UbiRep

There is a clear need for data sharing and collaboration support in a large number of computer applications in different domains. In this project, we focus on applications in the area of co-operative work that is carried out by people without access to a traditional wired fixed network. For example, a group of people outside their offices such as in the hotel lobby, while at the airport, in a building construction site, in the assembly line of a factory, etc. Such a group is highly dynamic, i.e. people get into and leave the group either at their own will or due to some communication problem. The networks thus formed are frequently called ad-hoc networks. A scenario of such co-operative work occurs, for example, in the manufacturing and maintenance of airplanes: a group of technicians perform a set of tasks that imply some co-ordination among them. This requires sharing of information among such people, each one using an info-appliance (portable, PDA, tablet, etc.), in an environment in which there is no wired fixed network. This kind of scenario occurs also when a group of engineers visit a bridge or a building still under construction.

By supporting data replication, the overall goal of this project is to facilitate the development and execution of distributed applications, running on top of info-appliances (such as portable computers and PDAs) that support data sharing and collaborative work in ad-hoc networks. Data replication improves not only performance but also data availability; it is very fast to access local data (i.e. co-located with the corresponding code running in some info-appliance) and, even when an info-appliance is isolated with no communication path to other equipment, the data replicated in its memory is still available.

To achieve the goal mentioned above, system-specific issues have to be solved. In particular, the memory management of replicas, their consistency and the resolution of conflicts, must be handled automatically (as much as possible) releasing the programmers of such tasks. Systems supporting replication do not consider such functionality in ad-hoc networks as most of them were first designed for stationary environments (with access to wired fixed network). Systems such as Bayou, Roam, etc., while providing very good ideas, either do not address the specific system issues we intend to work on, or do not evaluate the solutions we plan to explore.

Thus, the contribution of UbiRep is twofold. Concerning the memory management of replicas, we will study the garbage collection of objects in ad-hoc networks, with emphasis on distributed cycles. W.r.t. the consistency of replicas and the resolution of conflicts we intend to explore the notion of future-aware computing: using environmental information based on simple reasoning and forecasting derived from gathered environment properties (e.g. the users' calendar, policy specifications for replicas usage, etc.) so that the system is capable of anticipating future user behavior. For example, applications may advice the user (or even decide automatically) to create new replicas or merge them, thus preventing or solving conflicts.

Sponsoring bodies: FCT

Coordinator: Paulo Ferreira

Partners: INESC-ID

Homepage: N/A