Myung-Ki Shin, ETRI Hee-Jin Jang, Samsung AIT IETF-65 16ng BoF @ Dallas, TX

1. Transmission of IPv6 Packets over IEEE 802.16draft-shin-16ng-ipv6-transmission-00draft-shin-ipv6-ieee802.16-02 Myung-Ki Shin, ETRI Hee-Jin Jang, Samsung AIT IETF-65 16ng BoF @ Dallas, TX

2. Goal and Scope • The characteristics of IEEE 802.16 networks (links) put special considerations on how IPv6 is used • draft-jee-16ng-ps-goals-00 • Subnet models and scenarios have an influence on how to transmit IPv6 packets over IEEE 802.16 • Describe IPv6 operations and methods for IPv6 packet transmission over IEEE 802.16 • Maximum Transmission Unit • Stateless Autoconfiguration and Link-Local Addresses • Frame Format and Encapsulation • Unicast and Multicast Transmission Methods

3. IEEE 802.16 MAC Header • There is no source or destination MAC address in IEEE 802.16 the MAC header format. • Instead, CID is used to identify connections to equivalent peers in the MAC of the BS and the MS in IEEE 802.16 networks. • [IEEE802.16] defines several CSs for carrying IP packets. • Frame formats are different according to the CS types. • IP(v6) CS vs. Ethernet CS. 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |H|E| Type |R|C|EKS|R|LEN | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | LEN LSB | CID MSB | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | CID LSB | HCS | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

4. Frame Format 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |H|E| Type |R|C|EKS|R|LEN | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | LEN LSB | CID MSB | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | CID LSB | HCS | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | IPv6 | +- -+ | header | +- -+ | and | +- -+ / payload ... / +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |H|E| Type |R|C|EKS|R|LEN | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | LEN LSB | CID MSB | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | CID LSB | HCS | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Ethernet | +- -+ / header / / / +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | IPv6 | +- -+ | header | +- -+ | and | +- -+ / payload ... / +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Frame Format for IP(v6) CS Frame Format for Ethernet CS

5. Stateless Auto-configuration and Link-Local Addresses • Like other IEEE 802 interfaces, the Interface Identifier for an IEEE 802.16 interface is based on the EUI-64 identifier derived from the interface's built-in 48-bit IEEE 802 address. • The IPv6 link-local address for an IEEE 802.16 interface is formed by appending the Interface Identifier, to the prefix FE80::/64. 10 bits 54 bits 64 bits +----------+-----------------------+----------------------------+ |1111111010| (zeros) | Interface Identifier | +----------+-----------------------+----------------------------+

6. Address Mapping • IP(v6) CS • Mapping unicast or multicast addresses to IEEE 802.16 MAC addresses is unnecessary, since IEEE 802.16 MAC header does not contain any source or destination 802 MAC addresses. • To identify connections to equivalent peers in the MAC of the BS/router and the MS, CID is used. • Multicast CID may be provided for in the downlink. • Ethernet CS • Source and destination Ethernet addresses are required to form the frame for Ethernet CS. In such case, address mapping is the same as [RFC1972].

7. Network/Subnet Model A • A BS is integrated with a router, composing one box in view of implementation. • A subnet consists of only single BS/router and single MS. +-----+ | MS1 |<-------------+ +-----+ v +-----+ +-------+ +--------+ | MS2 |<---------->|BS/AR1 |---------| Edge | ISP +-----+ +-------+ | Router +==>Network +--------+ +-----+ +-------+ | | Ms3 |<---------->|BS/AR2 |-----------+ +-----+ +-------+ <---> IP termination

8. Network/Subnet Model B • A BS is integrated with a router, composing one box in view of implementation. • A subnet consists of only single BS/AR and multiple MSs. +-----+ | MS1 |<------+ +-----+ | +-----+ | +-------+ +--------+ | MS2 |<------+--->|BS/AR1 |---------| Edge | ISP +-----+ +-------+ | Router +==>Network +--------+ +-----+ +-------+ | | Ms3 |<---------->|BS/AR2 |-----------+ +-----+ +-------+ <---> IP termination

9. Network/Subnet Model C • A BS is separated from a router • A subnet consists of only single BS and single router and multiple MSs. +-----+ | MS1 |<------+ +-----+ | +-----+ | +-----+ +-----+ +--------+ | MSs |<------+----| BS1 |---->| AR |----| Edge | ISP +-----+ +-----+ +-----+ | Router +==>Network ^ +--------+ +-----+ +-----+ | | Mss |<-----------| BS2 |--------+ +-----+ +-----+ <---> IP termination

10. Network/Subnet Model D • A BS is integrated with a router, composing one box in view of implementation. • A subnet consists of multiple BS and multiple MSs. +-----+ +-----+ +-----+ ISP 1 | MS1 |<-----+ +->| AR1 |----| ER1 |===>Network +-----+ | | +-----+ +-----+ +-----+ | +-----+ | | MS2 |<-----+-----| BS1 |--| +-----+ +-----+ | +-----+ +-----+ ISP 2 +->| AR2 |----| ER2 |===>Network +-----+ +-----+ | +-----+ +-----+ | Ms3 |<-----------| BS2 |--+ +-----+ +-----+ <---> IP termination

11. Appendix – IPv6 Operations

12. Network/Subnet Model A • Unicast • Outbound packet from MS is always forwarded on a particular transport connection in the uplink direction to the BS/AR. • When BS/AR receives the packet destined to same subnet from MS, it does not relay the packet anymore. • Otherwise, BS/AR forwards the packet to the edge router. • Multicast • Outbound packet from the MS is always forwarded on a particular transport connection in the uplink direction to the BS/AR. • When BS/AR receives the packet with link-local scope from MS, it does not forward the packet anymore. • When BS/AR receives the packet with non-link-local scope from MS, it looks up the IPv6 multicast routing table and forwards the packets to the edge router.

13. Network/Subnet Model B • Unicast (Difference from Model A) • When BS/AR receives the packet destined to same subnet from MS but not to itself, it forwards to downlink after uplink CID is replaced with the corresponding downlink CID (which may be associated with the specified Ethernet addresses in Ethernet CS or IP addresses in Ethernet/IP CS) • Multicast (Difference from Model A) • When BS/AR receives the packet with link-local scope from MS, it sends back the packet to the downlink by using CID for multicast.

14. Network/Subnet Model C • Unicast (Difference from Model B) • When BS receives the packet destined to different subnet from MS, BS decapsulates the 802.16 header and forwards the packet to AR. • In case of Ethernet CS, it will be delivered to the AR naturally since the destination address in Ethernet header is the AR's address. • In case of IP CS, the BS is responsible to deliver it with the proper Ethernet header (L2 header swapping). AR performs routing and then forwards it to other BS or edge router according to the routing result.

15. Network/Subnet Model C, cont,. • Multicast (Difference from Model B) • When BS receives the packet with link-local scope from MS, it sends back the packet to the downlink by using CID for multicast. It also sends the packet to an AR after decapsulating 802.16 header • When BS receives the packet with non-link-local scope from MS, the packet is sent back to the downlink by using CID for multicast. It is also sent to the AR after decapsulating 802.16 header. • In case of Ethernet CS, it will be delivered to the AR naturally since the destination address in Ethernet header is the multicast address. • In case of IP CS, the BS is responsible to deliver it with the proper Ethernet header (L2 header swapping). AR looks up the IPv6 multicast routing table and then forwards the packets to other BSs or edge router according to the result.

16. Network/Subnet Model D • Unicast (Difference from Model C) • When BS/AR receives any packets from MS, BS decapsulates the 802.16 header and forwards the packet onto the wired network. • In case of Ethernet CS, if the packet is destined to one of ARs, it will be delivered to the AR naturally since the destination address in Ethernet header is the AR's address. If the packet is destined to MS under other BS, the target BS under which the MS exists should catch this packet instead by acting as a proxy of the MS and send it to MS by using corresponding downlink CID. • In case of IP CS, if the packet is destined to one of ARs, the BS is responsible to deliver it with the proper Ethernet header. If the packet is destined to MS under other BS, the target BS should catch this packet instead by acting as a proxy of the MS and send it to MS by using corresponding downlink CID.

17. Network/Subnet Model D, cont. • Multicast (Difference from Model C) • When BS receives the packet with link-local scope from MS, it sends back the packet to the downlink by using CID for multicast. It also sends the packet onto the wired network after decapsulating 802.16 header. ARs will get this, and other BSs will receive this and send it by using CID for multicast. • When BS receives the packet with non-link-local scope from MS, it sends back the packet to the downlink by using CID for multicast. It also sends the packet onto the wired network after decapsulating the 802.16 header. Other BSs will receive this and send it by using CID for multicast. The multicast router receiving this will look up the IPv6 multicast routing table and then forwards the packets to edge router according to the result.