The phenomenal growth and continuous innovation of mobile and wireless systems
has triggered the development of newer and more sophisticated mobile devices,
accompanied by an even greater growth of demanding applications with stringent
requirements for Quality-of-Service (QoS), performance, security, or
robustness. The situations where wireless systems dominate is endless;
consider, as an example, the ongoing research in sensor networks, mobile
ad-hoc computing, portable video game systems and video/audio players, or
robot teams.
Although modern mobile devices quickly grow in their capabilities, this growth
seems to be outpaced by the needs of sophisticated applications, such as video
streaming or multi-peer video conferences. Particularly, mobile applications
frequently lack the resources to perform a task with the desired QoS or
performance levels.
However, the ubiquitous computing problem addressed here goes beyond file
sharing, i.e., both resources and information (both referred to as services)
have to be discoverable and accessible, in a controlled and predictable
fashion. Nomadic users must be made aware of peers in their proximity, be able
to identify these peers' sharing capabilities, and be able to robustly and
efficiently access their services.
Currently, most mobile devices/users operate in isolation from one another,
i.e., they are unaware of the presence of devices in their proximity. At the
same time, there are numerous situations when proximity-awareness (i.e., a
device is able to spontaneously share information or resources with peers in
its proximity) could benefit a mobile user, e.g., when the own resources are
exhausted or information is lacking to perform a task with sufficient service
qualities (or at all).
The goal of the SPIRIT infrastructure (Figure right)
will therefore be to provide such a sharing environment. This combines goals
and problems found in areas such as Grid computing (resource sharing),
peer-to-peer computing (information discovery and access), and mobile computing
(nomadicity of users). However, this work complements research done in
these domains by addressing the systems challenges service sharing encounters
in highly dynamic mobile environments, particularly robustness and energy
constraints.
While the sharing philosophy and architecture proposed here are
applicable to both wired and wireless worlds, the focus of this work is
on wireless situations only, because of their increasing importance and
challenging constraints. Further, many sharing scenarios are only meaningful
in wireless domains, e.g., if a user wishes to estimate its own location by
requesting another user's GPS data, both have to be in physical proximity to
each other for the exchanged data to be useful.