobility 630 without a loss of connectivity. 632 The ESS mode requires a proxy ND operation at the AP. The proxy ND 633 operation must cover Duplicate Address Detection, Neighbor 634 Unreachability Detection, Address Resolution and Address Mobility to 635 transfer a role of ND proxy to the AP where a STA is associated 636 following the mobility of the STA. The proxy ND specification 637 associated to the address registration is 638 [I-D.ietf-6lo-backbone-router]. With that specification, the AP 639 participates to the protocol as a Backbone Router, typically 640 operating as a bridging proxy though the routing proxy operation is 641 also possible. As a bridging proxy, the proxy replies to NS lookups 642 with the MAC address of the STA, and then bridges packets to the STA 643 normally; as a routing proxy, it replies with its own MAC address and 644 then routes to the STA at the IP layer. The routing proxy reduces 645 the need to expose the MAC address of the STA on the wired side, for 646 a better stability and scalability of the bridged fabric. 648 5.3. Case of Mesh Under Technologies 650 The Mesh-Under provides a broadcast domain emulation with reflexive 651 and Transitive properties and defines a transit Link for IPv6 652 operations. It results that the model for IPv6 operation is similar 653 to that of a BSS, with the root of the mesh operating an Access Point 654 does in a BSS/ESS. While it is still possible to operate IPv6 ND, 655 the inefficiencies of the flooding operation make the IPv6 ND 656 operations even less desirable than in a BSS, and the use of WiND is 657 highly recommended. 659 5.4. Case of DMC radios 661 IPv6 over DMC radios uses P2P Links that can be formed and maintained 662 when a pair of DMC radios transmitters are in range from one another. 664 5.4.1. Using IPv6 ND only 666 DMC radios do not provide MAC level broadcast emulation. An example 667 of that is OCB (outside the context of a BSS), which uses IEEE Std. 668 802.11 transmissions but does not provide the BSS functions. 670 It is possible to form P2P IP Links between each individual pairs of 671 nodes and operate IPv6 ND over those Links with Link Local addresses. 672 DAD must be performed for all addresses on all P2P IP Links. 674 If special deployment care is taken so that the physical broadcast 675 domains of a collection of the nodes fully overlap, then it is also 676 possible to build an IP Subnet within that collection of nodes and 677 operate IPv6 ND. 679 The model can be stretched beyond the scope of IPv6 ND if an external 680 mechanism avoids duplicate addresses and if the deployment ensures 681 the connectivity between peers. This can be achieved for instance in 682 a Hub-and-Spoke deployment if the Hub is the only router in the 683 Subnet and the Prefix is advertised as not onlink. 685 5.4.2. Using Wireless ND 687 Though this can be achieved with IPv6 ND, WiND is the recommended 688 approach since it uses more unicast communications which are more 689 reliable and less impacting for other users of the medium. 691 Router and Hosts respectively send a compressed RA/NA with a SLLAO at 692 a regular period. The period can be indicated in a RA as in an RA- 693 Interval Option [RFC6275]. If available, the message can be 694 transported in a compressed form in a beacon, e.g., in OCB Basic 695 Safety Messages (BSM) that are nominally sent every 100ms. An active 696 beaconing mode is possible whereby the Host sends broadcast RS 697 messages to which a router can answer with a unicast RA. 699 A router that has Internet connectivity and is willing to serve as an 700 Internet Access may advertise itself as a default router [RFC4191] in 701 its RA. The NA/RA is sent over an Unspecified Link where it does not 702 conflict to anyone, so DAD is not necessary at that stage. 704 The receiver instantiates a Link where the sender's address is not a 705 duplicate. To achieve this, it forms an LLA that does not conflict 706 with that of the sender and registers to the sender using [RFC8505]. 707 If the sender sent an RA(PIO) the receiver can also autoconfigure an 708 address from the advertised prefix and register it. 710 6LoWPAN Node 6LR 711 (RPL leaf) (router) 712 | | 713 | LLN link | 714 | | 715 | IPv6 ND RS | 716 |-------------->| 717 |-----------> | 718 |------------------> 719 | IPv6 ND RA | 720 |<--------------| 721 | | 722 | NS(EARO) | 723 |-------------->| 724 | | 725 | NA(EARO) | 726 |<--------------| 727 | | 729 Figure 1: Initial Registration Flow 731 The lifetime in the registration should start with a small value 732 (X=RMin, TBD), and exponentially grow with each reregistration to a 733 mlarger value (X=Rmax, TBD). The IP Link is considered down when 734 (X=NbBeacons, TDB) expected messages are not received in a row. It 735 must be noted that the Link flapping does not affect the state of the 736 registration and when a Link comes back up, the active -lifetime not 737 elapsed- registrations are still usable. Packets should be held or 738 destroyed when the Link is down. 740 P2P Links may be federated in Hub-and-Spoke and then in Route-Over 741 MLSNs as described above. More details on the operation of WiND and 742 RPL over the MLSN can be found in section 3.1, 3.2, 4.1 and 4.2.2 of 743 [I-D.ietf-6tisch-architecture]. 745 6LoWPAN Node 6LR 6LBR 6BBR 746 (RPL leaf) (router) (root) 747 | | | | 748 | 6LoWPAN ND |6LoWPAN ND+RPL | 6LoWPAN ND | IPv6 ND 749 | LLN link |Route-Over mesh|Ethernet/serial| Backbone 750 | | | | 751 | IPv6 ND RS | | | 752 |-------------->| | | 753 |-----------> | | | 754 |------------------> | | 755 | IPv6 ND RA | | | 756 |<--------------| | | 757 | | | | 758 | NS(EARO) | | | 759 |-------------->| | | 760 | 6LoWPAN ND | Extended DAR | | 761 | |-------------->| | 762 | | | NS(EARO) | 763 | | |-------------->| 764 | | | | NS-DAD 765 | | | |------> 766 | | | | (EARO) 767 | | | | 768 | | | NA(EARO) | 769 | | |<--------------| 770 | | Extended DAC | | 771 | |<--------------| | 772 | NA(EARO) | | | 773 |<--------------| | | 774 | | | | 776 Figure 2: Initial Registration Flow over Multi-Link Subnet 778 An example Hub-and-Spoke is an OCB Road-Side Unit (RSU) that owns a 779 prefix, provides Internet connectivity using that prefix to On-Board 780 Units (OBUs) within its physical broadcast domain. An example of 781 Route-Over MLSN is a collection of cars in a parking lot operating 782 RPL to extend the connectivity provided by the RSU beyond its 783 physical broadcast domain. Cars may then operate NEMO [RFC3963] for 784 their own prefix using their address derived from the prefix of the 785 RSU as CareOf Address. 787 6. IANA Considerations 789 This specification does not require IANA action. 791 7. Security Considerations 793 This specification refers to the security sections of IPv6 ND and 794 WiND, respectively. 796 8. Acknowledgments 798 Many thanks to the participants of the 6lo WG where a lot of the work 799 discussed here happened. Also ROLL, 6TiSCH, and 6LoWPAN. 801 9. References 803 9.1. Normative References 805 [I-D.ietf-6lo-ap-nd] 806 Thubert, P., Sarikaya, B., Sethi, M., and R. Struik, 807 "Address Protected Neighbor Discovery for Low-power and 808 Lossy Networks", draft-ietf-6lo-ap-nd-12 (work in 809 progress), April 2019. 811 [I-D.ietf-6lo-backbone-router] 812 Thubert, P., Perkins, C., and E. Levy-Abegnoli, "IPv6 813 Backbone Router", draft-ietf-6lo-backbone-router-11 (work 814 in progress), February 2019. 816 [RFC3963] Devarapalli, V., Wakikawa, R., Petrescu, A., and P. 817 Thubert, "Network Mobility (NEMO) Basic Support Protocol", 818 RFC 3963, DOI 10.17487/RFC3963, January 2005, 819 . 821 [RFC4191] Draves, R. and D. Thaler, "Default Router Preferences and 822 More-Specific Routes", RFC 4191, DOI 10.17487/RFC4191, 823 November 2005, . 825 [RFC4861] Narten, T., Nordmark, E., Simpson, W., and H. Soliman, 826 "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, 827 DOI 10.17487/RFC4861, September 2007, 828 . 830 [RFC4862] Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless 831 Address Autoconfiguration", RFC 4862, 832 DOI 10.17487/RFC4862, September 2007, 833 . 835 [RFC6275] Perkins, C., Ed., Johnson, D., and J. Arkko, "Mobility 836 Support in IPv6", RFC 6275, DOI 10.17487/RFC6275, July 837 2011, . 839 [RFC8200] Deering, S. and R. Hinden, "Internet Protocol, Version 6 840 (IPv6) Specification", STD 86, RFC 8200, 841 DOI 10.17487/RFC8200, July 2017, 842 . 844 [RFC8505] Thubert, P., Ed., Nordmark, E., Chakrabarti, S., and C. 845 Perkins, "Registration Extensions for IPv6 over Low-Power 846 Wireless Personal Area Network (6LoWPAN) Neighbor 847 Discovery", RFC 8505, DOI 10.17487/RFC8505, November 2018, 848 . 850 9.2. Informative References 852 [I-D.bi-savi-wlan] 853 Bi, J., Wu, J., Wang, Y., and T. Lin, "A SAVI Solution for 854 WLAN", draft-bi-savi-wlan-16 (work in progress), November 855 2018. 857 [I-D.ietf-6tisch-architecture] 858 Thubert, P., "An Architecture for IPv6 over the TSCH mode 859 of IEEE 802.15.4", draft-ietf-6tisch-architecture-20 (work 860 in progress), March 2019. 862 [I-D.ietf-mboned-ieee802-mcast-problems] 863 Perkins, C., McBride, M., Stanley, D., Kumari, W., and J. 864 Zuniga, "Multicast Considerations over IEEE 802 Wireless 865 Media", draft-ietf-mboned-ieee802-mcast-problems-05 (work 866 in progress), April 2019. 868 [I-D.ietf-rift-rift] 869 Team, T., "RIFT: Routing in Fat Trees", draft-ietf-rift- 870 rift-05 (work in progress), April 2019. 872 [I-D.thubert-6lo-unicast-lookup] 873 Thubert, P. and E. Levy-Abegnoli, "IPv6 Neighbor Discovery 874 Unicast Lookup", draft-thubert-6lo-unicast-lookup-00 (work 875 in progress), January 2019. 877 [I-D.thubert-roll-unaware-leaves] 878 Thubert, P., "Routing for RPL Leaves", draft-thubert-roll- 879 unaware-leaves-07 (work in progress), April 2019. 881 [I-D.vyncke-6man-mcast-not-efficient] 882 Vyncke, E., Thubert, P., Levy-Abegnoli, E., and A. 883 Yourtchenko, "Why Network-Layer Multicast is Not Always 884 Efficient At Datalink Layer", draft-vyncke-6man-mcast-not- 885 efficient-01 (work in progress), February 2014. 887 [I-D.yourtchenko-6man-dad-issues] 888 Yourtchenko, A. and E. Nordmark, "A survey of issues 889 related to IPv6 Duplicate Address Detection", draft- 890 yourtchenko-6man-dad-issues-01 (work in progress), March 891 2015. 893 [IEEE802154] 894 IEEE standard for Information Technology, "IEEE Std. 895 802.15.4, Part. 15.4: Wireless Medium Access Control (MAC) 896 and Physical Layer (PHY) Specifications for Low-Rate 897 Wireless Personal Area Networks". 899 [IEEEstd8021] 900 IEEE standard for Information Technology, "IEEE Standard 901 for Information technology -- Telecommunications and 902 information exchange between systems Local and 903 metropolitan area networks Part 1: Bridging and 904 Architecture". 906 [IEEEstd80211] 907 IEEE standard for Information Technology, "IEEE Standard 908 for Information technology -- Telecommunications and 909 information exchange between systems Local and 910 metropolitan area networks-- Specific requirements Part 911 11: Wireless LAN Medium Access Control (MAC) and Physical 912 Layer (PHY) Specifications". 914 [IEEEstd802151] 915 IEEE standard for Information Technology, "IEEE Standard 916 for Information Technology - Telecommunications and 917 Information Exchange Between Systems - Local and 918 Metropolitan Area Networks - Specific Requirements. - Part 919 15.1: Wireless Medium Access Control (MAC) and Physical 920 Layer (PHY) Specifications for Wireless Personal Area 921 Networks (WPANs)". 923 [IEEEstd802154] 924 IEEE standard for Information Technology, "IEEE Standard 925 for Local and metropolitan area networks -- Part 15.4: 926 Low-Rate Wireless Personal Area Networks (LR-WPANs)". 928 [RFC4291] Hinden, R. and S. Deering, "IP Version 6 Addressing 929 Architecture", RFC 4291, DOI 10.17487/RFC4291, February 930 2006, . 932 [RFC4903] Thaler, D., "Multi-Link Subnet Issues", RFC 4903, 933 DOI 10.17487/RFC4903, June 2007, 934 . 936 [RFC6550] Winter, T., Ed., Thubert, P., Ed., Brandt, A., Hui, J., 937 Kelsey, R., Levis, P., Pister, K., Struik, R., Vasseur, 938 JP., and R. Alexander, "RPL: IPv6 Routing Protocol for 939 Low-Power and Lossy Networks", RFC 6550, 940 DOI 10.17487/RFC6550, March 2012, 941 . 943 [RFC6775] Shelby, Z., Ed., Chakrabarti, S., Nordmark, E., and C. 944 Bormann, "Neighbor Discovery Optimization for IPv6 over 945 Low-Power Wireless Personal Area Networks (6LoWPANs)", 946 RFC 6775, DOI 10.17487/RFC6775, November 2012, 947 . 949 [RFC7668] Nieminen, J., Savolainen, T., Isomaki, M., Patil, B., 950 Shelby, Z., and C. Gomez, "IPv6 over BLUETOOTH(R) Low 951 Energy", RFC 7668, DOI 10.17487/RFC7668, October 2015, 952 . 954 [RFC8273] Brzozowski, J. and G. Van de Velde, "Unique IPv6 Prefix 955 per Host", RFC 8273, DOI 10.17487/RFC8273, December 2017, 956 . 958 Author's Address 960 Pascal Thubert (editor) 961 Cisco Systems, Inc 962 Building D 963 45 Allee des Ormes - BP1200 964 MOUGINS - Sophia Antipolis 06254 965 FRANCE 967 Phone: +33 497 23 26 34 968 Email: pthubert@cisco.com