Futurewei Inc.

2330 Central Expressway Santa Clara 95050 CA USA +1-408-330-5305 charliep@computer.org

Cisco

170 West Tasman Drive San Jose CA 95134 USA sratliff@cisco.com

Cassidian

Celtic Springs Newport Wales NP10 8FZ United Kingdom John.Dowdell@Cassidian.com

Routing Mobile Ad hoc Networks Working Group RFC Request for Comments I-D Internet-Draft XML reactive protocol The revised Ad Hoc On-demand Distance Vector (AODVv2) routing protocol is intended for use by mobile routers in wireless, multihop networks. AODVv2 determines unicast routes among AODVv2 routers within the network in an on-demand fashion, offering on-demand convergence in dynamic topologies.

The revised Ad Hoc On-demand Distance Vector (AODVv2) routing protocol [formerly named DYMO] enables on-demand, multihop unicast routing among AODVv2 routers in mobile ad hod networks [MANETs]. The basic operations of the AODVv2 protocol are route discovery and route maintenance. Route discovery is performed when an AODVv2 router must transmit a packet towards a destination for which it does not have a route. Route maintenance is performed to avoid prematurely expunging routes from the route table, and to avoid dropping packets when an active route breaks. During route discovery, the originating AODVv2 router (RREQ_Gen) multicasts a Route Request message (RREQ) to find a route toward some target destination. Using a hop-by-hop retransmission algorithm, each AODVv2 router receiving the RREQ message records a route toward the originator. When the target's AODVv2 router (RREP_Gen) receives the RREQ, it records a route toward RREQ_Gen and generates a Route Reply (RREP) unicast toward RREQ_Gen. Each AODVv2 router that receives the RREP stores a route toward the target, and again unicasts the RREP toward the originator. When RREQ_Gen receives the RREP, routes have then been established between RREQ_Gen (the originating AODVv2 router) and RREP_Gen (the target's AODVv2 router) in both directions. Route maintenance consists of two operations. In order to maintain active routes, AODVv2 routers extend route lifetimes upon successfully forwarding a packet. When a data packet is received to be forwarded downstream but there is no valid route for the destination, then the AODVv2 router of the source of the packet is notified via a Route Error (RERR) message. Each upstream router that receives the RERR marks the route as broken. Before such an upstream AODVv2 router could forward a packet to the same destination, it would have to perform route discovery again for that destination. AODVv2 uses sequence numbers to assure loop freedom , similarly to AODV. Sequence numbers enable AODVv2 routers to determine the temporal order of AODVv2 route discovery messages, thereby avoiding use of stale routing information. Unlike AODV, AODVv2 uses RFC 5444 message and TLV formats.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in . This document also uses some terminology from . This document defines the following terminology: A bi-directional relationship between neighboring AODVv2 routers for the purpose of exchanging routing information. Not every pair of neighboring routers will necessarily form an adjacency. Neighboring routers may form an adjacency based on various information or other protocols; for example, exchange of AODVv2 routing messages, other protocols (e.g. NDP or NHDP ), or manual configuration. Loss of a routing adjacency may also be indicated by similar information; monitoring of adjacencies where packets are being forwarded is required (see ). An IP addressable device in the ad-hoc network that performs the AODVv2 protocol operations specified in this document. Same as Sequence Number. The current time as maintained by the AODVv2 router. Ignore for further processing (see ), and discard unless it is required to keep the message in the packet for purposes of authentication. HandlingRtr denotes the AODVv2 router receiving and handling an AODVv2 message. A link over which an AODVv2 router has received a message from an adjacent router. A Mobile Ad Hoc Network as defined in . An IP addressable device in the ad-hoc network. A node may be an AODVv2 router, or it may be a device in the network that does not perform any AODVv2 protocol operations. All nodes in this document are either AODVv2 Routers or else Router Clients. The Originating Node is the node that launched the application requiring communication with the Target Node. If OrigNode is not itself an AODVv2 router, its AODVv2 router (RREQ_Gen) has the responsibility to generate a AODVv2 RREQ message on behalf of OrigNode when necessary to discover a route. A protocol operation is said to be "reactive" if it is performed only in reaction to specific events. As used in this document, "reactive" is essentially synonymous with "on-demand". A routable unicast IP address is a unicast IP address that when put into the IP.DestinationAddress field is scoped sufficiently to be forwarded by a router. Globally-scoped unicast IP addresses and Unique Local Addresses (ULAs) are examples of routable unicast IP addresses. A RERR message is used to indicate that an AODVv2 router does not have a route toward one or more particular destinations. A RREP message is used to establish a route between the RREQ TargetNode and OrigNode, at all the AODVv2 routers between them. An AODVv2 router uses a RREQ message to discover a valid route to a particular destination address, called the TargetNode. An AODVv2 router processing a RREQ receives routing information for the RREQ OrigNode. An AODVv2 router may be configured with a list of other IP addresses and networks which correspond to other non-router nodes which require the services of the AODVv2 router for route discovery and maintenance. An AODVv2 router is always its own client, so that the list of client IP addresses is never empty. The RREP Generating Router is the AODVv2 router that serves TargNode. RREP_Gen generates the RREP message to advertise a route for TargNode. The RREQ Generating Router is the AODVv2 router that serves OrigNode. RREQ_Gen generates the RREQ message to discover a route for TargNode. AODVv2 mandates that each AODVv2 router maintain an unsigned integer known as the router's "Sequence Number". The Sequence Number guarantees the temporal order of routing information to maintain loop-free routes, and fulfills the same role as the "Destination Sequence Number" of DSDV, and as the AODV Sequence Number in RFC 3561. The value zero (0) is reserved to indicate that the Sequence Number for an address is unknown. The Target Node denotes the node for which a route is needed. A generic way to represent information as specified in . An UnreachableNode is a node for which a forwarding route is unknown. A route that can be used for forwarding; in other words a route that is not Broken or Expired.

This document uses the conventions found in to describe information in the fields from . Notation Information Location and/or Meaning Route[Address] A route table entry towards Address Route[Address].{field} A field in a route table entry -- -- <msg-hop-count> RFC 5444 Message Header <msg-hop-count> <msg-hop-limit> RFC 5444 Message Header <msg-hop-limit> AddrBlk an RFC 5444 Address TLV Block AddrBlk[1] The first address slot in AddrBlk AddrBlk[N] The Nth address slot in AddrBlk OrigNdx The index of OrigNode within the AddrBlk TargNdx The index of TargNode within the AddrBlk AddrBlk[OrigNode] AddrBlk[OrigNdx] AddrBlk[TargNode] AddrBlk[TargNdx] AddrTLV an RFC 5444 Address Block TLV AddrTLV[1] the first item in AddrTLV AddrTLV[N] the Nth item in AddrTLV AddrTLV[OrigNode] AddrTLV[OrigNdx] AddrTLV[TargNode] AddrTLV[TargNdx] Metric_TLV Metric AddrTLV for AddrBlk SeqNum_TLV Sequence Number AddrTLV for AddrBlk OrigSeqNum_TLV Originating Node Sequence Number AddrTLV TargSeqNum_TLV Target Node Sequence Number AddrTLV -- -- OrigNode Originating Node RREQ_Gen AODVv2 router originating an RREQ RREP_Gen AODVv2 router responding to an RREQ RteMsg Either RREQ or RREP RteMsg.{field} Field in RREQ or RREP HandlingRtr Handling Router TargNode Target Node UnreachableNode Unreachable Node

The AODVv2 routing protocol is designed for stub (i.e., non-transit) or disconnected (i.e., from the Internet) mobile ad hoc networks (MANETs). AODVv2 handles a wide variety of mobility patterns by determining routes on-demand. AODVv2 also handles a wide variety of traffic patterns. In networks with a large number of routers, AODVv2 is best suited for relatively sparse traffic scenarios where any particular router forwards packets to only a small percentage of the AODVv2 routers in the network, due to the on-demand nature of route discovery and route maintenance. AODVv2 supports routers with multiple interfaces, as long as each interface has its own (unicast routeable) IP address; the set of all network interfaces supporting AODVv2 is administratively configured in a list (namely, AODVv2_INTERFACES). Although AODVv2 is closely related to AODV , and has some of the features of DSR , AODVv2 is not interoperable with either of those other two protocols. AODVv2 is applicable to memory constrained devices, since little routing state is maintained in each AODVv2 router. Only routing information related to routes between active sources and destinations is maintained, in contrast to proactive routing protocols that require routing information to all routers within the MANET be maintained. In addition to routing for its own local applications, each AODVv2 router can also route on behalf of other non-routing nodes (i.e., "hosts", or, in this document, "clients"), reachable via those interfaces. Each AODVv2 router, if serving router clients other than itself, is configured with information about the IP addresses of its clients. No AODVv2 router is required to have information about the relationship between any other AODVv2 router and its router clients (see ). The coordination among multiple AODVv2 routers to distribute routing information correctly for a shared address (i.e. an address that is advertised and can be reached via multiple AODVv2 routers) is not described in this document. The AODVv2 router operation of shifting responsibility for a routing client from one AODVv2 router to another is mentioned in . Address assignment procedures are entirely out of scope for AODVv2. Any such node which is not itself an AODVv2 router SHOULD NOT be served by more than one AODVv2 router at any one time. Multi-homing is difficult unless the sequence number is expanded to include the AODVv2 router's IP address as well as SeqNum. Otherwise, comparing sequence numbers would not work to evaluate freshness. Even when the IP address is included, there isn't a good way to compare sequence numbers from different IP addresses, but at least a handling node can determine whether the two given sequence numbers are comparable. If the route table can store multiple routes for the same destination, then multi-homing can work with sequence numbers augmented by IP addresses. AODVv2 routers perform route discovery to find a route toward a particular destination. Therefore, AODVv2 routers MUST must be configured to respond to RREQs for a certain set of addresses. When AODVv2 is the only protocol interacting with the forwarding table, AODVv2 MAY be configured to perform route discovery for all unknown unicast destinations. AODVv2 only supports bidirectional links. In the case of possible unidirectional links, either blacklists (see ) or other means (e.g. adjacency establishment with only neighboring routers that have bidirectional communication as indicated by NHDP ) of assuring and monitoring bi-directionality are recommended. Otherwise, persistent packet loss or persistent protocol failures could occur. The cost of bidirectional link L (denoted Cost(L)) may depend upon the direction across the link for which the cost is measured. Metric information from incoming AODVv2 messages SHOULD NOT be used for route table updates unless it has been received over a link that is bidirectional. The routing algorithm in AODVv2 may be operated at layers other than the network layer, using layer-appropriate addresses. The routing algorithm makes of some persistent state; if there is no persistent storage available for this state, recovery can impose a performance penalty (e.g., in case of AODVv2 router reboots).

The route table entry is a conceptual data structure. Implementations may use any internal representation so long as it provides access to the information specified below. Conceptually, a route table entry has the following fields: The (host or network) destination address of the node(s) associated with the routing table entry The length of the netmask/prefix. If the value of the Route.PrefixLength is less than the length of addresses in the address family used by the AODVv2 routers, the associated address is a routing prefix, rather than a host address. A PrefixLength is stored for every route in the route table. The Sequence Number associated with destination of the route, as obtained from the last packet that updated the route table entry associated with a route table entry. An IP address of the adjacent AODVv2 router on the path toward the Route.Address The interface used to send packets toward the Route.Address The time that this route was last used The time at which this route must expire A flag indicating whether this Route is broken. This flag is set to true if the next-hop becomes unreachable or in response to processing to a RERR (see ) The type of the metric for the route towards Route.Address The cost of the route towards Route.Address A route table entry (i.e., a route) may be in one of the following states: An Active route is in current use for forwarding packets An Idle route can be used for forwarding packets, even though it is not in current use After a route has been idle for too long, it expires, and may no longer be used for forwarding packets A route marked as Broken cannot be used for forwarding packets but still has valid destination sequence number information. The expiration of a Timed route is controlled by the Route.ExpirationTime time of the route table entry (instead of MAX_IDLETIME). Until that time, a Timed route can be used for forwarding packets. Afterwards, the route must be Expired (or expunged). The route's state determines the operations that can be performed on the route table entry. During use, an Active route is maintained continuously by AODVv2 and is considered to remain active as long as it is used at least once during every ACTIVE_INTERVAL. When a route is no longer Active, it becomes an Idle route. After an idle route remains Idle for MAX_IDLETIME, it becomes an Expired route. An Expired route is not used for forwarding, but the sequence number information can be maintained until the destination sequence number has had no updates for MAX_SEQNUM_LIFETIME; after that time, old sequence number information is considered no longer valuable and the Expired route MUST BE expunged. MAX_SEQNUM_LIFETIME is the time after a reboot during which an AODVv2 router MUST NOT transmit any routing messages. Thus, if all other AODVv2 routers expunge routes to the rebooted router after that time interval, the rebooted AODVv2 router's sequence number will not be considered stale by any other AODVv2 router in the MANET. When the link to a route's next hop is broken, the route is marked as being Broken, and the route may no longer be used.

To avoid repeated failure of Route Discovery, an AODVv2 router (HandlingRtr) handling a RREP message SHOULD attempt to verify connectivity to the next upstream router towards AODVv2 router originating an RREQ message. This MAY be done by including the Acknowledgement Request (AckReq) message TLV (see ) in the RREP. Any unicast packet will satisfy the Acknowledgement Request, for example an ICMP REPLY message. If the verification is not received within UNICAST_MESSAGE_SENT_TIMEOUT, HandlingRtr SHOULD put the upstream neighbor in the blacklist. RREQs received from a blacklisted node, or any node over a link that is known to be incoming-only, SHOULD NOT be retransmitted by HandlingRtr. However, the upstream neighbor SHOULD NOT be permanently blacklisted; after a certain time (MAX_BLACKLIST_TIME), it SHOULD once again be considered as a viable upstream neighbor for route discovery operations. For this purpose, a list of blacklisted nodes along with their time of removal SHOULD be maintained: The IP address of the node that did not verify bidirectional connectivity. The time at which Blacklist.Node will be removed from the blacklist.

An AODVv2 router may offer routing services to other nodes that are not AODVv2 routers. AODVv2 defines the Sequence Number to be the same for the AODVv2 router and each of its clients. For this purpose, CLIENT_ADDRESSES must be configured on each AODVv2 router with the following information: The IP address of the node that requires routing service from the AODVv2 router. The length of the routing prefix associated with the client IP address. If the Client Prefix Length is not the full length of the Client IP address, then the prefix defines a Client Network. If an AODVv2 router is configured to serve a Client Network, then the AODVv2 router MUST serve every node that has an address within the range defined by the routing prefix of the Client Network. The list of Routing Clients for an AODVv2 router is never empty, since an AODVv2 router is always its own client as well.

In its default mode of operation, AODVv2 transmits UDP packets using the parameters for port number and IP protocol specified in to carry protocol packets. By default, AODVv2 packets are sent with the IP destination address set to the link-local multicast address LL-MANET-Routers unless otherwise specified. Therefore, all AODVv2 routers MUST subscribe to LL-MANET-Routers to receiving AODVv2 messages. In order to reduce multicast overhead, retransmitting multicast packets in MANETs SHOULD be done according to methods specified in . AODVv2 does not specify which method should be used to restrict the set of AODVv2 routers that have the responsibility to retransmit multicast packets. Note that multicast packets MAY be sent via unicast. For example, this may occur for certain link-types (non-broadcast media), for manually configured router adjacencies, or in order to improve robustness. The IPv4 TTL (IPv6 Hop Limit) field for all packets containing AODVv2 messages is set to 255. If a packet is received with a value other than 255, any AODVv2 message contained in the packet MUST be disregarded by AODVv2. This mechanism, known as "The Generalized TTL Security Mechanism" (GTSM) helps to assure that packets have not traversed any intermediate routers. IP packets containing AODVv2 protocol messages SHOULD be given priority queuing and channel access. AODVv2 messages are transmitted in packets that conform to the packet and message format specified in . Here is a brief summary of the format. A packet formatted according to RFC 5444 contains zero or more messages. A message contains a message header, message TLV block, and zero or more address blocks. Each address block may also have an associated TLV block; this TLV block may encode multiple TLVs. According to RFC 5444, each such TLV may itself include an array of values. If a packet contains only a single AODVv2 message and no packet TLVs, it need only include a minimal Packet-Header . The length of an address (32 bits for IPv4 and 128 bits for IPv6) inside an AODVv2 message is indicated by the msg-addr-length (MAL) in the msg-header, as specified in . When multiple messages are aggregated into a single packet according to RFC 5444 formatting, and the aggregation of messages is also authenticated (e.g., with IPsec), and the IP destination is multiple hops away, it becomes infeasible to delete individual messages. In such cases, instead of deleting individual messages, they are maintained in the aggregation of messages, but simply ignored for further processing. In such cases where individual messages cannot be deleted, in this document "disregarded" means "ignored". Otherwise, any such "disregarded" AODVv2 messages SHOULD be deleted from the aggregated messages in the RFC 5444 packet.

Sequence Numbers allow AODVv2 routers to evaluate the freshness of routing information. Each AODVv2 router in the network MUST maintain its own sequence number. Each RREQ and RREQ generated by an AODVv2 router includes that sequence number. Each AODVv2 router MUST make sure that its sequence number is unique and monotonically increasing. This can be achieved by incrementing it with every RREQ or RREP it generates. Every router receiving a RREQ or RREP can thus use the Sequence Number of a RREQ or RREP as information concerning the freshness of the packet's route update: If the new packet's Sequence Number is lower than the one already stored in the routing table, its information is considered stale. As a consequence, loop freedom is assured. An AODVv2 router increments its SeqNum as follows. Most of the time, SeqNum is incremented by simply adding one (1). But when the SeqNum has the value of the largest possible number representable as a 16-bit unsigned integer (i.e., 65,535), it MUST be incremented by setting to one (1). In other words, the sequence number after 65,535 is 1. An AODVv2 router SHOULD maintain its SeqNum in persistent storage. If an AODVv2 router's SeqNum is lost, it MUST take the following actions to avoid the danger of routing loops. First, the AODVv2 router MUST invalidate all route table entries, by setting Route.Broken for each entry. Furthermore the AODVv2 router MUST wait for at least MAX_SEQNUM_LIFETIME before transmitting or retransmitting any AODVv2 RREQ or RREP messages. If an AODVv2 protocol message is received during this waiting period, the AODVv2 router SHOULD perform normal route table entry updates, but not forward the message to other nodes. If a data packet is received for forwarding to another destination during this waiting period, the AODVv2 router MUST transmit a RERR message indicating that no route is available. At the end of the waiting period the AODVv2 router sets its SeqNum to one (1) and begins performing AODVv2 protocol operations again.

AODVv2 route selection in MANETs depends upon associating metric information with each route table entry. When presented with candidate route update information, deciding whether to use the update involves evaluating the metric. Some applications may require metric information other than Hop Count, which has traditionally been the default metric associated with routes in MANET. Unfortunately, it is well known that reliance on Hop Count can cause selection of the worst possible route in many situations. It is beyond the scope of this document to describe how applications specify route selection at the time they launch processing. One possibility would be to provide a route metric preference as part of the library routines for opening sockets. In view of the above considerations, it is important to enable route selection based on metric information other than Hop Count -- in other words, based on "alternate metrics". Each such alternate metric measures a "cost" of using the associated route, and there are many different kinds of cost (latency, delay, monetary, energy, etc.). The most significant change when enabling use of alternate metrics is to require the possibility of multiple routes to the same destination, where the "cost" of each of the multiple routes is measured by a different metric. Moreover, the method by which route updates are tested for usefulness has to be slightly generalized to depend upon a more abstract method of evaluation which, in this document, is named "Cost(R)", where 'R' is the route for which the Cost is to be evaluated. From the above, the route table information for 'R' must always include the type of metric by which Cost(R) is evaluated, so the metric type does not have to be shown as a distinct parameter for Cost(R). Since determining loop freedom is known to depend on comparing the Cost(R) of route update information to the Cost(R) of an existing stored route using the same metric, AODVv2 must also be able to invoke an abstract routine which in this document is called "LoopFree(R1, R2)". LoopFree(R1, R2) returns TRUE when, (under the assumption of nondecreasing SeqNum during Route Discovery) given that R2 is loop-free and Cost(R2) is the cost of route R2, Cost(R1) is known to guarantee loop freedom of the route R1. In this document, LoopFree(R1,R2) will only be invoked for routes R1 and R2 to the same destination which use the same metric. Generally, HopCount may still be considered the default metric for use in MANETs, notwithstanding the above objections. Each metric has to have a Metric Type, and the Metric Type is allocated by IANA as specified in . Each Route has to include the Metric Type as part of the route table entry for that route. Hop Count has Metric Type assignment 3. The Cost of a route using Metric Type 3 is simply the hop count between the router and the destination. For routes R1 and R2 using Metric Type 3, LoopFree (R1, R2) is TRUE when Cost(R2) ≤ (Cost(R1) + 1). The specification of Cost(R) and LoopFree(R1,R2) for metric types other than 3 is beyond the scope of this document. Whenever an AODV router receives metric information in an incoming message, the value of the metric is as measured by the transmitting router, and does not reflect the cost of traversing the incoming link. In order to simplify the description of storing accrued route costs in the route table, the Cost() function is also defined to return the value of traversing a link 'L'. In other words, the domain of the Cost() function is enlarged to include links as well as routes. For Metric Type 3, (i.e., the HopCount metric) Cost(L) = 1 for all links. The specification of Cost(L) for metric types other than 3 is beyond the scope of this document. Whether the argument of the Cost() function is a link or a route will, in this document, always be clear. As a natural result of the way routes are looked up according to conformant metric type, all intermediate routers handling a RteMsg will assign the same metric type to all metric information in the RteMsg. For some metrics, a maximum value is defined, namely MAX_METRIC[i] where 'i' is the Metric Type. AODVv2 does not store routes that cost more than MAX_METRIC[i]. MAX_METRIC[3] is defined to be MAX_HOPCOUNT, where as before 3 is the Metric Type of the HopCount metric. MAX_HOPCOUNT MUST be larger than the AODVv2 network diameter. Otherwise, AODVv2 protocol messages may not reach their intended destinations.

Two incoming RREQ messages are considered to be "comparable" if they were generated by the same AODVv2 router in order to discover a route for the same destination with the same metric type. According to that notion of comparability, when RREQ messages are flooded in a MANET, an AODVv2 router may well receive comparable RREQ messages from more than one of its neighbors. A router, after receiving an RREQ message, MUST check against previous RREQs to assure that its response message would contain information that is not redundant. Otherwise, multicast RREQs are likely to be retransmitted again and again with almost no additional benefit, but generating a great deal of unnecessary signaling traffic and interference. To avoid transmission of redundant RREQ messages, while still enabling the proper handling of earlier RREQ messages that may have somehow been delayed in the network, it is needed for each AODVv2 router to keep a list of the certain information about RREQ messages which it has recently received. This list is called the AODVv2 Received RREQ Table -- or, more briefly, the RREQ Table. Two AODVv2 RREQ messages are comparable if: they have the same metric type they have the same OrigNode and TargNode addresses Each entry in the RREQ Table has the following fields: Metric Type OrigNode address TargNode address Sequence Number Metric Timestamp The RREQ Table is maintained so that no two entries in the RREQ Table are comparable -- that is, all RREQs represented in the RREQ Table either have different OrigNode addresses, different TargNode addresses, or different metric types. If two RREQs have the same metric type and OrigNode and Targnode addresses, the information from the one with the older Sequence Number is not needed in the table; in case they have the same Sequence Number, the one with the greater Metric value is not needed; in case they have the same Metric as well, it does not matter which table entry is maintained. Whenever a RREQ Table entry is updated, its Timestamp field should also be updated to reflect the Current_Time. When optional multicast RREP (see ) is used to enable selection from among multiple possible return routes, an AODVv2 router can eliminate redundant RREP messages using the analogous mechanism along with a RREP Table. Nevertheless, the description in this section only refers to RREQ multicast messages. Protocol handling of RERR messages eliminates the need for tracking RERR messages, since the rules for RERR regeneration prevent the phenomenon of redundant retansmission that affects RREQ and RREP multicast.

In this section, operations are specified for updating the route table due to timeouts and route updates within AODVv2 messages. Route update information in AODVv2 messages includes IP addresses, along with the SeqNum and prefix length associated with each IP address, and including the Metric measured from the node transmitting the AODVv2 message to the IP address in the route update. IP addresses and prefix length are encoded within an RFC 5444 AddrBlk, and the SeqNum and Metric associated with each address in the AddrBlk are encoded in RFC 5444 AddrTLVs. In this section, RteMsg is either RREQ or RREP, RteMsg.Address[i] denotes the [i]th address in an RFC 5444 AddrBlk of the RteMsg. RteMsg.PrefixLength[i] denotes the associated prefix length for RteMsg.Address[i], and RteMsg.{field} denotes the corresponding value in the named AddrTLV block associated with RteMsg.Address[i]. All SeqNum comparisons use signed 16-bit arithmetic.

If the incoming RteMsg does not have a Metric Type Message TLV, then the metric information contained by RteMsg is considered to be of type DEFAULT_METRIC_TYPE -- which is 3 (for HopCount) unless changed by administrative action. Whenever an AODVv2 router (HandlingRtr) handles an incoming RteMsg (i.e., RREQ or RREP), for every relevant address (RteMsg.Address) in the RteMsg, HandlingRtr searches its route table to see if there is a route table entry with the same Metric Type of the RteMsg, matching RteMsg.Address. If not, HandlingRtr creates a route table entry for RteMsg.Address as described in . Otherwise, HandlingRtr compares the incoming routing information in RteMsg against the already stored routing information in the route table entry (Route) for RteMsg.Address, as described below. Suppose Route[RteMsg.Address] uses the same metric type as the incoming routing information, and the route entry contains Route.SeqNum, Route.Metric, and Route.Broken. Suppose the incoming routing information for Route.Address is RteMsg.SeqNum and RteMsg.Metric. Define RteMsg.Cost to be (RteMsg.Metric + Cost(L)), where L is the incoming link. The incoming routing information is classified as follows: If RteMsg.SeqNum < Route.SeqNum the incoming information is stale. Using stale routing information is not allowed, since that might result in routing loops. HandlingRtr MUST NOT update the route table entry using the routing information for RteMsg.Address. If RteMsg is not Stale (as in (1) above), RteMsg.Cost is next considered to insure loop freedom. If (TRUE != LoopFree (RteMsg, Route)) (see ), then the incoming RteMsg information is not guaranteed to prevent routing loops, and it MUST NOT be used to update any route table entry. = Route.Metric) && (Route.Broken==FALSE)]]> When RteMsg.SeqNum is the same as in a valid route table entry, and LoopFree (RteMsg, Route) assures loop freedom, incoming information still does not offer any improvement over the existing route table information if RteMsg.Cost ≥ Route.Metric. Using such incoming routing information to update a route table entry is not recommended. Incoming routing information that does not match any of the above criteria is better than existing routing table information and SHOULD be used to improve the route table. The following pseudo-code illustrates whether incoming routing information should be used to update an existing route table entry as described in .

Route.SeqNum) OR {(RteMsg.SeqNum == Route.SeqNum) AND [(RteMsg.Cost < Route.Metric) OR ((Route.Broken == TRUE) && LoopFree (RteMsg, Route))]} ]]>

The above logic corresponds to placing the following conditions on the incoming route update (compared to the existing route table entry) before it can be used: it is more recent, or it is not stale and is less costly, or it can safely repair a broken route.

To apply the route update, a route table entry for RteMsg.Address is either found to already exist in the route table, or else a new route table entry for RteMsg.Address is created and inserted into the route table. The route table entry is populated with the following information: If RteMsg.PrefixLength exists, then Route.PrefixLength := RteMsg.PrefixLength. Otherwise, Route.PrefixLength := maximum length for address family (either 32 or 128). Route.SeqNum := RteMsg.SeqNum Route.NextHopAddress := IP.SourceAddress (i.e., an address of the node from which the RteMsg was received) Route.NextHopInterface is set to the interface on which RteMsg was received Route.Broken flag := FALSE If RteMsg.MetricType is included, then Route.MetricType := RteMsg.MetricType. Otherwise, Route.MetricType := DEFAULT_METRIC_TYPE. Route.Metric := (RteMsg.Metric + Cost(L)), where L is the incoming link. Route.LastUsed := Current_Time If RteMsg.VALIDITY_TIME is included, then Route.ExpirationTime := Current_Time + RteMsg.VALIDITY_TIME, otherwise, Route.ExpirationTime := Current_Time + (ACTIVE_INTERVAL + MAX_IDLETIME). With these assignments to the route table entry, a route has been made available, and the route can be used to send any buffered data packets and subsequently to forward any incoming data packets for Route.Address. An updated route entry also fulfills any outstanding route discovery (RREQ) attempts for Route.Address.

During normal operation, AODVv2 does not require any explicit timeouts to manage the lifetime of a route. However, the route table entry MUST be examined before using it to forward a packet, as discussed in . Any required expiry or deletion can occur at that time. Nevertheless, it is permissible to implement timers and timeouts to achieve the same effect. At any time, the route table can be examined and route table entries can be expunged according to their current state at the time of examination, as follows. An Active route MUST NOT be expunged. An Idle route SHOULD NOT be expunged. An Expired route MAY be expunged (least recently used first). A route MUST be expunged if (Current_Time - Route.LastUsed) >= MAX_SEQNUM_LIFETIME. A route MUST be expunged if Current_Time >= Route.ExpirationTime If precursor lists are maintained for the route (as described in ) then the precursor lists must also be expunged at the same time that the route itself is expunged.

AODVv2 message types RREQ and RREP are together known as Routing Messages (RteMsgs) and are used to discover a route between an Originating and Target Node, denoted here by OrigNode and TargNode. The constructed route is bidirectional, enabling packets to flow between OrigNode and TargNode. RREQ and RREP have similar information and function, but have some differences in their rules for handling. The main difference between the two messages is that RREQ messages are typically multicast to solicit a RREP, whereas RREP is typically unicast as a response to RREQ. When an AODVv2 router needs to forward a data packet from a node (OrigNode) in its set of router clients, and it does not have a forwarding route toward the packet's IP destination address (TargNode), the AODVv2 router (RREQ_Gen) generates a RREQ (as described in ) to discover a route toward TargNode. Subsequently RREQ_Gen awaits reception of an RREP message (see ) or other route table update (see ) to establish a route toward TargNode. Optionally, RREQ_Gen MAY specify that only the router serving TargNode is allowed to generate an RREP message, by including the DestOnly message TLV (see ). The RREQ message contains routing information to enable RREQ recipients to route packets back to OrigNode, and the RREP message contains routing information enabling RREP recipients to route packets to TargNode.

After issuing a RREQ, as described above RREQ_Gen awaits a RREP providing a bidirectional route toward Target Node. If the RREP is not received within RREQ_WAIT_TIME, RREQ_Gen may retry the Route Discovery by generating another RREQ. Route Discovery SHOULD be considered to have failed after DISCOVERY_ATTEMPTS_MAX and the corresponding wait time for a RREP response to the final RREQ. After the attempted Route Discovery has failed, RREQ_Gen MUST wait at least RREQ_HOLDDOWN_TIME before attempting another Route Discovery to the same destination. To reduce congestion in a network, repeated attempts at route discovery for a particular Target Node SHOULD utilize a binary exponential backoff. Data packets awaiting a route SHOULD be buffered by RREQ_Gen. This buffer SHOULD have a fixed limited size (BUFFER_SIZE_PACKETS or BUFFER_SIZE_BYTES). Determining which packets to discard first is a matter of policy at each AODVv2 router; in the absence of policy constraints, by default older data packets SHOULD be discarded first. Buffering of data packets can have both positive and negative effects (albeit usually positive). Nodes without sufficient memory available for buffering SHOULD be configured to disable buffering by configuring BUFFER_SIZE_PACKETS == 0 and BUFFER_SIZE_BYTES == 0. Doing so will affect the latency required for launching TCP applications to new destinations. If a route discovery attempt has failed (i.e., DISCOVERY_ATTEMPTS_MAX attempts have been made without receiving a RREP) to find a route toward the Target Node, any data packets buffered for the corresponding Target Node MUST BE dropped and a Destination Unreachable ICMP message (Type 3) SHOULD be delivered to the source of the data packet. The code for the ICMP message is 1 (Host unreachable error). If RREQ_Gen is not the source (OrigNode), then the ICMP is sent over the interface from which OrigNode sent the packet to the AODVv2 router.

RteMsgs have the following general format:

The RteMsg MUST contain the following: <msg-hop-limit> Metric Type Message TLV, if Metric Type != 3 The RteMsg may contain the following: <msg-hop-count> DestOnly TLV (RREQ only: no Intermediate RREP) Metric Type TLV (Metric Type for Metric AddrTLV) AckReq TLV (Acknowledgement Requested) This Address Block contains the IP addresses for RREQ Originating and Target Node (OrigNode and TargNode). For both RREP and RREQ, OrigNode and TargNode are as identified in the context of the RREQ message originator. At least one of OrigSeqNum or TargSeqNum Address Block TLV is REQUIRED and carries the destination sequence numbers associated with either OrigNode or TargNode. Both may appear when SeqNum information is available for both OrigNode and TargNode. The Metric AddrTLV is REQUIRED and carries the route metric information associated with either OrigNode or TargNode. RteMsgs carry information about OrigNode and TargNode. Since their addresses may appear in arbitrary order within the RFC 5444 AddrBlk, the OrigSeqNum and/or TargSeqNum TLVs must be used to distinguish the nature of the node addresses present in the AddrBlk. In each RteMsg, either the OrigSeqNum TLV or TargSeqNum TLV MUST appear. Both TLVs MAY appear in the same RteMsg, but each one MUST NOT appear more than once, because there is only one OrigNode and only one TargNode address in the AddrBlk. If the OrigSeqNum TLV appears, then the address range for the OrigSeqNum TLV MUST be limited to a single position in the AddrBlk. That position is used as the OrigNdx, identifying the OrigNode address. The other address in the AddrBlk is, by elimination, the TargNode address, and TargNdx is set appropriately. Otherwise, if the TargSeqNum TLV appears, then the address range for the TargSeqNum TLV MUST be limited to a single position in the AddrBlk. That position is used as the TargNdx, identifying the TargNode address. The other address in the AddrBlk is, by elimination, the OrigNode address, and OrigNdx is set appropriately.

The AODVv2 router generating the RREQ (RREQ_Gen) on behalf of its client OrigNode follows the steps in this section. OrigNode MUST be a unicast address. The order of protocol elements is illustrated schematically in . RREQ_Gen MUST increment its SeqNum by one (1) according to the rules specified in . This assures that each node receiving the RREQ will update its route table using the information in the RREQ. If RREQ_Gen requires that only the router providing connectivity to TargNode is allowed to generate a RREP, then RREQ_Gen includes the "Destination RREP Only" (DestOnly) TLV as part of the RFC 5444 message header. This also assures that RREP_Gen increments its sequence number. Otherwise, (if the optional behavior is enabled) other AODVv2 routers MAY respond to the RREQ if they have a valid route to TargNode (see ). <msg-hop-limit> SHOULD be set to MAX_HOPCOUNT. <msg-hop-count>, if included, MUST be set to 0. This RFC 5444 constraint causes certain RREQ payloads to incur additional enlargement (otherwise, <msg-hop-count> could often be used as the metric). RREQ.AddrBlk := {OrigNode.Addr, TargNode.Addr} Let OrigNdx and TargNdx denote the indexes of OrigNode and TargNode respectively in the RREQ.AddrBlk list. If Route[OrigNode].PrefixLength/8 is equal to the number of bytes in the addresses of the RREQ (4 for IPv4, 16 for IPv6), then no <prefix-length> is included with the RREQ.AddrBlk. Otherwise, RREQ.PrefixLength[OrigNdx] := Route[OrigNode].PrefixLength according to the rules of RFC 5444 AddrBlk encoding. RREQ.OrigSeqNum_TLV[OrigNdx] := RREQ_Gen's SeqNum RREQ.TargSeqNum_TLV[TargNdx] := TargNode's SeqNum (only if known) RREQ_Gen SHOULD include TargNode's SeqNum, if a previous value of the TargNode's SeqNum is known (e.g., from an invalid routing table entry using longest-prefix matching). If TargNode's SeqNum is not included, AODVv2 routers handling the RREQ assume that RREQ_Gen does not have that information. If ENABLE_IRREP is enabled, then any route to TargNode will satisfy the RREQ . RREQ.Metric_TLV[OrigNdx] := Route[OrigNode].Metric An example RREQ message format is illustrated in .

This section specifies the generation of an RREP by an AODVv2 router (RREP_Gen) that provides connectivity for the Target Node (TargNode) of a RREQ, thus enabling the establishment of a route between OrigNode and TargNode. If TargNode is not a unicast IP address the RREP MUST NOT be generated, and processing for the RREQ is complete. Before transmitting a RREP, the routing information of the RREQ is processed as specified in ; after such processing, RREP_Gen has an updated route to OrigNode as well as TargNode. The basic format of an RREP conforms to the structure for RteMsgs as shown in . RREP_Gen generates the RREP as follows: RREP_Gen checks the RREQ against recently received RREQ information as specified in . If a previously received RREQ has made the information in the incoming RREQ to be redundant, no RREP is generated and processing is complete. RREP_Gen MUST increment its SeqNum by one (1) according to the rules specified in . RREP.AddrBlk := {OrigNode.Addr, TargNode.Addr} Let OrigNdx and TargNdx denote the indexes of OrigNode and TargNode respectively in the RREQ.AddrBlk list. RREP.TargSeqNum_TLV[TargNdx] := RREP_Gen's SeqNum If Route[TargNode].PrefixLength/8 is equal to the number of bytes in the addresses of the RREQ (4 for IPv4, 16 for IPv6), then no <prefix-length> is included with the RREP.AddrBlk. Otherwise, RREP.PrefixLength[TargNdx] := Route[TargNode].PrefixLength according to the rules of RFC 5444 AddrBlk encoding. RREP.MetricType[TargNdx] := Route[TargNode].MetricType RREP.Metric_TLV[TargNdx] := Route[TargNode].Metric <msg-hop-count>, if included, MUST be set to 0. <msg-hop-limit> SHOULD be set to RREQ.<msg-hop-count>. IP.DestinationAddr := Route[OrigNode].NextHop An example message format for RREP is illustrated in .

Before an AODVv2 router can make use of a received RteMsg (i.e., RREQ or RREP), the router first must verify that the RteMsg is permissible according to the following steps. OrigNdx and TargNdx are set according to the rules in . For RREQ, RteMsg.Metric is Metric_TLV[OrigNdx]. For RREP, RteMsg.Metric is Metric_TLV[TargNdx]. In this section (unless qualified by additional description such as "upstream" or "neighboring") all occurrences of the term "router" refer to the AODVv2 router handling the received RteMsg. A router MUST handle RteMsgs only from neighbors as specified in . RteMsgs from other sources MUST be disregarded. The router examines the RteMsg to ascertain that it contains the required information: <msg-hop-limit>, TargNode.Addr, OrigNode.Addr, RteMsg.Metric, and either RteMsg.OrigSeqNum or RteMsg.TargSeqNum. If the required information does not exist, the message is disregarded. The router checks that OrigNode.Addr and TargNode.Addr are valid routable unicast addresses. If not, the message is disregarded. The router checks the Metric Type MsgTLV (if present) to assure that the Metric Type associated with the Metric AddrTLV information in the RREQ or RREP is known. If not, the message is disregarded. DISCUSSION: or, can change the AddrBlk metric to use HopCount, e.g., measured from <msg-hop-count>. If (MAX_METRIC[RteMsg.MetricType] - Cost(L)) ≤ RteMsg.Metric, the RteMsg is disregarded, where Cost(L) denotes the cost of traversing the incoming link (i.e., as measured by the network interface receiving the incoming RteMsg). An AODVv2 router handles a permissible RteMsg according to the following steps. The router MUST process the routing information for OrigNode and TargNode contained in the RteMsg as specified in . If RteMsg.<msg-hop-limit> is zero (0), no further action is taken, and the RteMsg is not retransmitted. Otherwise, the router MUST decrement RteMsg.<msg-hop-limit>. If the RteMsg.<msg-hop-count> is present, and MAX_HOPCOUNT <= <msg-hop-count>, then no further action is taken. Otherwise, the router MUST increment RteMsg.<msg-hop-count> Further actions to transmit an updated RteMsg depend upon whether the incoming RteMsg is an RREP or an RREQ.

By sending a RREQ, a router advertises that it will route for addresses contained in the RteMsg based on the information enclosed. The router MAY choose not to send the RREQ, though not resending the RREQ could decrease connectivity in the network or result in nonoptimal paths. The circumstances under which a router might choose not to re-transmit a RREQ are not specified in this document. Some examples might include the following: The router is already heavily loaded and does not want to advertise routing for more traffic The router recently transmitted identical routing information (e.g. in a RREQ advertising the same metric) The router is low on energy and has to reduce energy expended for sending protocol messages or packet forwarding Unless the router is prepared to send a RREQ, it halts processing. If the upstream router sending a RREQ is in the Blacklist, and Current_Time < Blacklist.RemoveTime, then the router receiving that RREQ MUST NOT transmit any outgoing RteMsg, and processing is complete. Otherwise, if the upstream router is in the Blacklist, and Current_Time ≥ Blacklist.RemoveTime, then the upstream router SHOULD be removed from the Blacklist, and message processing continued. The incoming RREQ MUST be checked against previously received information from the RREQ Table . If the information in the incoming RteMsg is redundant, then then no further action is taken. If TargNode is a client of the router receiving the RREQ, then the router generates a RREP message as specified in , and subsequently processing for the RREQ is complete. Otherwise, processing continues as follows. RREQ.MetricType := Route[OrigNode].MetricType RREQ.Metric_TLV[OrigNdx] := Route[OrigNode].Metric The RREQ (with updated fields as specified above>) SHOULD be sent to the IP multicast address LL-MANET-Routers . If the RREQ is unicast, the IP.DestinationAddress is set to Route[RREQ.TargNode].NextHopAddress.

As before, OrigNode and TargNode are named in the context of RREQ_Gen (i.e., the router originating the RREQ for which the RREP was generated) (see ). OrigNdx and TargNdx are set according to the rules in . If no forwarding route exists to OrigNode, then a RERR SHOULD be transmitted to RREP.AddrBlk[TargNdx]. Otherwise, if HandlingRtr is not RREQ_Gen then the outgoing RREP is sent to the Route.NextHopAddress for the RREP.AddrBlk[OrigNdx]. If HandlingRtr is RREQ_Gen then the RREP satisfies RREQ_Gen's earlier RREQ, and RREP processing is completed. Any packets buffered for OrigNode should be transmitted.

Since RREQ messages are multicast, there are common circumstances in which an AODVv2 router might transmit a redundant response (RREQ or RREP), duplicating the information transmitted in response to some other recent RREQ (see ). Before responding, an AODVv2 router MUST suppress such redundant RREQ messages. This is done by checking the list of recently received RREQs to determine whether the incoming RREQ contains new information, as follows: The AODVv2 router searches the RREQ Table for recent entries with the same OrigNode, TargNode, and Metric Type. If there is no such entry, the incoming RREQ message is not suppressed. A new entry for the incoming RREQ is created in the RREQ Table. If there is such an entry, and the incoming RREQ has a newer sequence number, the incoming RREQ is not suppressed, and the existing table entry MUST be updated to reflect the new Sequence Number and Metric. Similarly, if the Sequence Numbers are the same, and the incoming RREQ offers a better Metric, the incoming RREQ is not suppressed, and the RREQ Table entry MUST be updated to reflect the new Metric. Otherwise, the incoming RREQ is suppressed.

AODVv2 routers attempt to maintain active routes. When a routing problem is encountered, an AODVv2 router (denoted RERR_Gen) attempts to quickly notify upstream routers. Two kinds of routing problems may trigger generation of a RERR message. The first case happens when the router receives a packet but does not have a route for the destination of the packet. The second case happens immediately upon detection of a broken link (see ) of an Active route, to quickly notify upstream AODVv2 routers that that route is no longer available.

Before using a route to forward a packet, an AODVv2 router MUST check the status of the route as follows. If the route is marked has been marked as Broken, it cannot be used for forwarding. If Current_Time > Route.ExpirationTime, the route table entry has expired, and cannot be used for forwarding. Similarly, if (Route.ExpirationTime == MAXTIME), and if (Current_Time - Route.LastUsed) > (ACTIVE_INTERVAL + MAX_IDLETIME), the route has expired, and cannot be used for forwarding. Furthermore, if Current_Time - Route.LastUsed > (MAX_SEQNUM_LIFETIME), the route table entry MUST be expunged. If any of the above route error conditions hold true, the route cannot be used to forward the packet, and an RERR message MUST be generated (see ). Otherwise, Route.LastUsed := Current_Time, and the packet is forwarded to the route's next hop. Optionally, if a precursor list is maintained for the route, see for precursor lifetime operations.

AODVv2 routers SHOULD monitor connectivity to adjacent routers along active routes. This monitoring can be accomplished by one or several mechanisms, including: Neighborhood discovery Route timeout Lower layer trigger that a link is broken TCP timeouts Promiscuous listening Other monitoring mechanisms or heuristics If a next-hop AODVv2 router has become unreachable, RERR_Gen follows the procedures specified in .

An RERR message is generated by a AODVv2 router (i.e., RERR_Gen) in order to notify upstream routers that packets cannot be delivered to certain destinations. An RERR message has the following general structure:

| +---------------------------------------------------------------+ | UnreachableNode AddrBlk (Unreachable Node addresses) | +---------------------------------------------------------------+ | UnreachableNode SeqNum AddrBlk TLV | +---------------------------------------------------------------+ ]]>

The RERR MUST contain the following: <msg-hop-limit> PktSource Message TLV (see ), if the RERR is unicast Metric Type Message TLV (see ), if Metric Type != 3 The RERR SHOULD contain the following: <msg-hop-count> This Address Block contains the IP addresses unreachable by AODVv2 router transmitting the RERR. This Address Block TLV carries the destination sequence number associated with each UnreachableNode when that information is available. The prefix length associated with an UnreachableNode. There are two kinds of events indicating that packets cannot be delivered to certain destinations. The two cases differ in the way that the neighboring IP destination address for the RERR is chosen, and in the way that the set of UnreachableNodes is identified. In both cases, the <msg-hop-limit> MUST be included and SHOULD be set to MAX_HOPCOUNT. <msg-hop-count> SHOULD be included and set to 0, to facilitate use of various route repair strategies including expanding rings multicast and Intermediate RREP .

The first case happens when the router receives a packet from another AODVv2 router but does not have a valid route for the destination of the packet. In this case, there is exactly one UnreachableNode to be included in the RERR's AddrBlk (either IP.DestinationAddress from a data packet or RREP.AddrBlk[OrigNode]). The RERR SHOULD be sent to the multicast address LL-MANET-Routers, but RERR_Gen MAY instead send the RERR to the next hop towards the source IP address of the packet which was undeliverable. For unicast RERR, the PktSource Message TLV MUST be included, containing the the source IP address of the undeliverable packet, or the IP address of TargRtr in case the undeliverable packet was an RREP message generated by TargRtr. If a Sequence Number for UnreachableNode is known, that Sequence Number SHOULD be included in a Seqnum AddrTLV the RERR. Otherwise all nodes handling the RERR will assume their route through RERR_Gen towards the UnreachableNode is no longer valid and flag those routes as broken, regardless of the Sequence Number information for those routes. RERR_Gen MUST discard the packet or message that triggered generation of the RERR. If an AODVv2 router receives an ICMP packet from the address of one of its client nodes, it simply relays the packet to the ICMP packet's destination address, and does not generate any RERR message.

The second case happens when the link breaks to an active adjacent AODVv2 router (i.e., the next hop of an active route). In this case, the RERR MUST be sent to the multicast address LL-MANET-Routers, except when the optional feature of maintaining precursor lists is used as specified in . All routes (Active, Idle and Expired) that use the broken link MUST be marked as Broken. The set of UnreachableNodes is initialized by identifying those Active routes which use the broken link. For each such Active Route, Route.Dest is added to the set of Unreachable Nodes. After the Active Routes using the broken link have all been included as UnreachableNodes, Idle routes MAY also be included, if allowed by the setting of ENABLE_IDLE_UNREACHABLE, as long as the packet size of the RERR does not exceed the MTU (interface "Maximum Transfer Unit") of the physical medium. If the set of UnreachableNodes is empty, no RERR is generated. Otherwise, RERR_Gen generates a new RERR, and the address of each UnreachableNode is inserted into an AddrBlock. The value for each UnreachableNode's SeqNum (UnreachableNode.SeqNum) MUST be placed in the SeqNum AddrTLV. If none of UnreachableNode.Addr entries are associated with routing prefixes (i.e., no prefix length is shorter than the maximum length for the address family), then the AddrBlk SHOULD NOT include any prefix-length information. Every broken route reported in the RERR MUST have the same Metric Type. If the Metric Type is not 3, then the RERR message MUST contain a Metric Type MsgTLV indicating the Metric Type of the broken route(s).

When an AODVv2 router (HandlingRtr) receives a RERR message, it uses the information provided to invalidate affected routes. If the information in the RERR may be relevant to upstream neighbors using those routes, HandlingRtr subsequently sends another RERR to those neighbors. This operation has the effect of retransmitting the RERR information and is counted as another "hop" for purposes of properly modifying <msg-hop-limit> and <msg-hop-count> in the RERR message header. HandlingRtr examines the incoming RERR to assure that it contains <msg-hop-limit> and at least one UnreachableNode.Address. If the required information does not exist, the incoming RERR message is disregarded and further processing stopped. Otherwise, for each UnreachableNode.Address, HandlingRtr searches its route table for a route using longest prefix matching. If no such Route is found, processing is complete for that UnreachableNode.Address. Otherwise, HandlingRtr verifies the following: The UnreachableNode.Address is a routable unicast address. Route.NextHopAddress is the same as RERR IP.SourceAddress. Route.NextHopInterface is the same as the interface on which the RERR was received. The UnreachableNode.SeqNum is unknown, OR Route.SeqNum <= UnreachableNode.SeqNum (using signed 16-bit arithmetic). If the Route satisfies all of the above conditions, HandlingRtr checks whether Route.PrefixLength is the same as the prefix length for UnreachableNode.Address. If so, HandlingRtr simply sets the Route.Broken flag for that Route. Otherwise, HandlingRtr creates a new route (call it BrokenRoute) with the same PrefixLength as the prefix length for UnreachableNode.Address, and sets the BrokenRoute.Broken flag for that new route. If the prefix length for BrokenRoute is shorter than Route.PrefixLength, then Route MUST be expunged from the routing table (since it is a subroute of the larger route which is reported to be broken). Furthermore, if <msg-hop-limit> is greater than 0, then HandlingRtr adds the UnreachableNode address and TLV information to an AddrBlk for delivery in the outgoing RERR message. If there are no UnreachableNode addresses to be transmitted in an RERR to upstream routers, HandlingRtr MUST discard the RERR, and no further action is taken. Otherwise, <msg-hop-limit> is decremented by one (1) and processing continues as follows: (Optional) If precursor lists are maintained, the outgoing RERR SHOULD be sent to the active precursors of the broken route as specified in . Otherwise, if the incoming RERR message was received at the LL-MANET-Routers multicast address, the outgoing RERR SHOULD also be sent to LL-MANET-Routers. Otherwise, if the PktSource Message TLV is present, and HandlingRtr has a Route to PktSource.Addr, then HandlingRtr MUST send the outgoing RERR to Route[PktSource.Addr].NextHop. Otherwise, the outgoing RERR MUST be sent to LL-MANET-Routers.

If a message with an unknown type is received, the message is disregarded. For handling of messages that contain unknown TLV types, ignore the information for processing, but preserve it unmodified for forwarding.

Simple Internet attachment means attachment of a stub (i.e., non-transit) network of AODVv2 routers to the Internet via a single Internet AODVv2 router (called IAR). As in any Internet-attached network, AODVv2 routers, and their clients, wishing to be reachable from hosts on the Internet MUST have IP addresses within the IAR's routable and topologically correct prefix (e.g. 191.0.2.0/24).

When an AODVv2 router within the AODVv2 MANET wants to discover a route toward a node on the Internet, it uses the normal AODVv2 route discovery for that IP Destination Address. The IAR MUST respond to RREQ on behalf of all Internet destinations. When a packet from a node on the Internet destined for a node in the AODVv2 MANET reaches the IAR, if the IAR does not have a route toward that destination it will perform normal AODVv2 route discovery for that destination.

AODVv2 may be used with multiple interfaces; therefore, the particular interface over which packets arrive MUST be known whenever a packet is received. Whenever a new route is created, the interface through which the Route.Address can be reached is also recorded in the route table entry. When multiple interfaces are available, a node transmitting a multicast packet with IP.DestinationAddress set to LL-MANET-Routers SHOULD send the packet on all interfaces that have been configured for AODVv2 operation. Similarly, AODVv2 routers SHOULD subscribe to LL-MANET-Routers on all their AODVv2 interfaces.

To avoid messaging overload, each AODVv2 router's rate of packet/message generation SHOULD be limited. The rate and algorithm for limiting messages (CONTROL_TRAFFIC_LIMITS) is left to the implementor and should be administratively configurable. AODVv2 messages SHOULD be discarded in the following order of preference: RREQ, RREP, and finally RERR.

Some optional features of AODVv2, associated with AODV, are not required by minimal implementations. These features are expected to apply in networks with greater mobility, or larger node populations, or requiring reduced latency for application launches. The optional features are as follows: Expanding Rings Multicast Intermediate RREPs (iRREPs): Without iRREP, only the destination can respond to a RREQ. Precursor lists. Reporting Multiple Unreachable Nodes. An RERR message can carry more than one Unreachable Destination node for cases when a single link breakage causes multiple destinations to become unreachable from an intermediate router. RREP_ACK. Message Aggregation.

For multicast RREQ, <msg-hop-limit> MAY be set in accordance with an expanding ring search as described in to limit the RREQ propagation to a subset of the local network and possibly reduce route discovery overhead.

This specification has been published as a separate Internet Draft .

This section specifies an interoperable enhancement to AODVv2 (and possibly other reactive routing protocols) enabling more economical notifications to active sources of traffic upon determination that a route needed to forward such traffic to its destination has become Broken.

In many circumstances, there can be several sources of traffic for a certain destination. Each such source of traffic is known as a "precursor" for the destination, as well as all upstream routers between the forwarding AODVv2 router and the traffic source. For each active destination, an AODVv2 router MAY choose to keep track of the upstream neighbors that have provided traffic for that destination; there is no need to keep track of upstream routers any farther away than the next hop. Moreover, any particular link to an adjacent AODVv2 router may be a path component of multiple routes towards various destinations. The precursors for all destinations using the next hop across any link are collectively known as the precursors for that next hop. When an AODVv2 router determines that an active link to one of its downstream neighbors has broken, the AODVv2 router detecting the broken link must mark multiple routes as Broken, for each of the newly unreachable destinations, as described in . Each route that relies on the newly broken link is no longer valid. Furthermore, the precursors of the broken link should be notified (using RERR) about the change in status of their route to a destination downstream along the broken next hop.

During normal operation, each AODVv2 router wishing to maintain precursor lists as described above, maintains a precursor table and updates the table whenever the node forwards traffic to one of the destinations in its route table. For each precursor in the precursor list, a record must be maintained to indicate whether the precursor has been used for recent traffic (in other words, whether the precursor is an Active precursor). So, when traffic arrives from a precursor, the Current_Time is used to mark the time of last use for the precursor list element associated with that precursor. When an AODVv2 router detects that a link is broken, then for each precursor using that next hop, the node MAY notify the precursor using either unicast or multicast RERR: This option is applicable when there are few Active precursors compared to the number of neighboring AODVv2 routers. RERR_PRECURSORS is, by default, LL-MANET-Routers . This option is typically preferable when there are many precursors, since fewer packet transmissions are required. Each active upstream neighbor (i.e., precursor) MAY then execute the same procedure until all active upstream routers have received the RERR notification.

The RREQ Target Router (RREP_Gen) MAY, as an alternative to unicasting a RREP, be configured to distribute routing information about the route toward the RREQ TargNode (RREP_Gen's client) more widely. That is, RREP_Gen MAY be configured respond to a route discovery by generating a RREP, using the procedure in , but multicasting the RREP to LL-MANET-Routers (subject to similar suppression algorithm for redundant RREP multicasts as described in ). The redundant message suppression must occur at every router handling the multicast RREP. Afterwards, RREP_Gen processing for the incoming RREQ is complete. Broadcast RREP response to incoming RREQ was originally specified to handle unidirectional links, but it is expensive. Due to the significant overhead, AODVv2 routers MUST NOT use multicast RREP unless configured to do so by setting the administrative parameter USE_MULTICAST_RREP.

Instead of relying on existing mechanisms for requesting verification of link bidirectionality during Route Discovery, RREP_Ack is provided as an optional feature and modeled on the RREP_Ack message type from AODV . Since the RREP_ACK is simply echoed back to the node from which the RREP was received, there is no need for any additional RFC 5444 address information (or TLVs). Considerations of packet TTL are as specified in . An example message format is illustrated in section .

The aggregation of multiple messages into a packet is specified in RFC 5444 . Implementations MAY choose to briefly delay transmission of messages for the purpose of aggregation (into a single packet) or to improve performance by using jitter .

AODVv2 uses various configurable parameters of various types: Timers Protocol constants Administrative (functional) controls Other administrative parameters and lists The tables in the following sections show the parameters along their definitions and default values (if any). Note: several fields have limited size (bits or bytes). These sizes and their encoding may place specific limitations on the values that can be set. For example, <msg-hop-count> is a 8-bit field and therefore MAX_HOPCOUNT cannot be larger than 255.

AODVv2 requires certain timing information to be associated with route table entries. The default values are as follows, subject to future experience: Name Default Value ACTIVE_INTERVAL 5 second MAX_IDLETIME 200 seconds MAX_BLACKLIST_TIME 200 seconds MAX_SEQNUM_LIFETIME 300 seconds RREQ_WAIT_TIME 2 seconds UNICAST_MESSAGE_SENT_TIMEOUT 1 second RREQ_HOLDDOWN_TIME 10 seconds The above timing parameter values have worked well for small and medium well-connected networks with moderate topology changes. The timing parameters SHOULD be administratively configurable for the network where AODVv2 is used. Ideally, for networks with frequent topology changes the AODVv2 parameters should be adjusted using either experimentally determined values or dynamic adaptation. For example, in networks with infrequent topology changes MAX_IDLETIME may be set to a much larger value.

AODVv2 protocol constants typically do not require changes. The following table lists these constants, along with their values and a reference to the specification describing their use. Name Default Value Description DISCOVERY_ATTEMPTS_MAX 3 MAX_HOPCOUNT 20 hops MAX_METRIC[i] Specified only for HopCount MAXTIME [TBD] Maximum expressible clock time

The following administrative controls may be used to change the operation of the network, by enabling optional behaviors. These options are not required for correct routing behavior, although they may potentially reduce AODVv2 protocol messaging in certain situations. The default behavior is to NOT enable most such options, options. Packet buffering is enabled by default. Name Description DEFAULT_METRIC_TYPE 3 {Hop Count (see )} ENABLE_IDLE_UNREACHABLE ENABLE_IRREP USE_MULTICAST_RREP

The following table lists contains AODVv2 parameters which should be administratively configured for each specific network. Name Default Value Cross Reference AODVv2_INTERFACES BUFFER_SIZE_PACKETS 2 BUFFER_SIZE_BYTES MAX_PACKET_SIZE [TBD] CLIENT_ADDRESSES AODVv2_INTERFACES CONTROL_TRAFFIC_LIMIT TBD [50 packets/sec?]

This section specifies several message types, message tlv-types, and address tlv-types. Also, a new registry of 16-bit alternate metric types is specified.

Name Type (TBD) Route Request (RREQ) 10 Route Reply (RREP) 11 Route Error (RERR) 12 Route Reply Acknowledgement (RREP_ACK) 13

Name Type (TBD) Length in octets Cross Reference Acknowledgment Request (AckReq) 10 0 Destination RREP Only (DestOnly) 11 0 Packet Source (PktSource) 12 4 or 16 Metric Type 13 1

Name Type (TBD) Length Value Metric 10 depends on Metric Type Sequence Number (SeqNum) 11 2 octets Originating Node Sequence Number (OrigSeqNum) 12 2 octets Target Node Sequence Number (TargSeqNum) 13 2 octets VALIDITY_TIME 1 1 octet

Metric types are identified according to the assignments as specified in . The metric type of the Hop Count metric is assigned to be 3, in order to maintain compatibility with that existing table of values from RFC 6551. Non-addititve metrics are not supported in this draft. Name Type Metric Size Unallocated 0 -- 2 TBD Hop Count 3 - TBD 1 octet Unallocated 4 -- 254 TBD Reserved 255 Undefined

The objective of the AODVv2 protocol is for each router to communicate reachability information about addresses for which it is responsible. Positive routing information (i.e. a route exists) is distributed via RteMsgs and negative routing information (i.e. a route does not exist) via RERRs. AODVv2 routers that handle these messages store the contained information to properly forward data packets, and they generally provide this information to other AODVv2 routers. This section does not mandate any specific security measures. Instead, this section describes various security considerations and potential avenues to secure AODVv2 routing. The most important security mechanisms for AODVv2 routing are integrity/authentication and confidentiality. In situations where routing information or router identity are suspect, integrity and authentication techniques SHOULD be applied to AODVv2 messages. In these situations, routing information that is distributed over multiple hops SHOULD also verify the integrity and identity of information based on originator of the routing information. A digital signature could be used to identify the source of AODVv2 messages and information, along with its authenticity. A nonce or timestamp SHOULD also be used to protect against replay attacks. S/MIME and OpenPGP are two authentication/integrity protocols that could be adapted for this purpose. In situations where confidentiality of AODVv2 messages is important, cryptographic techniques can be applied. In certain situations, for example sending a RREP or RERR, an AODVv2 router could include proof that it has previously received valid routing information to reach the destination, at one point of time in the past. In situations where routers are suspected of transmitting maliciously erroneous information, the original routing information along with its security credentials SHOULD be included. Note that if multicast is used, any confidentiality and integrity algorithms used MUST permit multiple receivers to handle the message. Routing protocols, however, are prime targets for impersonation attacks. In networks where the node membership is not known, it is difficult to determine the occurrence of impersonation attacks, and security prevention techniques are difficult at best. However, when the network membership is known and there is a danger of such attacks, AODVv2 messages must be protected by the use of authentication techniques, such as those involving generation of unforgeable and cryptographically strong message digests or digital signatures. While AODVv2 does not place restrictions on the authentication mechanism used for this purpose, IPsec Authentication Message (AH) is an appropriate choice for cases where the nodes share an appropriate security association that enables the use of AH. In particular, routing messages SHOULD be authenticated to avoid creation of spurious routes to a destination. Otherwise, an attacker could masquerade as that destination and maliciously deny service to the destination and/or maliciously inspect and consume traffic intended for delivery to the destination. RERR messages SHOULD be authenticated in order to prevent malicious nodes from disrupting active routes between communicating nodes. If the mobile nodes in the ad hoc network have pre-established security associations, the purposes for which the security associations are created should include that of authorizing the processing of AODVv2 control packets. Given this understanding, the mobile nodes should be able to use the same authentication mechanisms based on their IP addresses as they would have used otherwise. If the mobile nodes in the ad hoc network have pre-established security associations, the purposes for which the security associations Most AODVv2 messages are transmitted to the multicast address LL-MANET-Routers . It is therefore required for security that AODVv2 neighbors exchange security information that can be used to insert an ICV into the AODVv2 message block . This enables hop-by-hop security, which is proper for these message types that may have mutable fields. For destination-only RREP discovery procedures, AODVv2 routers that share a security association SHOULD use the appropriate mechanisms as specified in RFC 6621. The establishment of these security associations is out of scope for this document.

AODVv2 is a descendant of the design of previous MANET on-demand protocols, especially AODV and DSR . Changes to previous MANET on-demand protocols stem from research and implementation experiences. Thanks to Elizabeth Belding-Royer for her long time authorship of AODV. Additional thanks to Derek Atkins, Emmanuel Baccelli, Ramon Caceres, Thomas Clausen, Christopher Dearlove, Ulrich Herberg, Henner Jakob, Luke Klein-Berndt, Lars Kristensen, Tronje Krop, Koojana Kuladinithi, Alexandru Petrescu, Henning Rogge, Fransisco Ros, Pedro Ruiz, Christoph Sommer, Lotte Steenbrink, Romain Thouvenin, Jiazi Yi, Seung Yi, and Cong Yuan, for their reviews AODVv2 and DYMO, as well as several specification suggestions.

The following subsections show example RFC 5444-compliant packets for AODVv2 message types RREQ, RREP, RERR, and RREP-Ack. These proposed message formats are designed based on expected savings from IPv6 addressable MANET nodes, and a layout for the Address TLVs that may be viewed as natural, even if perhaps not the absolute most compact possible encoding. For RteMsgs, the msg-hdr fields are followed by at least one and optionally two Address Blocks. The first AddrBlk contains OrigNode and TargNode. For each AddrBlk, there must be AddrTLVs of type Metric and one of the SeqNum types (i.e, OrigSeqNum, TargSeqNum, or Seqnum). There is no Metric Type Message TLV present, so the Metric AddrTLV measures HopCount. The Metric AddrTLV also provides a way for the AODV router generating the RREQ or RREP to supply an initial nonzero cost for the route to its client node (OrigNode or TargNode, for RREQ or RREP respectively). In all cases, the length of an address (32 bits for IPv4 and 128 bits for IPv6) inside an AODVv2 message is indicated by the msg-addr-length (MAL) in the msg-header, as specified in . The RFC 5444 header preceding AODVv2 messages in this document has the format illustrated in .

The fields in are to be interpreted as follows: PV=0 (Packet Header Version = 0) PF=0 (Packet Flags = 0)

illustrates an example RREQ message format.

The fields in are to be interpreted as follows: msg-type=RREQ (first [and only] message is of type RREQ) MF=4 (Message Flags = 4 [only msg-hop-limit field is present]) MAL=3 (Message Address Length indicator [3 for IPv4, 15 for IPv6]) msg-size=28 (octets -- counting MsgHdr, MsgTLVs, and AddrBlks) msg-hop-limit (initially MAX_HOPCOUNT by default) msg.tlvs-length=0 (no Message TLVs) num-addr=2 (OrigNode and TargNode addresses in RteMsg AddrBlock) AddrBlk flags: bit 0 (ahashead): 1 bit 1 (ahasfulltail): 0 bit 2 (ahaszerotail): 0 bit 3 (ahassingleprelen): 0 bit 4 (ahasmultiprelen): 0 bits 5-7: RESERVED head-length=3 (length of head part of each address is 3 octets) Head (3 initial bytes for both Originating & Target addresses) Orig.Tail (4th byte of Originating Node IP address) Target.Tail (4th byte of Target Node IP address) addr.TLV.len = 11 (length in bytes for OrigSeqNum and Metric TLVs type=OrigSeqNum (type of first AddrBlk TLV, value 2 octets) AddrTLV flags for the OrigSeqNum TLV: bit 0 (thastypeext): 0 bit 1 (thassingleindex): 1 bit 2 (thasmultiindex): 0 bit 3 (thasvalue): 1 bit 4 (thasextlen): 0 bit 5 (tismultivalue): 0 bits 6-7: RESERVED Index-start=0 (OrigSeqNum TLV value applies at index 0) tlv-length=2 (so there is only one TLV value, [1 = 2/2]) Orig.Node Sequence # (TLV value for the OrigSeqNum TLV type=Metric (AddrTLV type of second AddrBlk TLV, values 1 octet) AddrTLV flags for Metric_TLV: bit 0 (thastypeext): 0 bit 1 (thassingleindex): 1 bit 2 (thasmultiindex): 0 bit 3 (thasvalue): 1 bit 4 (thasextlen): 0 bit 5 (tismultivalue): 0 bits 6-7: RESERVED Index-start=0 (Metric TLV values start at index 0) tlv-length=1 (so there is only one TLV value, [1 = 1/1]) OrigNodeHopCt (first [and only] TLV value for the Metric TLV)

illustrates a packet format for an example RREP message.

The fields in are to be interpreted as follows: msg-type=RREP (first [and only] message is of type RREP) MF=4 (Message Flags = 4 [only msg-hop-limit field is present]) MAL=3 (Message Address Length indicator [3 for IPv4, 15 for IPv6]) msg-size=28 (octets -- counting MsgHdr, MsgTLVs, and AddrBlks) msg-hop-limit (initially MAX_HOPCOUNT by default) msg.tlvs-length=0 (no Message TLVs) num-addr=2 (OrigNode and TargNode addresses in RteMsg AddrBlock) AddrBlk flags: bit 0 (ahashead): 1 bit 1 (ahasfulltail): 0 bit 2 (ahaszerotail): 0 bit 3 (ahassingleprelen): 0 bit 4 (ahasmultiprelen): 0 bits 5-7: RESERVED head-length=3 (length of head part of each address is 3 octets) Head (3 initial bytes for both Originating & Target addresses) Orig.Tail (4th byte of Originating Node IP address) Target.Tail (4th byte of Target Node IP address) addr.TLV.len = 11 (length in bytes for TargSeqNum TLV and Metric TLV type=TargSeqNum (type of first AddrBlk TLV, value 2 octets) AddrTLV flags for the TargSeqNum TLV: bit 0 (thastypeext): 0 bit 1 (thassingleindex): 1 bit 2 (thasmultiindex): 0 bit 3 (thasvalue): 1 bit 4 (thasextlen): 0 bit 5 (tismultivalue): 0 bits 6-7: RESERVED Index-start=1 (TargSeqNum TLV value applies to address at index 1) tlv-length=2 (there is one TLV value, 2 bytes in length) Targ.Node Sequence # (value for the TargSeqNum TLV) type=Metric (AddrTLV type of second AddrBlk TLV, value 1 octet) AddrTLV flags for the Metric TLV [01010000, same as for TargSeqNum TLV] Index-start=1 (Metric TLV values start at index 1) tlv-length=1 (there is one TLV value, 1 byte in length) TargNodeHopCt (first [and only] TLV value for Metric TLV)

illustrates an example RERR message format.

The fields in are to be interpreted as follows: msg-type=RERR (first [and only] message is of type RERR) MF=4 (Message Flags = 4 [only msg-hop-limit field is present]) MAL=3 (Message Address Length indicator [3 for IPv4, 15 for IPv6]) msg-size=24 (octets -- counting MsgHdr, MsgTLVs, and AddrBlks) msg-hop-limit (initially MAX_HOPCOUNT by default) msg.tlvs-length=0 (no Message TLVs) num-addr=2 (OrigNode and TargNode addresses in RteMsg AddrBlock) AddrBlk flags == 10000000 [same as RREQ and RREP AddrBlk examples] head-length=3 (length of head part of each address is 3 octets) Head (3 initial bytes for both Unreachable Nodes, Dest_1 and Dest_2) Dest_1.Tail (4th byte of Dest_1 IP address) Dest_2.Tail (4th byte of Dest_2 IP address) addr.TLV.len = 7 (length in bytes for SeqNum TLV type=SeqNum (AddrTLV type of AddrBlk TLV, values 2 octets each) AddrTLV flags for SeqNum TLV: bit 0 (thastypeext): 0 bit 1 (thassingleindex): 0 bit 2 (thasmultiindex): 1 bit 3 (thasvalue): 1 bit 4 (thasextlen): 0 bit 5 (tismultivalue): 1 bits 6-7: RESERVED tlv-length=4 (so there are two TLV values, [2 = 4/2]) Dest_1 Sequence # (first of two TLV values for the SeqNum TLV) Dest_2 Sequence # (second of two TLV values for the SeqNum TLV)

The figure below illustrates a packet format for an example RREP_ACK message.

This section lists the changes since AODVv2 revision ...-02.txt The "Added Node" feature was removed. This feature was intended to enable additional routing information to be carried within a RREQ or a RREP message, thus increasing the amount of topological information available to nodes along a routing path. However, enlarging the packet size to include information which might never be used can increase congestion of the wireless medium. The feature can be included as an optional feature at a later date when better algorithms are understood for determining when the inclusion of additional routing information might be worthwhile. Numerous editorial improvements and improvements to consistent terminology were made. Instances of OrigNodeNdx and TargNodeNdx were replaced by OrigNdx and TargNdx, to be consistent with the terminology shown in . Example RREQ and RREP message formats shown in the Appendices were changed to use OrigSeqNum and TargSeqNum message TLVs instead of using the SeqNum message TLV. Inclusion of the OrigNode's SeqNum in the RREP message is not specified. The processing rules for the OrigNode's SeqNum were incompletely specified in previous versions of the draft, and very little benefit is foreseen for including that information, since reverse path forwarding is used for the RREP. Additional acknowledgements were included, and contributors names were alphabetized. Definitions in the Terminology section capitalize the term to be defined. Uncited bibliographic entries deleted. Ancient "Changes" sections were deleted.

Only one AODVv2 router within a MANET SHOULD be responsible for a particular address at any time. If two AODVv2 routers dynamically shift the advertisement of a network prefix, correct AODVv2 routing behavior must be observed. The AODVv2 router adding the new network prefix must wait for any existing routing information about this network prefix to be purged from the network. Therefore, it must wait at least ROUTER_SEQNUM_AGE_MAX_TIMEOUT after the previous AODVv2 router for this address stopped advertising routing information on its behalf.