Self-Organising Semantic Resource Discovery for Pervasive Systems

Pervasive context-aware computing networks call for designing algorithms for information propagation and reconfiguration that promote self-adaptation, namely, which can guarantee - at least to a probabilistic extent - certain reliability and robustness properties in spite of unpredicted changes and conditions. The possibility of formally analysing their properties is obviously an essential engineering requirement, calling for general-purpose models and tools. As proposed in recent works, several such algorithms can be modelled by the notion of "computational field": a dynamically evolving spatial data structure mapping every node of the network to a data value. Based on this idea, as a contribution toward formally verifying properties of pervasive computing systems, in this article we propose a specification language to model computational fields, and a framework based on PRISM stochastic model checker explicitly targeted at supporting temporal property verification, exploited for quantitative analysis of systems running on networks composed of hundreds of nodes.