Load Balancing and Clustering in Large-Scale Overlays

Overla can facilitate the collaboration between subsets of nodes/hosts of a network, towards common goals (e.g. resource sharing, community structures and more) Clustering in such networks is important: it can facilitate achieving locality and scalability for services –e.g. routing, resource finding and sharing in neighborhoods---, realizing interest-based/social groups and more; clustering can extend to security issues, with trust as a parameter. When it comes to resource sharing and community service, load balancing is aligned with clustering; it forms an incentive to achieve fairness and collaboration. This project aims to design and cluster overlay networks in dynamic environments and balance the load within them. One of our basic instruments will be random walks, which have been shown to be effective in the searching, dissemination and construction of overlay networks as a statistical process that associates with important parameters and properties of the corresponding graph. The proposing teams have complementary expertise on working with large networks (self-stabilization, locality-based algorithms respectively), as well as common interests and closely related recent work on large networks. Through our collaboration, which will be enabled by mutual visits, we will be able to combine our expertise and achieve even stronger results, as well as to project these on our future plans.

Participants

Marina Papatriantafilou (contact)

Docent at Computer Science and Engineering, Networks and Systems (Chalmers)

Funding

Swedish Foundation for Strategic Research (SSF)

Funding years 2011–2013

More information

Latest update

2016-12-15