Publication details

Self-organizing Collaborative Environments

Investor logo


Year of publication 2010
MU Faculty or unit

Faculty of Informatics

Description The main goal of this thesis is to study ad-hoc collaborative environments and present benefits which can be brought by employment of self-organization principles into the area. We show that manual orchestration of collaborative environments is not only extremely cumbersome but also fails to address especially the needs for robustness and resiliency. Therefore we study especially possibilities of automated orchestration of component based collaborative environments. Since we consider collaborative environments featuring uncompressed or low-compression media with high-bandwidth demands we also study possibilities of automated network infrastructures configuration and orchestration for the specific purposes of the collaborative environments. In this context we focus on advanced network infrastructures providing lambda-services and propose benefits of their integration with collaborative environments. Based on our findings we propose a theoretical framework to address these issues. As our work is strongly motivated by the needs of real-world collaborative environments, we present a novel collaborative environments middleware named CoUniverse and its architecture which is based on above mentioned framework to demonstrate its features and benefits. Latency-sensitive data distribution and media streaming is a part of many network-centric applications. When bandwidth demands of media streams are comparable to capacity of network links, the implicit additive assumption no longer holds and the system needs to feature sophisticated planning of the media streams transmissions over individual network links. Based on our previous experience with interactive high-quality media distribution and orchestration of collaborative environments, we are proposing an automated planning of media streams distribution in environments where bandwidth of streams is comparable to network link capacities. We have successfully adopted methods of constraint programming to deal with this problem. We provide an algebraic description of the media streams planning problem and show its complexity. Based on the algebraic description of the problem we propose a comprehensive constraint model. To formulate the model, a~new variant of the link based model for traffic placement problem is applied. An~evaluation of the implemented solver is presented on a number of experimental problems to show effectiveness of the formulation and proposed constraints. In this thesis we also address the problem of automated user-empowered control of dynamic network infrastructures to satisfy the demands of orchestrated collaborative environments. To do so we take advantage of the advanced features of Dynamic Circuit Networks (DCN) and proposed framework providing self-organization to orchestrated collaborative environments. We show augmentation of the framework capabilities by interfacing with DCN enabled network infrastructures in order to automatically control the lambda-services allocation, deallocation and modifications based on collaborative environments and application needs. We have carried out a proof of the concept implementation of the CoUniverse middleware to verify the above mentioned concepts. We provide an evaluation of the most important concepts of the CoUniverse middleware with respect to its ability to establish and provide self-organization features to the collaborative environments. We also evaluate the scalability of the CoUniverse middleware analysing the size of the collaborative environments it is able to orchestrate taking into the account the interactive nature of the collaborative environments in question. We have already used the prototype implementation of the CoUniverse middleware to orchestrate various types of collaborative environments and applications in a~number of demonstrations and events. Most of these demonstrations we also discuss in this thesis featured CoUniverse deployment on a network infrastructures spanning USA and Europe.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.

More info