Network coding network design study considering coding nodes and dormant nodes

Abstract

As a wireless mesh network becomes one of the major architecture for a wireless network, it intrigues a research question for a higher throughput of the wireless mesh network. The network coding is one of the candidates which can satisfy the requirement for a higher throughput. However, the computational complexity at a coding node is too high to deploy a network coding function to all intermediate nodes in a multi-hop network. It is proved that coding at several nodes can achieve the maximum throughput. However, finding the number of minimal coding node is a NP-hard problem. In this thesis, heuristic algorithm is proposed which selects the placement of small number of network coding nodes achieving given throughput. The concept of a dormant node is introduced and the worst case throughput at each stage of the algorithm is estimated by using the average number of arrival packet at each node. Then the selection algorithm is proposed which selects a node as network coding node or dormant node which can maximize the worst case throughput. The proposed algorithm unblocks a throughput bottleneck of a network by changing a bottleneck node to a network coding node or dormant node, where the dormant node of a flow is a node which discards all packets for the flow. In our case studies, it shows that unblocking bottleneck can increases throughput. By introducing a dormant node, the given throughput is achieved with the small number of network coding nodes, even with the small number of store-and-forward nodes.