0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Type (=137)  |  Code (=0/1)  |          Checksum             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                   Reserved                    | Prefix Length |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      +                                                               +
      |                                                               |
      +                       Target Address                          +
      |                                                               |
      +                                                               +
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      +                                                               +
      |                                                               |
      +                     Destination Address                       +
      |                                                               |
      +                                                               +
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |   Options ...
      +-+-+-+-+-+-+-+-+-+-+-+-

Figure 8: AERO Redirect/Predirect Message Format

3.17.4. Sending Predirects

When a Client forwards a packet with a source address from one of its ACPs toward a destination address covered by an ASP (i.e., toward another AERO Client connected to the same AERO link), the source Client MAY send a Predirect message forward toward the destination Client via the Server.

In the reference operational scenario, when Client ('C1') forwards a packet toward Client ('C2'), it MAY also send a Predirect message forward toward Client ('C2'), subject to rate limiting (see Section 8.2 of [RFC4861]). Client ('C1') prepares the Predirect message as follows:

• the link-layer source address is set to 'L2(C1)' (i.e., the link-layer address of Client ('C1')).
• the link-layer destination address is set to 'L2(S1)' (i.e., the link-layer address of Server ('S1')).
• the network-layer source address is set to fe80::2001:db8:0:0 (i.e., the AERO address of Client ('C1')).
• the network-layer destination address is set to fe80::2001:db8:1:0 (i.e., the AERO address of Client ('C2')).
• the Type is set to 137.
• the Code is set to 1 to indicate "Predirect".
• the Prefix Length is set to the length of the prefix to be assigned to the Target Address.
• the Target Address is set to fe80::2001:db8:0:0 (i.e., the AERO address of Client ('C1')).
• the Destination Address is set to the source address of the originating packet that triggered the Predirection event. (If the originating packet is an IPv4 packet, the address is constructed in IPv4-compatible IPv6 address format).
• the message includes one or more TLLAOs with Link ID and DSCPs set to appropriate values for Client ('C1')'s underlying interfaces, and with UDP Port Number and IP Address set to 0'.
• the message SHOULD include a Timestamp option and a Nonce option.
• the message includes a Redirected Header Option (RHO) that contains the originating packet truncated if necessary to ensure that at least the network-layer header is included but the size of the message does not exceed 1280 bytes.

Note that the act of sending Predirect messages is cited as "MAY", since Client ('C1') may have advanced knowledge that the direct path to Client ('C2') would be unusable or otherwise undesirable. If the direct path later becomes unusable after the initial route optimization, Client ('C1') simply allows packets to again flow through Server ('S1').

3.17.5. Re-encapsulating and Relaying Predirects

When Server ('S1') receives a Predirect message from Client ('C1'), it first verifies that the TLLAOs in the Predirect are a proper subset of the Link IDs in Client ('C1')'s neighbor cache entry. If the Client's TLLAOs are not acceptable, Server ('S1') discards the message. Otherwise, Server ('S1') validates the message according to the Redirect message validation rules in Section 8.1 of [RFC4861], except that the Predirect has Code=1. Server ('S1') also verifies that Client ('C1') is authorized to use the Prefix Length in the Predirect when applied to the AERO address in the network-layer source address by searching for the AERO address in the neighbor cache. If validation fails, Server ('S1') discards the Predirect; otherwise, it copies the correct UDP Port numbers and IP Addresses for Client ('C1')'s links into the (previously empty) TLLAOs.

Server ('S1') then examines the network-layer destination address of the Predirect to determine the next hop toward Client ('C2') by searching for the AERO address in the neighbor cache. Since Client ('C2') is not one of its neighbors, Server ('S1') re-encapsulates the Predirect and relays it via Relay ('R1') by changing the link-layer source address of the message to 'L2(S1)' and changing the link-layer destination address to 'L2(R1)'. Server ('S1') finally forwards the re-encapsulated message to Relay ('R1') without decrementing the network-layer TTL/Hop Limit field.

When Relay ('R1') receives the Predirect message from Server ('S1') it determines that Server ('S2') is the next hop toward Client ('C2') by consulting its forwarding table. Relay ('R1') then re-encapsulates the Predirect while changing the link-layer source address to 'L2(R1)' and changing the link-layer destination address to 'L2(S2)'. Relay ('R1') then relays the Predirect via Server ('S2').

When Server ('S2') receives the Predirect message from Relay ('R1') it determines that Client ('C2') is a neighbor by consulting its neighbor cache. Server ('S2') then re-encapsulates the Predirect while changing the link-layer source address to 'L2(S2)' and changing the link-layer destination address to 'L2(C2)'. Server ('S2') then forwards the message to Client ('C2').

3.17.6. Processing Predirects and Sending Redirects

When Client ('C2') receives the Predirect message, it accepts the Predirect only if the message has a link-layer source address of one of its Servers (e.g., L2(S2)). Client ('C2') further accepts the message only if it is willing to serve as a redirection target. Next, Client ('C2') validates the message according to the Redirect message validation rules in Section 8.1 of [RFC4861], except that it accepts the message even though Code=1 and even though the network-layer source address is not that of it's current first-hop router.

In the reference operational scenario, when Client ('C2') receives a valid Predirect message, it either creates or updates a dynamic neighbor cache entry that stores the Target Address of the message as the network-layer address of Client ('C1') , stores the link-layer addresses found in the TLLAOs as the link-layer addresses of Client ('C1') and stores the Prefix Length as the length to be applied to the network-layer address for forwarding purposes. Client ('C2') then sets AcceptTime for the neighbor cache entry to ACCEPT_TIME.

After processing the message, Client ('C2') prepares a Redirect message response as follows:

• the link-layer source address is set to 'L2(C2)' (i.e., the link-layer address of Client ('C2')).
• the link-layer destination address is set to 'L2(S2)' (i.e., the link-layer address of Server ('S2')).
• the network-layer source address is set to fe80::2001:db8:1:0 (i.e., the AERO address of Client ('C2')).
• the network-layer destination address is set to fe80::2001:db8:0:0 (i.e., the AERO address of Client ('C1')).
• the Type is set to 137.
• the Code is set to 0 to indicate "Redirect".
• the Prefix Length is set to the length of the prefix to be applied to the Target Address.
• the Target Address is set to fe80::2001:db8:1:0 (i.e., the AERO address of Client ('C2')).
• the Destination Address is set to the destination address of the originating packet that triggered the Redirection event. (If the originating packet is an IPv4 packet, the address is constructed in IPv4-compatible IPv6 address format).
• the message includes one or more TLLAOs with Link ID and DSCPs set to appropriate values for Client ('C2')'s underlying interfaces, and with UDP Port Number and IP Address set to '0'.
• the message SHOULD include a Timestamp option and MUST echo the Nonce option received in the Predirect (i.e., if a Nonce option is included).
• the message includes as much of the RHO copied from the corresponding Predirect message as possible such that at least the network-layer header is included but the size of the message does not exceed 1280 bytes.

After Client ('C2') prepares the Redirect message, it sends the message to Server ('S2').

3.17.7. Re-encapsulating and Relaying Redirects

When Server ('S2') receives a Redirect message from Client ('C2'), it first verifies that the TLLAOs in the Redirect are a proper subset of the Link IDs in Client ('C2')'s neighbor cache entry. If the Client's TLLAOs are not acceptable, Server ('S2') discards the message. Otherwise, Server ('S2') validates the message according to the Redirect message validation rules in Section 8.1 of [RFC4861]. Server ('S2') also verifies that Client ('C2') is authorized to use the Prefix Length in the Redirect when applied to the AERO address in the network-layer source address by searching for the AERO address in the neighbor cache. If validation fails, Server ('S2') discards the Redirect; otherwise, it copies the correct UDP Port numbers and IP Addresses for Client ('C2')'s links into the (previously empty) TLLAOs.

Server ('S2') then examines the network-layer destination address of the Redirect to determine the next hop toward Client ('C1') by searching for the AERO address in the neighbor cache. Since Client ('C1') is not a neighbor, Server ('S2') re-encapsulates the Redirect and relays it via Relay ('R1') by changing the link-layer source address of the message to 'L2(S2)' and changing the link-layer destination address to 'L2(R1)'. Server ('S2') finally forwards the re-encapsulated message to Relay ('R1') without decrementing the network-layer TTL/Hop Limit field.

When Relay ('R1') receives the Redirect message from Server ('S2') it determines that Server ('S1') is the next hop toward Client ('C1') by consulting its forwarding table. Relay ('R1') then re-encapsulates the Redirect while changing the link-layer source address to 'L2(R1)' and changing the link-layer destination address to 'L2(S1)'. Relay ('R1') then relays the Redirect via Server ('S1').

When Server ('S1') receives the Redirect message from Relay ('R1') it determines that Client ('C1') is a neighbor by consulting its neighbor cache. Server ('S1') then re-encapsulates the Redirect while changing the link-layer source address to 'L2(S1)' and changing the link-layer destination address to 'L2(C1)'. Server ('S1') then forwards the message to Client ('C1').

3.17.8. Processing Redirects

When Client ('C1') receives the Redirect message, it accepts the message only if it has a link-layer source address of one of its Servers (e.g., ''L2(S1)'). Next, Client ('C1') validates the message according to the Redirect message validation rules in Section 8.1 of [RFC4861], except that it accepts the message even though the network-layer source address is not that of it's current first-hop router. Following validation, Client ('C1') then processes the message as follows.

In the reference operational scenario, when Client ('C1') receives the Redirect message, it either creates or updates a dynamic neighbor cache entry that stores the Target Address of the message as the network-layer address of Client ('C2'), stores the link-layer addresses found in the TLLAOs as the link-layer addresses of Client ('C2') and stores the Prefix Length as the length to be applied to the network-layer address for forwarding purposes. Client ('C1') then sets ForwardTime for the neighbor cache entry to FORWARD_TIME.

Now, Client ('C1') has a neighbor cache entry with a valid ForwardTime value, while Client ('C2') has a neighbor cache entry with a valid AcceptTime value. Thereafter, Client ('C1') may forward ordinary network-layer data packets directly to Client ('C2') without involving any intermediate nodes, and Client ('C2') can verify that the packets came from an acceptable source. (In order for Client ('C2') to forward packets to Client ('C1'), a corresponding Predirect/Redirect message exchange is required in the reverse direction; hence, the mechanism is asymmetric.)

3.17.9. Server-Oriented Redirection

In some environments, the Server nearest the target Client may need to serve as the redirection target, e.g., if direct Client-to-Client communications are not possible. In that case, the Server prepares the Redirect message the same as if it were the destination Client (see: Section 3.17.6), except that it writes its own link-layer address in the TLLAO option. The Server must then maintain a dynamic neighbor cache entry for the redirected source Client.

3.17.10. Route Optimization Policy

Although the Client is responsible for initiating route optimization through the transmission of Predirect messages, the Server is the policy enforcement point that determines whether route optimization is permitted. For example, on some AERO links route optimization would allow traffic to circumvent critical network-based traffic interception points. In those cases, the Server can deny route optimization requests by simply discarding any Predirect messages instead of forwarding them.

3.17.11. Route Optimization and Multiple ACPs

Clients that receive multiple non-contiguous ACP delegations must perform route optimization for each of the individual ACPs based on demand of traffic with source addresses taken from those prefixes. For example, if Client C1 has already performed route optimization for destination ACP X on behalf of its source ACP Y, it must also perform route optimization for X on behalf of its source ACP Z. As a result, source route optimization state cannot be shared between non-contiguous ACPs and must be managed separately.

3.18. Neighbor Unreachability Detection (NUD)

AERO nodes perform Neighbor Unreachability Detection (NUD) by sending unicast NS messages to elicit solicited NA messages from neighbors the same as described in [RFC4861]. NUD is performed either reactively in response to persistent L2 errors (see Section 3.14) or proactively to test existing neighbor cache entries.

When an AERO node sends an NS/NA message, it MUST use its link-local address as the IPv6 source address and the link-local address of the neighbor as the IPv6 destination address. When an AERO node receives an NS message or a solicited NA message, it accepts the message if it has a neighbor cache entry for the neighbor; otherwise, it ignores the message.

When a source Client is redirected to a target Client it SHOULD proactively test the direct path by sending an initial NS message to elicit a solicited NA response. While testing the path, the source Client can optionally continue sending packets via the Server, maintain a small queue of packets until target reachability is confirmed, or (optimistically) allow packets to flow directly to the target. The source Client SHOULD thereafter continue to test the direct path to the target Client (see Section 7.3 of [RFC4861]) periodically in order to keep dynamic neighbor cache entries alive.

In particular, while the source Client is actively sending packets to the target Client it SHOULD also send NS messages separated by RETRANS_TIMER milliseconds in order to receive solicited NA messages. If the source Client is unable to elicit a solicited NA response from the target Client after MAX_RETRY attempts, it SHOULD set ForwardTime to 0 and resume sending packets via one of its Servers. Otherwise, the source Client considers the path usable and SHOULD thereafter process any link-layer errors as a hint that the direct path to the target Client has either failed or has become intermittent.

When ForwardTime for a dynamic neighbor cache entry expires, the source Client resumes sending any subsequent packets via a Server and may (eventually) attempt to re-initiate the AERO redirection process. When AcceptTime for a dynamic neighbor cache entry expires, the target Client discards any subsequent packets received directly from the source Client. When both ForwardTime and AcceptTime for a dynamic neighbor cache entry expire, the Client deletes the neighbor cache entry.

3.19. Mobility Management

3.19.1. Announcing Link-Layer Address Changes

When a Client needs to change its link-layer address, e.g., due to a mobility event, it issues an immediate Rebind to each of its Servers using the new link-layer address as the source address and with an ALREQ that includes the correct Link ID and DSCP values. If authentication succeeds, the Server then updates its neighbor cache and sends a Reply. Note that if the Client does not issue a Rebind before the prefix delegation lifetime expires (e.g., if the Client has been out of touch with the Server for a considerable amount of time), the Server's Reply will report NoBinding and the Client must re-initiate the DHCPv6 PD procedure.

Next, the Client sends Predirect messages to each of its correspondent Client neighbors using the same procedures as specified in Section 3.17.4. The Client sends the Predirect messages via a Server the same as if it was performing the initial route optimization procedure with the correspondent. The Predirect message will update the correspondent' link layer address mapping for the Client.

3.19.2. Bringing New Links Into Service

When a Client needs to bring a new underlying interface into service (e.g., when it activates a new data link), it issues an immediate Rebind to each of its Servers using the new link-layer address as the source address and with an ALREQ that includes the new Link ID and DSCP values. If authentication succeeds, the Server then updates its neighbor cache and sends a Reply. The Client MAY then send Predirect messages to each of its correspondent Clients to inform them of the new link-layer address as described in Section 3.19.1.

3.19.3. Removing Existing Links from Service

When a Client needs to remove an existing underlying interface from service (e.g., when it de-activates an existing data link), it issues an immediate Rebind to each of its Servers over any available link with an ALDEL that includes the deprecated Link ID. If authentication succeeds, the Server then updates its neighbor cache and sends a Reply. The Client SHOULD then send Predirect messages to each of its correspondent Clients to inform them of the deprecated link-layer address as described in Section 3.19.1.

3.19.4. Moving to a New Server

When a Client associates with a new Server, it performs the Client procedures specified in Section 3.15.2.

When a Client disassociates with an existing Server, it sends a Release message via a new Server to the unicast link-local network layer address of the old Server. The new Server then writes its own link-layer address in the Release message IP source address and forwards the message to the old Server.

When the old Server receives the Release, it first authenticates the message. Next, it resets the Client's neighbor cache entry lifetime to 3 seconds, rewrites the link-layer address in the neighbor cache entry to the address of the new Server, then returns a Reply message to the Client via the old Server. When the lifetime expires, the old Server withdraws the IP route from the AERO routing system and deletes the neighbor cache entry for the Client. The Client can then use the Reply message to verify that the termination signal has been processed, and can delete both the default route and the neighbor cache entry for the old Server. (Note that since Release/Reply messages may be lost in the network the Client MUST retry until it gets a Reply indicating that the Release was successful. If the Client does not receive a Reply after MAX_RETRY attempts, the old Server may have failed and the Client should discontinue its Release attempts.)

Clients SHOULD NOT move rapidly between Servers in order to avoid causing excessive oscillations in the AERO routing system. Such oscillations could result in intermittent reachability for the Client itself, while causing little harm to the network. Examples of when a Client might wish to change to a different Server include a Server that has gone unreachable, topological movements of significant distance, etc.

3.19.5. Packet Queueing for Mobility

AERO Clients and Servers should maintain a samll queue of packets they have recently sent to an AERO neighbor, e.g., a 1 second window. If the AERO neighbor moves, the AERO node MAY retransmit the queued packets to ensure that they are delviered to the AERO neighbor's new location.

Note that this may have performance implications for asymmetric paths. For example, if the AERO neighbor moves from a 50mbps link to a 128kbps link, retransmitting a 1 second window could lead to significant congestion.

3.20. Proxy AERO

Proxy Mobile IPv6 (PMIPv6) [RFC5213][RFC5844][RFC5949] presents a network-based localized mobility management scheme for use within an access network domain. It is typically used in WiFi and cellular wireless access networks, and allows Mobile Nodes (MNs) to receive and retain an IP address that remains stable within the access network domain without needing to implement any special mobility protocols. In the PMIPv6 architecture, access network devices known as Mobility Access Gateways (MAGs) provide MNs with an access link abstraction and receive prefixes for the MNs from a Local Mobility Anchor (LMA).

In a proxy AERO domain, a proxy AERO Client (acting as a MAG) can similarly provide proxy services for MNs that do not participate in AERO messaging. The proxy Client presents an access link abstraction to MNs, and performs DHCPv6 PD exchanges over the AERO interface with an AERO Server (acting as an LMA) to receive ACPs for address provisioning of new MNs that come onto an access link. This scheme assumes that proxy Clients act as fixed (non-mobile) infrastructure elements under the same administrative trust basis as for Relays and Servers.

When an MN comes onto an access link within a proxy AERO domain for the first time, the proxy Client authenticates the MN and obtains a unique identifier that it can use as a DHCPv6 DUID then sends a Solicit message to its Server. When the Server delegates an ACP and returns a Reply, the proxy Client creates an AERO address for the MN and assigns the ACP to the MN's access link. The proxy Client then configures itself as a default router for the MN and provides address autoconfiguration services (e.g., SLAAC, DHCPv6, DHCPv4, etc.) for provisioning MN addresses from the ACP over the access link. Since the proxy Client may serve many such MNs simultaneously, it may receive multiple ACP prefix delegations and configure multiple AERO addresses, i.e., one for each MN.

When two MNs are associated with the same proxy Client, the Client can forward traffic between the MNs without involving a Server since it configures the AERO addresses of both MNs and therefore also has the necessary routing information. When two MNs are associated with different proxy Clients, the source MN's Client can initiate standard AERO intradomain route optimization to discover a direct path to the target MN's Client through the exchange of Predirect/Redirect messages.

When an MN in a proxy AERO domain leaves an access link provided by an old proxy Client, the MN issues an access link-specific "leave" message that informs the old Client of the link-layer address of a new Client on the planned new access link. This is known as a "predictive handover". When an MN comes onto an access link provided by a new proxy Client, the MN issues an access link-specific "join" message that informs the new Client of the link-layer address of the old Client on the actual old access link. This is known as a "reactive handover".

Upon receiving a predictive handover indication, the old proxy Client sends a Solicit message directly to the new Client and queues any arriving data packets addressed to the departed MN. The Solicit message includes the MN's ID as the DUID, the ACP in an IA_PD option, the old Client's address as the link-layer source address and the new Client's address as the link-layer destination address. When the new Client receives the Solicit message, it changes the link-layer source address to its own address, changes the link-layer destination address to the address of its Server, and forwards the message to the Server. At the same time, the new Client creates access link state for the ACP in anticipation of the MN's arrival (while queuing any data packets until the MN arrives), creates a neighbor cache entry for the old Client with AcceptTime set to ACCEPT_TIME, then sends a Redirect message back to the old Client. When the old Client receives the Redirect message, it creates a neighbor cache entry for the new Client with ForwardTime set to FORWARD_TIME, then forwards any queued data packets to the new Client. At the same time, the old Client sends a Release message to its Server. Finally, the old Client sends unsolicited Redirect messages to any of the ACP's correspondents with a TLLAO containing the link-layer address of the new Client.

Upon receiving a reactive handover indication, the new proxy Client creates access link state for the MN's ACP, sends a Solicit message to its Server, and sends a Release message directly to the old Client. The Release message includes the MN's ID as the DUID, the ACP in an IA_PD option, the new Client's address as the link-layer source address and the old Client's address as the link-layer destination address. When the old Client receives the Release message, it changes the link-layer source address to its own address, changes the link-layer destination address to the address of its Server, and forwards the message to the Server. At the same time, the old Client sends a Predirect message back to the new Client and queues any arriving data packets addressed to the departed MN. When the new Client receives the Predirect, it creates a neighbor cache entry for the old Client with AcceptTime set to ACCEPT_TIME, then sends a Redirect message back to the old Client. When the old Client receives the Redirect message, it creates a neighbor cache entry for the new Client with ForwardTime set to FORWARD_TIME, then forwards any queued data packets to the new Client. Finally, the old Client sends unsolicited Redirect messages to correspondents the same as for the predictive case.

When a Server processes a Solicit message, it creates a neighbor cache entry for this ACP if none currently exists. If a neighbor cache entry already exists, however, the Server changes the link-layer address to the address of the new proxy Client (this satisfies the case of both the old Client and new Client using the same Server).

When a Server processes a Release message, it resets the neighbor cache entry lifetime for this ACP to 3 seconds if the cached link-layer address matches the old proxy Client's address. Otherwise, the Server ignores the Release message (this satisfies the case of both the old Client and new Client using the same Server).

When a correspondent Client receives an unsolicited Redirect message, it changes the link-layer address for the ACP's neighbor cache entry to the address of the new proxy Client.

From an architectural perspective, in addition to the use of DHCPv6 PD and IPv6 ND signaling the AERO approach differs from PMIPv6 in its use of the NBMA virtual link model instead of point-to-point tunnels. This provides a more agile interface for Client/Server and Client/Client coordinations, and also facilitates simple route optimization. The AERO routing system is also arranged in such a fashion that Clients get the same service from any Server they happen to associate with. This provides a natural fault tolerance and load balancing capability such as desired for distributed mobility management.

3.21. Extending AERO Links Through Security Gateways

When an enterprise mobile device moves from a campus LAN connection to a public Internet link, it must re-enter the enterprise via a security gateway that has both a physical interface connection to the Internet and a physical interface connection to the enterprise internetwork. This most often entails the establishment of a Virtual Private Network (VPN) link over the public Internet from the mobile device to the security gateway. During this process, the mobile device supplies the security gateway with its public Internet address as the link-layer address for the VPN. The mobile device then acts as an AERO Client to negotiate with the security gateway to obtain its ACP.

In order to satisfy this need, the security gateway also operates as an AERO Server with support for AERO Client proxying. In particular, when a mobile device (i.e., the Client) connects via the security gateway (i.e., the Server), the Server provides the Client with an ACP in a DHCPv6 PD exchange the same as if it were attached to an enterprise campus access link. The Server then replaces the Client's link-layer source address with the Server's enterprise-facing link-layer address in all AERO messages the Client sends toward neighbors on the AERO link. The AERO messages are then delivered to other devices on the AERO link as if they were originated by the security gateway instead of by the AERO Client. In the reverse direction, the AERO messages sourced by devices within the enterprise network can be forwarded to the security gateway, which then replaces the link-layer destination address with the Client's link-layer address and replaces the link-layer source address with its own (Internet-facing) link-layer address.

After receiving the ACP, the Client can send IP packets that use an address taken from the ACP as the network layer source address, the Client's link-layer address as the link-layer source address, and the Server's Internet-facing link-layer address as the link-layer destination address. The Server will then rewrite the link-layer source address with the Server's own enterprise-facing link-layer address and rewrite the link-layer destination address with the target AERO node's link-layer address, and the packets will enter the enterprise network as though they were sourced from a device located within the enterprise. In the reverse direction, when a packet sourced by a node within the enterprise network uses a destination address from the Client's ACP, the packet will be delivered to the security gateway which then rewrites the link-layer destination address to the Client's link-layer address and rewrites the link-layer source address to the Server's Internet-facing link-layer address. The Server then delivers the packet across the VPN to the AERO Client. In this way, the AERO virtual link is essentially extended *through* the security gateway to the point at which the VPN link and AERO link are effectively grafted together by the link-layer address rewriting performed by the security gateway. All AERO messaging services (including route optimization and mobility signaling) are therefore extended to the Client.

In order to support this virtual link grafting, the security gateway (acting as an AERO Server) must keep static neighbor cache entries for all of its associated Clients located on the public Internet. The neighbor cache entry is keyed by the AERO Client's AERO address the same as if the Client were located within the enterprise internetwork. The neighbor cache is then managed in all ways as though the Client were an ordinary AERO Client. This includes the AERO IPv6 ND messaging signaling for Route Optimization and Neighbor Unreachability Detection.

Note that the main difference between a security gateway acting as an AERO Server and an enterprise-internal AERO Server is that the security gateway has at least one enterprise-internal physical interface and at least one public Internet physical interface. Conversely, the enterprise-internal AERO Server has only enterprise-internal physical interfaces. For this reason security gateway proxying is needed to ensure that the public Internet link-layer addressing space is kept separate from the enterprise-internal link-layer addressing space. This is afforded through a natural extension of the security association caching already performed for each VPN client by the security gateway.

3.22. Extending IPv6 AERO Links to the Internet

When an IPv6 host ('H1') with an address from an ACP owned by AERO Client ('C1') sends packets to a correspondent IPv6 host ('H2'), the packets eventually arrive at the IPv6 router that owns ('H2')s prefix. This IPv6 router may or may not be an AERO Client ('C2') either within the same home network as ('C1') or in a different home network.

If Client ('C1') is currently located outside the boundaries of its home network, it will connect back into the home network via a security gateway acting as an AERO Server. The packets sent by ('H1') via ('C1') will then be forwarded through the security gateway then through the home network and finally to ('C2') where they will be delivered to ('H2'). This could lead to sub-optimal performance when ('C2') could instead be reached via a more direct route without involving the security gateway.

Consider the case when host ('H1') has the IPv6 address 2001:db8:1::1, and Client ('C1') has the ACP 2001:db8:1::/64 with underlying IPv6 Internet address of 2001:db8:1000::1. Also, host ('H2') has the IPv6 address 2001:db8:2::1, and Client ('C2') has the ACP 2001:db8:2::/64 with underlying IPv6 address of 2001:db8:2000::1. Client ('C1') can determine whether 'C2' is indeed also an AERO Client willing to serve as a route optimization correspondent by resolving the AAAA records for the DNS FQDN that matches ('H2')s prefix, i.e.:

'0.0.0.0.2.0.0.0.8.b.d.0.1.0.0.2.aero.linkupnetworks.net'

If ('C2') is indeed a candidate correspondent, the FQDN lookup will return a PTR resource record that contains the domain name for the AERO link that manages ('C2')s ASP. Client ('C1') can then attempt route optimization using an approach similar to the Return Routability procedure specified for Mobile IPv6 (MIPv6) [RFC6275]. In order to support this process, both Clients MUST intercept and decapsulate packets that have a subnet router anycast address corresponding to any of the /64 prefixes covered by their respective ACPs.

To initiate the process, Client ('C1') creates a specially-crafted encapsulated Predirect message that will be routed through its home network then through ('C2')s home network and finally to ('C2') itself. Client ('C1') prepares the initial message in the exchange as follows:

• The encapsulating IPv6 header source address is set to 2001:db8:1:: (i.e., the IPv6 subnet router anycast address for ('C1')s ACP)
• The encapsulating IPv6 header destination address is set to 2001:db8:2:: (i.e., the IPv6 subnet router anycast address for ('C2')s ACP)
• The encapsulating IPv6 header is followed by any additional encapsulation headers
• The encapsulated IPv6 header source address is set to fe80::2001:db8:1:0 (i.e., the AERO address for ('C1'))
• The encapsulated IPv6 header destination address is set to fe80::2001:db8:2:0 (i.e., the AERO address for ('C2'))
• The encapsulated Predirect message includes all of the securing information that would occur in a MIPv6 "Home Test Init" message (format TBD)

Client ('C1') then further encapsulates the message in the encapsulating headers necessary to convey the packet to the security gateway (e.g., through IPsec encapsulation) so that the message now appears "double-encapsulated". ('C1') then sends the message to the security gateway, which re-encapsulates and forwards it over the home network from where it will eventually reach ('C2').

At the same time, ('C1') creates and sends a second encapsulated Predirect message that will be routed through the IPv6 Internet without involving the security gateway. Client ('C1') prepares the message as follows:

• The encapsulating IPv6 header source address is set to 2001:db8:1000:1 (i.e., the Internet IPv6 address of ('C1'))
• The encapsulating IPv6 header destination address is set to 2001:db8:2:: (i.e., the IPv6 subnet router anycast address for ('C2')s ACP)
• The encapsulating IPv6 header is followed by any additional encapsulation headers
• The encapsulated IPv6 header source address is set to fe80::2001:db8:1:0 (i.e., the AERO address for ('C1'))
• The encapsulated IPv6 header destination address is set to fe80::2001:db8:2:0 (i.e., the AERO address for ('C2'))
• The encapsulated Predirect message includes all of the securing information that would occur in a MIPv6 "Care-of Test Init" message (format TBD)

('C2') will receive both Predirect messages through its home network then return a corresponding Redirect for each of the Predirect messages with the source and destination addresses in the inner and outer headers reversed. The first message includes all of the securing information that would occur in a MIPv6 "Home Test" message, while the second message includes all of the securing information that would occur in a MIPv6 "Care-of Test" message (formats TBD).

When ('C1') receives the Redirect messages, it performs the necessary security procedures per the MIPv6 specification. It then prepares an encapsulated NS message that includes the same source and destination addresses as for the "Care-of Test Init" Predirect message, and includes all of the securing information that would occur in a MIPv6 "Binding Update" message (format TBD) and sends the message to ('C2').

When ('C2') receives the NS message, if the securing information is correct it creates or updates a neighbor cache entry for ('C1') with fe80::2001:db8:1:0 as the network-layer address, 2001:db8:1000::1 as the link-layer address and with AcceptTime set to ACCEPT_TIME. ('C2') then sends an encapsulated NA message back to ('C1') that includes the same source and destination addresses as for the "Care-of Test" Redirect message, and includes all of the securing information that would occur in a MIPv6 "Binding Acknowledgement" message (format TBD) and sends the message to ('C1').

When ('C1') receives the NA message, it creates or updates a neighbor cache entry for ('C2') with fe80::2001:db8:2:0 as the network-layer address and 2001:db8:2:: as the link-layer address and with ForwardTime set to FORWARD_TIME, thus completing the route optimization in the forward direction.

('C1') subsequently forwards encapsulated packets with outer source address 2001:db8:1000::1, with outer destination address 2001:db8:2::, with inner source address taken from the 2001:db8:1::, and with inner destination address taken from 2001:db8:2:: due to the fact that it has a securely-established neighbor cache entry with non-zero ForwardTime. ('C2') subsequently accepts any such encapsulated packets due to the fact that it has a securely-established neighbor cache entry with non-zero AcceptTime.

In order to keep neighbor cache entries alive, ('C1') periodically sends additional NS messages to ('C2') and receives any NA responses. If ('C1') moves to a different point of attachment after the initial route optimization, it sends a new secured NS message to ('C2') as above to update ('C2')s neighbor cache.

If ('C2') has packets to send to ('C1'), it performs a corresponding route optimization in the opposite direction following the same procedures described above. In the process, the already-established unidirectional neighbor cache entries within ('C1') and ('C2') are updated to include the now-bidirectional information. In particular, the AcceptTime and ForwardTime variables for both neighbor cache entries are updated to non-zero values, and the link-layer address for ('C1')s neighbor cache entry for ('C2') is reset to 2001:db8:2000::1.

Note that two AERO Clients can use full security protocol messaging instead of Return Routability, e.g., if strong authentication and/or confidentiality are desired. In that case, security protocol key exchanges such as specified for MOBIKE [RFC4555] would be used to establish security associations and neighbor cache entries between the AERO clients. Thereafter, NS/NA messaging can be used to maintain neighbor cache entries, test reachability, and to announce mobility events. If reachability testing fails, e.g., if both Clients move at roughly the same time, the Clients can tear down the security association and neighbor cache entries and again allow packets to flow through their home network.

3.23. Encapsulation Protocol Version Considerations

A source Client may connect only to an IPvX underlying network, while the target Client connects only to an IPvY underlying network. In that case, the target and source Clients have no means for reaching each other directly (since they connect to underlying networks of different IP protocol versions) and so must ignore any redirection messages and continue to send packets via their Servers.

3.24. Multicast Considerations

When the underlying network does not support multicast, AERO Clients map link-scoped multicast addresses to the link-layer address of a Server, which acts as a multicast forwarding agent. The AERO Client also serves as an IGMP/MLD Proxy for its EUNs and/or hosted applications per [RFC4605] while using the link-layer address of the Server as the link-layer address for all multicast packets.

When the underlying network supports multicast, AERO nodes use the multicast address mapping specification found in [RFC2529] for IPv4 underlying networks and use a TBD site-scoped multicast mapping for IPv6 underlying networks. In that case, border routers must ensure that the encapsulated site-scoped multicast packets do not leak outside of the site spanned by the AERO link.

3.25. Operation on AERO Links Without DHCPv6 Services

When Servers on the AERO link do not provide DHCPv6 services, operation can still be accommodated through administrative configuration of ACPs on AERO Clients. In that case, administrative configurations of AERO interface neighbor cache entries on both the Server and Client are also necessary. However, this may interfere with the ability for Clients to dynamically change to new Servers, and can expose the AERO link to misconfigurations unless the administrative configurations are carefully coordinated.

3.26. Operation on Server-less AERO Links

In some AERO link scenarios, there may be no Servers on the link and/or no need for Clients to use a Server as an intermediary trust anchor. In that case, each Client acts as a Server unto itself to establish neighbor cache entries by performing direct Client-to-Client IPv6 ND message exchanges, and some other form of trust basis must be applied so that each Client can verify that the prospective neighbor is authorized to use its claimed ACP.

When there is no Server on the link, Clients must arrange to receive ACPs and publish them via a secure alternate prefix delegation authority through some means outside the scope of this document.

3.27. Manually-Configured AERO Tunnels

In addition to the dynamic neighbor discovery procedures for AERO link neighbors described above, AERO encapsulation can be applied to manually-configured tunnels. In that case, the tunnel endpoints use an administratively-assigned link-local address and exchange NS/NA messages the same as for dynamically-established tunnels.

3.28. Intradomain Routing

After a tunnel neighbor relationship has been established, neighbors can use a traditional dynamic routing protocol over the tunnel to exchange routing information without having to inject the routes into the AERO routing system.

4. Implementation Status

User-level and kernel-level AERO implementations have been developed and are undergoing internal testing within Boeing.

An initial public release of the AERO source code was announced on the intarea mailing list on August 21, 2015, and a pointer to the code is available in the list archives.

5. IANA Considerations

The IANA has assigned a 4-octet Private Enterprise Number "45282" for AERO in the "enterprise-numbers" registry.

The IANA has assigned the UDP port number "8060" for an earlier experimental version of AERO [RFC6706]. This document obsoletes [RFC6706] and claims the UDP port number "8060" for all future use.

No further IANA actions are required.

6. Security Considerations

AERO link security considerations are the same as for standard IPv6 Neighbor Discovery [RFC4861] except that AERO improves on some aspects. In particular, AERO uses a trust basis between Clients and Servers, where the Clients only engage in the AERO mechanism when it is facilitated by a trust anchor. Unless there is some other means of authenticating the Client's identity (e.g., link-layer security), AERO nodes SHOULD also use DHCPv6 securing services (e.g., DHCPv6 authentication, Secure DHCPv6 [I-D.ietf-dhc-sedhcpv6], etc.) for Client authentication and network admission control. In particular, Clients SHOULD include authenticating information on each Solicit/Rebind/Release message they send, but omit authenticating information on Renew messages. Renew messages are exempt due to the fact that the Renew must already be checked for having a correct link-layer address and does not update any link-layer addresses. Therefore, asking the Server to also authenticate the Renew message would be unnecessary and could result in excessive processing overhead.

Redirect, Predirect and unsolicited NA messages SHOULD include a Timestamp option (see Section 5.3 of [RFC3971]) that other AERO nodes can use to verify the message time of origin. Predirect, NS and RS messages SHOULD include a Nonce option (see Section 5.3 of [RFC3971]) that recipients echo back in corresponding responses.

AERO links must be protected against link-layer address spoofing attacks in which an attacker on the link pretends to be a trusted neighbor. Links that provide link-layer securing mechanisms (e.g., IEEE 802.1X WLANs) and links that provide physical security (e.g., enterprise network wired LANs) provide a first line of defense that is often sufficient. In other instances, additional securing mechanisms such as Secure Neighbor Discovery (SeND) [RFC3971], IPsec [RFC4301] or TLS [RFC5246] may be necessary.

AERO Clients MUST ensure that their connectivity is not used by unauthorized nodes on their EUNs to gain access to a protected network, i.e., AERO Clients that act as routers MUST NOT provide routing services for unauthorized nodes. (This concern is no different than for ordinary hosts that receive an IP address delegation but then "share" the address with unauthorized nodes via some form of Internet connection sharing.)

On some AERO links, establishment and maintenance of a direct path between neighbors requires secured coordination such as through the Internet Key Exchange (IKEv2) protocol [RFC5996] to establish a security association.

An AERO Client's link-layer address could be rewritten by a link-layer switching element on the path from the Client to the Server and not detected by the DHCPv6 security mechanism. However, such a condition would only be a matter of concern on unmanaged/unsecured links where the link-layer switching elements themselves present a man-in-the-middle attack threat. For this reason, IP security MUST be used when AERO is employed over unmanaged/unsecured links.

7. Acknowledgements

Discussions both on IETF lists and in private exchanges helped shape some of the concepts in this work. Individuals who contributed insights include Mikael Abrahamsson, Mark Andrews, Fred Baker, Stewart Bryant, Brian Carpenter, Wojciech Dec, Ralph Droms, Adrian Farrel, Sri Gundavelli, Brian Haberman, Joel Halpern, Tom Herbert, Sascha Hlusiak, Lee Howard, Andre Kostur, Ted Lemon, Andy Malis, Satoru Matsushima, Tomek Mrugalski, Alexandru Petrescu, Behcet Saikaya, Joe Touch, Bernie Volz, Ryuji Wakikawa and Lloyd Wood. Members of the IESG also provided valuable input during their review process that greatly improved the document. Discussions on the v6ops list in the December 2015 through January 2016 timeframe further helped clarify AERO multi-addressing capabilities. Special thanks go to Stewart Bryant, Joel Halpern and Brian Haberman for their shepherding guidance during the publication of the AERO first edition.

This work has further been encouraged and supported by Boeing colleagues including M. Wayne Benson, Dave Bernhardt, Cam Brodie, Balaguruna Chidambaram, Irene Chin, Bruce Cornish, Claudiu Danilov, Wen Fang, Anthony Gregory, Jeff Holland, Ed King, Gen MacLean, Rob Muszkiewicz, Sean O'Sullivan, Kent Shuey, Brian Skeen, Mike Slane, Brendan Williams, Julie Wulff, Yueli Yang, and other members of the BR&T and BIT mobile networking teams. Wayne Benson is especially acknowledged for his outstanding work in converting the AERO proof-of-concept implementation into production-ready code.

Earlier works on NBMA tunneling approaches are found in [RFC2529][RFC5214][RFC5569].

Many of the constructs presented in this second edition of AERO are based on the author's earlier works, including:

• The Internet Routing Overlay Network (IRON) [RFC6179][I-D.templin-ironbis]
• Virtual Enterprise Traversal (VET) [RFC5558][I-D.templin-intarea-vet]
• The Subnetwork Encapsulation and Adaptation Layer (SEAL) [RFC5320][I-D.templin-intarea-seal]
• AERO, First Edition [RFC6706]

Note that these works cite numerous earlier efforts that are not also cited here due to space limitations. The authors of those earlier works are acknowledged for their insights.

8. References

8.1. Normative References

8.2. Informative References

Appendix A. AERO Alternate Encapsulations

When GUE encapsulation is not needed, AERO can use common encapsulations such as IP-in-IP [RFC2003][RFC2473][RFC4213], Generic Routing Encapsulation (GRE) [RFC2784][RFC2890] and others. The encapsulation is therefore only differentiated from non-AERO tunnels through the application of AERO control messaging and not through, e.g., a well-known UDP port number.

As for GUE encapsulation, alternate AERO encapsulation formats may require encapsulation layer fragmentation. For simple IP-in-IP encapsulation, an IPv6 fragment header is inserted directly between the inner and outer IP headers when needed, i.e., even if the outer header is IPv4. The IPv6 Fragment Header is identified to the outer IP layer by its IP protocol number, and the Next Header field in the IPv6 Fragment Header identifies the inner IP header version. For GRE encapsulation, a GRE fragment header is inserted within the GRE header [I-D.templin-intarea-grefrag].

Figure 9 shows the AERO IP-in-IP encapsulation format before any fragmentation is applied:

     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |     Outer IPv4 Header     |      |    Outer IPv6 Header      |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |IPv6 Frag Header (optional)|      |IPv6 Frag Header (optional)|
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |      Inner IP Header      |      |       Inner IP Header     |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                           |      |                           |
     ~                           ~      ~                           ~
     ~    Inner Packet Body      ~      ~     Inner Packet Body     ~
     ~                           ~      ~                           ~
     |                           |      |                           |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+

     Minimal Encapsulation in IPv4      Minimal Encapsulation in IPv6

     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |        Outer IP Header        |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |          GRE Header           |
     | (with checksum, key, etc..)   |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     | GRE Fragment Header (optional)|
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |        Inner IP Header        |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                               |
     ~                               ~
     ~      Inner Packet Body        ~
     ~                               ~
     |                               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+