Reasoning about Context in Ambient Intelligence Environments

1. Contextual Defeasible Logic The study of ambient computing (AmI) environments and pervasive computing systems has introduced new research challenges in the field of KR. These are mainly caused by the imperfect nature of context information, and the need to provide distributed reasoning capabilities. [4] characterizes four types of imperfect context information: unknown, ambiguous (inconsistent), imprecise, and erroneous. These imperfections may be caused by hardware, communication and sensor failures, and the need to integrate information from various sources. So far, most ambient computing frameworks have followed fully centralized approaches, while others have used blackboard and shared memory paradigms. Collecting the reasoning tasks in a central entity certainly has advantages in terms of control and coordination between. However, such solutions cannot meet the demanding requirements of ambient environments. The dynamics of the network and the unreliable and restricted (by the range of the transmitters) wireless communications call for fully distributed solutions. In previous work, the authors have presented a novel approach to reasoning about context in ambient intelligence environments, called Contextual Defeasible Logic (CDL) [1]. They adopted ideas of and the MultiContext Systems [3], which consist of a set of contexts and a set of inference rules (known as mapping or bridge rules) that enable information flow between different contexts. These were extended by local nonmonotonic (defeasible) theories, defeasible bridge rules that query other contexts, and the use of trust information about the reliability of information sources [1]. Contextual reasoning proceeds roughly as follows: when a peer P processes a query q, it may query through bridge rules other peers, which in turn may pass on queries to further peers. Based on the information collected, P builds a support set and a blocking set for the query q; these sets contain information about the peers from which (supporting or attacking) information was received. These are compared to each other, based on the trust P places to other peers, and a positive or negative conclusion is drawn. After these conceptual and formal works were completed, the authors moved on to realize the vision described in those works by implementing CDL on a number of devices, including small devices such as mobile phones, and using these implementations to develop and evaluate sample application scenarios in real, not simulated AmI environments. The aim of this paper is to report on the initial findings of this practical work.