Context-Aware Reflective mIddleware System for Mobile Applications
People
Dr. Licia Capra
Research Problem
The increasing popularity of wireless devices, such as mobile phones and personal digital assistants, is enabling new classes of applications that present challenging problems to designers. Applications have to be aware of, and adapt to, variations in the execution context, such as fluctuating network bandwidth, decreasing battery power, changes in location or device capabilities, and so on.We argue that middleware solutions developed for wired distributed systems cannot be used in a mobile setting, as the principle of transparency that has driven their design runs counter to the new degrees of awareness imposed by mobility.
We believe the construction of adaptive and context-aware mobile applications can be supported using a reflective middleware. The middleware provides software engineers with primitives that enable dynamic customisation of service delivery. In particular, applications define associations between the services they are willing to customise, the policies that must be used in order to deliver the services, and the context conditions that must hold in order for a policy to be enabled. These associations are stored in application profiles; through reflection, they are made available to applications for both run-time inspection and modification. Middleware uses the information encoded in the profiles to dynamically adapt its behaviour to changing context conditions and applications' needs: each time a service is invoked, the profile of the application that requests it is consulted to determine which policy can be applied in the current context.
Although enabling the development of adaptive and context-aware middleware, reflection also opens the door to conflicts: different policies, associated to the same service, may be enabled at the same time, i.e., in the same context. These conflicts cannot be resolved statically at the time applications are designed, but, rather, need to be resolved at execution time. We have designed a method by which these policy conflicts can be treated. This method uses a micro-economic approach that relies on a particular type of sealed-bid auction: applications associate values to each conflicting policy, based on private information (i.e., private non-functional requirements); middleware then selects and applies the policy that maximises "social welfare'', that is, the one that delivers (on average) the best quality of service, according to current applications' needs.
Selected Publications