Preliminary definition of CORTEX interaction model

Abstract

As scheduled in the Technical Annex, WP2-D3 comprises work on the basic communication abstractions and the context and environmental awareness. It is structured in an introduction, providing a short survey of the content and four technical chapters. Chapter 2 describes the notion of event channels as a basic middleware abstraction of the interaction model. The concept of event channels accommodates an event-based, generative, many-to-many, anonymous communication model. It contributes to the resolution of the trade-off between autonomy and the need of coordination. Rather than explicitly coordinating actions by transferring control, an event channel allows interaction via a shared data space, thereby maintaining the autonomy of components. A comparison with alternative schemes is presented in chapter 3. Here, the impact of the interaction scheme on the modelling and implementation of a complex robotic application is analysed. It provides additional arguments in favour of a publisher/subscriber communication architecture. One of the challenges in CORTEX is to integrate the cooperation of components through the environment into the general interaction concept. The sensor capabilities of the sentient components and their ability to interact with the environment open new ways of cooperation. A mechanism called Stigmergy which is borrowed from biology and discussed in the CORTEX context is presented in chapter 4. Any activity which is carried out in the physical world needs to adapt to the pace and dependability requirements dictated by the environment. In technical terms this means that non-functional properties of the system, as timeliness and reliability of operation have to be included. These Quality of Service (QoS) attributes have to be guaranteed even in an environment where unanticipated dynamic change is one of the inherent properties. Chapter 5 introduces an adaptive QoS mechanism based on a reliable and timely system service. This service, called the Timely Computing Base (TCB) is able to monitor distributed system activities and to provide an "early warning system" for temporal and functional failures. The TCB thus provides part of the context and environmental awareness needed for adaptation