Use this URL to cite or link to this record in EThOS:
Title: Supporting management interaction and composition of self-managed cells
Author: Schaeffer Filho, Alberto E.
ISNI:       0000 0004 2678 9759
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2009
Availability of Full Text:
Access from EThOS:
Access from Institution:
Management in ubiquitous systems cannot rely on human intervention or centralised decision-making functions because systems are complex and devices are inherently mobile and cannot refer to centralised management applications for reconfiguration and adaptation directives. Management must be devolved, based on local decision-making and feedback control-loops embedded in autonomous components. Previous work has introduced a Self-Managed Cell (SMC) as an infrastructure for building ubiquitous applications. An SMC consists of a set of hardware and software components that implement a policy-driven feedback control-loop. This allows SMCs to adapt continually to changes in their environment or in their usage requirements. Typical applications include body-area networks for healthcare monitoring, and communities of unmanned autonomous vehicles (UAVs) for surveillance and reconnaissance operations. Ubiquitous applications are typically formed from multiple interacting autonomous components, which establish peer-to-peer collaborations, federate and compose into larger structures. Components must interact to distribute management tasks and to enforce communication strategies. This thesis presents an integrated framework which supports the design and the rapid establishment of policy-based SMC interactions by systematically composing simpler abstractions as building elements of a more complex collaboration. Policy-based interactions are realised – subject to an extensible set of security functions – through the exchanges of interfaces, policies and events, and our framework was designed to support the specification, instantiation and reuse of patterns of interaction that prescribe the manner in which these exchanges are achieved. We have defined a library of patterns that provide reusable abstractions for the structure, task-allocation and communication aspects of an interaction, which can be individually combined for building larger policy-based systems in a methodical manner. We have specified a formal model to ensure the rigorous verification of SMC interactions before policies are deployed in physical devices. A prototype has been implemented that demonstrates the practical feasibility of our framework in constrained resources.
Supervisor: Lupu, Emil Sponsor: Engineering and Physical Sciences Research Council ; International Technology Alliance ; EC IST EMANICS Network of Excellence
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral