Distributed Defeasible Contextual Reasoning in Ambient Computing

The study of ambient computing environments and pervasive computing systems has introduced new research challenges in the field of Distributed Artificial Intelligence. The imperfect nature of context, the different viewpoints from which the ambient agents face the available context, and their heterogeneity with respect to the language and inference system that they use cannot be efficiently handled by the classical centralized reasoning approaches followed by most of the systems presented so far. The current paper proposes a distributed reasoning approach from the field of Multi-Context Systems (MCS) that handles these requirements by modeling ambient agents as peers in a P2P system, local context knowledge as local rule theories, and mapping rules through which an ambient agent imports context knowledge from other ambient agents as defeasible rules. To resolve potential inconsistencies that may derive from the interaction of context theories through the mappings, it uses a preference relation, which may express the trust that an agent has in the knowledge imported by other ambient agents. The paper also describes a specific distributed algorithm for query evaluation in the proposed MCS framework, analyzes its formal properties, and demonstrates its use in three use case scenarios from the Ambient Intelligence domain.