Introduction

While the wired Internet reaches many homes and businesses, the wireless Internet has the potential to not just reach, but encompass all the spaces that people use to live, work, and travel. Wireless data services (e.g., 802.11b, GPRS, 3G cellular) will soon provide the potential for ubiquitous, though heterogeneous, coverage. To realize this potential, users will want both seamless connectivity (flows uninterrupted by mobility) and continuous reachability (the ability of other hosts to contact the user's host despite mobility). These services would enable users to run applications such as IP telephony, instant messaging, and audio streaming while mobile.

Unfortunately, the standard Internet cannot provide these services. The fundamental problem is that the Internet uses IP addresses to combine the notion of unique host identifier with location in the network topology. For a mobile host to have seamless connectivity and continuous reachability, it must retain its identifier while changing its location. Previous mobility proposals decouple this binding by introducing a fixed indirection point (e.g., Mobile IP [1]), redirecting through the DNS (e.g., TCP Migrate [2]), or using indirection at the link layer (e.g., cellular mobility schemes).

However, these proposals lack one or more of the following properties to fully realize the promise of ubiquitous mobility:

• Efficient routing: packets should be routed on paths with latency close to the shortest path provided by IP routing.

• Efficient handoff: the loss of packets during handoff should be minimized and avoided, if possible.

• Fault tolerance: communication between mobile hosts should not be more vulnerable to faults than communication between stationary hosts.

• Location privacy: the host's topological location should not be revealed to other end-hosts.

• Simultaneous mobility: end hosts should be able to move simultaneously without breaking an ongoing session between them.

• Personal/session mobility: a user should be able to redirect a new session or migrate an active one from one application or device to another one when a better choice becomes available [3,4,5].

• Link layer independence: users should be able to seamlessly operate across heterogeneous link layer technologies, not all of which support the same link layer mobility scheme (e.g., GSM mobility).

In this paper, we propose (to the best of our knowledge) the first solution to achieve all of these properties. Our solution, called Robust Overlay Architecture for Mobility (ROAM), is built on top of the Internet Indirection Infrastructure ($i3$) [6]. $i3$is implemented as an overlay network on top of IP, and provides a rendezvous-based communication abstraction. In $i3$, each packet is sent to an identifier. To receive a packet, a receiver inserts a trigger, which is an association between the packet's identifier and the receiver's address. The trigger is stored at an $i3$node (server). Each packet is routed through the overlay network until it reaches the $i3$server which stores the trigger. Once the matching trigger is found the packet is forwarded to the address specified by the trigger. Thus, the trigger plays the role of an indirection point that relays packets from the sender to the receiver.

ROAM addresses each of the properties described above. For instance, since an $i3$identifier can be bound to a host, session, or person (unlike Mobile IP, where an IP address can only be bound to a host), personal/session mobility applications can leverage the ROAM infrastructure for efficiency, fault tolerance, and privacy. Section 4 discusses in detail how ROAM achieves the above properties.

At the architectural level, this paper makes two contributions. First, it demonstrates the benefit of giving end-hosts control on the placement of the indirection points. This allows, end-hosts to optimize the routing and handoff efficiency. Second, it demonstrates the benefits of a mobility architecture based on a shared overlay network. Such a solution leverages the robustness of the overlay networks.

In addition, we use a proxy based solution to transparently support unmodified applications on an unmodified Linux kernel. Using our prototype implementation, we show that our solution can perform rapid soft handoffs with no noticeable disruption of TCP throughput.

The paper is organized as follows. Section 2 presents the related work, and Section 3 gives an overview of $i3$. Section 4 discusses the design of ROAM, and Section 5 presents the ROAM support for legacy applications. Section 6 presents some implementation details. Section 7 presents simulation and experimental results. Finally, Section 8 discusses some open issues, and Section 9 concludes the paper.