Pervasive computing deals with heterogeneous mobile agents attached to ubiquitous micro-devices. In such scenarios, what one agent knows about the environment is based on perception components it uses or has access to, and it can be significantly different from another agent's knowledge. Furthermore, transient conditions and uncertainty affects perceptions and communication, aggravating the need to cope with the lack of complete and reliable information. Current solutions in the Internet of Things (IoT) are mostly based on centralized data collection and analysis and on top-down agent orchestration, with obvious limitations in latency, connection availability and data confidentiality. This thesis proposes a novel distributed knowledge-based framework named object (b)logging to tackle the above issues. The approach is conceived as a general-purpose evolution of the IoT, able to associate semantic annotations to real-world objects and events as well as to trigger complex objects choreography through advanced resource discovery. It envisions several smart entities organized in social networks, interacting autonomously and sharing information, cooperating and orchestrating resources through a published micro-blog. Ontology-referred context annotations produced and shared by individual smart objects in mobile ad-hoc networks are merged by means of novel Concept Fusion and enhanced Concept Integration reasoning services in Description Logics, specifically devised for context-aware multi-agent systems and tailored to resource-constrained devices. Management of incomplete information, reconciliation of inconsistencies in context descriptions, quick adaptation to changes and robustness against spurious or inaccurate information allow to progressively enrich a node's core knowledge in a private micro-log. Then it becomes able to identify on-the-fly the task(s) needed to change its own configuration or the environment state and automatically infer what useful capabilities it can provide to or needs from other entities in order to enact them, in a decentralized and collaborative fashion. A novel semantic-enhanced blockchain infrastructure underlies the dissemination, discovery and selection of services and resources. These tasks have been revisited as smart contracts with opportunistic and distributed execution, exploiting validation by consensus. The introduced paradigm ideally applies to pervasive cyber-physical systems, where several mobile heterogeneous micro-devices cooperate to connote and modify appropriately the environment they are dipped in, as demonstrated by relevant case studies and extensive experimental evaluations.
|Titolo:||Object (B)logging: a Semantic-Based Self-Description for Cyber-Physical Systems|
|Data di pubblicazione:||2020|
|Appare nelle tipologie:||5.14 Tesi di dottorato|