Distributed Mobility Management without tunnels or specialized protocols
Why mobility is hard today? A good mobility management solution requires at least two sets of identifiers: i) application names that do not change as the mobile host moves, so that communication endpoints have stable identities regardless of their location; and ii) addresses that change as the host moves, to reflect the position of the device in the network and allow routing to scale. The fundamental problem of mobility management in the Internet is that the only identifier assigned to an entity in a Mobile Host (MH) that has scope greater than the system itself is the IP address. Hence there is a conflict: a single identifier cannot satisfy both properties at the same time.
What about RINA networks? Mobility management is a simpler problem to solve with a complete naming and addressing architecture in place. Mobility in RINA doesn\textquoteright t require setting up tunnels, re-writing packet headers or using special protocols: it is just achieved by utilizing the tools the architecture provides and that are used for normal operation, albeit using them a bit more frequently. Mobility is supported by a combination of routing updates, changing addresses of IPCPs and designing the number and size of layers in different parts of the network to accommodate the load, scale, and rate of change of the (in this case) mobile terminals to be supported. But nothing more than what one would do to design a network for any other purpose. All standard procedures that can be performed in any RINA network, mobility is no special case.
Mobile devices can roam through multiple provider networks and multiple access media without causing service disruption at the application level: service continuity is preserved as long as the destination applications are reachable. Moreover, different applications may access different services from multiple providers, utilising multiple underlying physical media simultaneously or as a failover. Mobility, therefore, is not restricted to a single layer: since all layers have the same infrastructure and protocol frameworks, all of them are capable of dealing with systems that move using the same tools. This property can be used by network designers to adapt to the density of users and rate of mobility that their networks must support.
Do you want to learn more?
- Read the paper "Mobility management in RINA networks: experimental validation of architectural properties" presented at the IEEE WCNC 2018 conference.
- Read chapter 6 of the final report of ARCFIRE's experimentation activities, pages 119-147.
- Read the paper "On supporting Mobility and Multi-Homing in Recursive InterNetwork Architectures", published at Elsevier Computer Communications (2012).
- Look at the slides below, presented at the IEEE WCNC 2018 conference.
- Try ARCFIRE experiments yourself following the instructions of IRATI tutorials 10 and 11.