Sociality and Self-Organisation in Next-Generation Distributed Environments

The proliferation of computationally powerful, interconnected devices entails a new generation of networked applications and social utilities characterized by a strong growth in scale and dynamics.<br />Distributed virtual environments constitute a privileged example, involving a high degree of interactivity as well as tightened constraints and requirements. As a response to these issues, this dissertation explores and substantiates sociality as a fundamental principle both in and for the design of such systems.<br />A specialized, dual peer-to-peer architecture is introduced, combining a highly-structured backbone overlay with a loosely-structured geometric client overlay synergistically complementing each other. To enable a global-scale, single-instanced environment, it is imperative to include as many client-side resources as possible and unburden the backbone. The focus of this dissertation therefore lies upon the latter, geometric overlay.<br />By taking an interdisciplinary perspective and leveraging different aspects of sociality, a series of self-organized approaches addressing major problem areas are proposed: a collaborative filtering mechanism for the handling of information overload created from the soaring amounts of users and objects; a confidentiality framework for the protection of sensitive data more likely exposed due to an increased interactivity; and two resource allocation schemes for fairly distributing surplus capacities in the face of critical regional surges.<br />Detailed evaluations show that these decentralized algorithms operate robustly and effectively, while yielding well-converging results in comparison to optimal, global-knowledge scenarios.