Towards an adaptive execution of applications in heterogeneous embedded networks

Abstract

Embedded networks are emerging in many application fields, such as the automotive or building and factory automation sector. Compared to other distributed systems, embedded networks offer a new challenge for developers: heterogeneity and resource constraints. The nodes contained in these networks can differ greatly w.r.t. their storage, processing and sensing/acting capabilities, ranging from very simple sensor devices with very limited resources over programmable logic controllers to very powerful nodes such as PCs. In order to achieve an efficient execution of applications running on such a network, a middleware is required that automatically adapts the embedded network to the requirements of the installed applications. In this paper, we will present a model driven development approach that allows the specification of application requirements, and a corresponding middleware solution that supports the automatic adaptation of the application execution based on these requirements and the characteristics of the underlying hardware.

Self-awareness in autonomous automotive systems

“…Within the RACE project [16], a flexible software and partially redundant hardware architecture is proposed as a solution to the challenge of creating highly available and dependable automated vehicle systems. However, detection of sensor failures is limited to performing a set of boundary checks for the respective sensors.…”

RACE: A Centralized Platform Computer Based Architecture for Automotive Applications

“…Part of the deployment is the location of the applications on the platform as well as the interconnection of components, the so called wiring [12] or logical routing. The wiring is a matching based on the specified topics and attributes of the component ports.…”