A Goal Model for Collective Adaptive Systems

A collective adaptive system is composed of large numbers of autonomous and self-adaptive entities which are able to provide benefits for other participants and have to cooperate with each other to accomplish their individual goals. When entities adapt their own behaviour due to unexpected situations, this may result in breaking the overall collaboration if these are not executed simultaneously and do not respect some collective constraints. The objective of this paper is to present a goal model that allows to express requirements for collective adaptability. It includes (i) how to specify the behaviour of the entities that encodes cooperation with other entities, (ii) how to link collaborative goals and individual goals, and (iii) how to specify constraints that are imposed on different entities to preserve some form of collective consistency while they are adapted.

I. Introduction

Collective Adaptive Systems (CASs) consist of diverse heterogeneous entities (i.e., computers, services, devices, sensors, humans, networks, etc.) that are autonomous but have to cooperate with each other to accomplish collective tasks [12]. As each entity is autonomous, it is characterized by its own behaviour. To enable a collaboration, it can also expose functionalities to the outside world. As such, while entities preserve individuality, they still can participate in a collaboration. As a rule, the environment in which such system operates, continuously changes, starting from the changes of the context in which entities live, to the availability of new entities (or their exit from the system), to changes in the system requirements and preferences. Dynamic changes might affect the operation of a CAS and they have to be properly handled at run time by adapting system configuration and behaviour of constituent entities. This continuous adaptation becomes the key feature of CASs when it comes to operating in continuously changing environment.