Data synchronization via random network coding

Duong Nguyen-Huu, Thinh Nguyen
2014 Proceedings of the fourth ACM international symposium on Development and analysis of intelligent vehicular networks and applications - DIVANet '14  
We investigate the problem of efficient data synchronization for a large number of nodes with disparate data. The synchronization problem arises naturally in many large scale systems, including Peer-to-Peer networks, data centers, and distributed storage systems with asynchronous updates. It is intimately related to the index coding problem in network coding, and can be described abstractly as follows. There are a number of nodes (receivers) who share a broadcast channel. Each node keeps a
more » ... h node keeps a random fraction of a given file in terms of a number of packets. In addition, there is a broadcast node who has the entire file, and can broadcast a packet in a time slot to every node. The goal is to find a broadcast scheme that minimizes the number of time slots until all the nodes receive the entire file. In this paper, we propose and analyze a number of random network coding algorithms and verify their performances via theoretical analysis and simulations.
doi:10.1145/2656346.2656361 dblp:conf/mswim/Nguyen-HuuN14 fatcat:cfstxsxgozexdhgf4m4jagfgb4