Mobility as a Public Network Service
Posted: 3 Apr 2013
Date Written: March 31, 2013
Enabling mobility in the Internet has the potential to be more important, transformative, and disruptive than the transition from narrowband to broadband. Historically, mobility in the Internet has largely been relegated to treatment as an access problem. The overarching paradigm is that a user/application initiates an end-to-end session with a network and if the user moves, the connection either needs to be re-established at the new location or recourse needs to be made to a network service that supports mobility, but that service is typically controlled by/implemented by the user's access network provider via a "gateway" that addresses the challenge of mobility as a special case. Fundamentally, it assumes a fixed host-server model.
MobilityFirst, an evolving future Internet architecture, takes seriously the proposition that mobility is a fundamental capability that needs to be embedded in the core of the Internet. Support for mobility needs to be much richer than just providing support for moving end nodes; and taking mobility support seriously, also requires addressing the special challenges of wireless access. Enabling mobility as basic functionality means allowing users, applications, and networks to move in network, time, and spectrum space. This requires a fundamental rethinking of Internet architecture, and if adopted, implies profound challenges and opportunities for the Internet ecosystem.
In this paper, we explain why legacy approaches for supporting mobility in the Internet are inadequate. Using the MobilityFirst architecture to frame our discussion, we articulate what a richer notion of mobility ought to entail in practical terms (i.e., the user capabilities/scenarios that should be supported) and then map these to a set of network requirements, including support for varying wireless link quality and disconnection, multi-homing, ad hoc networking, flexible autonomous system boundaries and spectrum coordination. In short, we make the case for why a public mobility service represents basic functionality that should be incorporated in the narrow waist of the Internet.
From both technical and economic perspectives, we examine the challenges and benefits of a public mobility service through three important usage cases: • Ad hoc networks: A mobility service provides the ability to add and move mobile networks for dynamic reconfiguration. While public safety and vehicular networks are the obvious applications, ad hoc networks also have implications for carrier-less networking and user-deployed networks for specialized services. • Multihoming: The GNRS enables users and devices to obtain the diversity benefits of multiple simultaneous networks connections in a manner transparent to network applications. The technical challenge is a low latency implementation using a distributed mobility service. In addition, multihoming allows applications to flexibly control what network used for what traffic, allowing arbitrage of pricing, QoS, or other service terms. • Spectrum management: As we transition to small cell architectures, the challenge of coordinating frequencies and other network elements (distributed antenna systems, backhaul, power) creates difficult coordination problems. While these problems are traditionally solved by a mobile operator, MobilityFirst GUIDs for spectrum resources can enable alternative end-user solutions.
In principle, each of these use cases could be implemented with extensions to current IP networks using application layer protocols at the network edge in addition to protocol stack tweaks at the mobile terminals. However, we explain why enabling this as a public mobility service as anticipated in MobilityFirst offers important benefits in terms of facilitating competition
Keywords: Mobility, Internet, Architecture
Suggested Citation: Suggested Citation