Amazones

Context

With the emergence of ambient computing and network of communicating objects, a new class of software has emerged. Human activities are more and more relying on digital information, acquired, stored, processed and delivered using a growing variety of smart devices in more and more intelligent environments. This ambient intelligent environment is made of a collection of heterogeneous hardware, from very specific communicating sensor nodes up to Personal Digital Assistant. Most of the applications that are deployed in such environments do not meet strong requirements as the ones found in critical embedded systems. These applications are deployed and executed in rather constrained environment with reduced CPU and memory and intermittent communication capabilities. In ambient computing, resource are managed to guarantee an average behavior, with specific constraints coming from the embedded world. We think that ambient systems are typical best-effort systems that need to be designed like critical ones. This problem raises a new scientific ground where new models and techniques must be found to bridge the two software design approaches that address those specific environments. The AMAZONES project deals with challenges in the context of software engineering for this new class of resource constrained communicating hardware in ambient environment. The aim is to study new models and design tools to leverage development of applications with trade-offs between dynamicity and predictable behavior.

Scientific foundation

Ambient applications suppose that computing processes and services are available at any time, anywhere. They exploit surrounding context and provide user-friendly applications. Services in ambient environment are running in very heterogeneous environment, in terms of hardware platforms, operating systems and communicating patterns. They need to inherit properties from the embedded world but achieve best-effort behaviours as large scale distributed applications. Designing such applications is still an issue, and most of the time, pure ad-hoc approaches are used. One challenge of designing services for ambient environments comes from the convergence of these two visions of services design: generic services dynamically adaptable and embeddable services for platforms with scarce resources and dependability constraints. Therefore, the need for middleware platform simplifying application design and integration becomes obvious. The middleware acts as glue between and within the hardware layers and the service layers. AMAZONES objective is to model and design middleware architectures as a convergence layer between embedded approaches and dynamic services. More precisely, we focus our activity on two frontiers. The south-bound which deals with a smart coupling between operating systems and hardware devices and the north-bound which aims at designing better framework support for ambient world. Finally, we think that efforts have to be made on the ground of this middleware layer by coupling south-bound issues and north-bound issues. This coupling may be very strong and static, when the middleware is very thin on very small devices and more dynamic when the middleware includes a virtual machine deployed on more powerful devices.

For more information

A more detailed presentation of the AMAZONES team can be found at this address: http://amazones.gforge.inria.fr/.

People

Head

Faculty

PhD students

Trainees