Session Management Algorithms for Solving K-Token Dissemination Using Network Coding

Examensarbete för masterexamen

Please use this identifier to cite or link to this item: https://hdl.handle.net/20.500.12380/193794
Download file(s):
File Description SizeFormat 
193794.pdfFulltext771.77 kBAdobe PDFView/Open
Type: Examensarbete för masterexamen
Master Thesis
Title: Session Management Algorithms for Solving K-Token Dissemination Using Network Coding
Authors: GARCÍA, GUILLERMO BARREDO
Abstract: In this thesis report we consider k-token dissemination algorithms that makes use of network coding. The k-token dissemination problem consists of propagating a total number of k tokens to all nodes in the network. The tokens are distributed between one or more nodes before the algorithm is executed, and the final goal is that all nodes must eventually have the same set of k-tokens. Network coding is a recent technique that, implemented in a proper way, helps to save bandwidth and improves the speed of distributed computation. The network model consists of a network that can change completely from round to round, therefore the nodes do not know anything about their neighbours. Moreover, when a node broadcasts a message, it does not know which are the receivers, thus, it is not possible for the nodes to know which are the tokens that their neighbours need. By using network coding, the time needed to achieve the final goal (all nodes possess the same k tokens) is drastically reduced in comparison to a simple random forwarding algorithm, which, as its name infers, randomly broadcasts the tokens possessed by a node. We study the session management problem, and present two algorithms to solve it. The problem consists of limiting the total number of sessions that concurrently coexist in the system. In the context of dissemination problems and network coding, we consider a session as an index according to which the information is codded by the session initiator. Since in dynamic networks nodes can crash and recover, we wish to allow sessions to accomplish their tasks, while limiting the amount of overall system resources in use. We propose solutions for the session management problem, and by that facilitate the solution of the k-token dissemination problem in dynamic networks. By solving the session management problem, we can also solve the k-token dissemination problem in a more robust way, in the sense that the algorithms can deal with several types of failures, such as, crashes and crashes-recoveries.
Keywords: Datavetenskap (datalogi);Informations- och kommunikationsteknik;Computer Science;Information & Communication Technology
Issue Date: 2014
Publisher: Chalmers tekniska högskola / Institutionen för data- och informationsteknik (Chalmers)
Chalmers University of Technology / Department of Computer Science and Engineering (Chalmers)
URI: https://hdl.handle.net/20.500.12380/193794
Collection:Examensarbeten för masterexamen // Master Theses



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.