Network Coding can be used in applications like Peer to Peer (P2P) Networks, Wireless Ad Hoc and others. Moreover, the related theory and application research contributes to Informatics, Coding Theory, Complexity Theory, Graph theory, Matrix Theory and other subjects.
Application Layer multicast is an alternative method to the Network Layer Multicast. While the information in network layer multicast is forwarded by the router, the information in application layer multicast is forward by the terminal host (PC or server). As the host usually has powerful computational resources, it is a compatible environment for network coding.
One typical application is P2P file sharing software Avalanche developed by Microsoft [Gka04]. The file is divided into n blocks and encoded with Random Linear Coding at every host. As the linear coefficients are added to the encoded blocks, one host could decode the entire file once sufficient coded blocks are received. Meanwhile, the possibility of complete delivery is increased despite the joining or leaving of hosts dynamically. In fact, Avalanche applies network coding on a unique time-parameterized graph as shown in Figure 4., rather than the physical network [Yeu08]. The variable t denotes the unit time and the link stands for the delivery of packets between hosts.
Time-Parameterized Graph of Avlanche type system
Due to the unreliability and multicasting feature in the physical layer of wireless network, network coding could resolve the issue in the traditional routing and cross-layer design. Applied network coding in the wireless network could increase the multicast throughput, reduce the number of hops and decrease the energy required for emission. In particular, with Random Network Coding, the original data could be retrieved at the terminal nodes even if some of the intermediate nodes or links are disabled.
One throughput optimization framework is proposed for multi-hop networks across network and physical layers. The network coding scheme provides for data routing and wireless power allocation[Yua05]. Another experimental on 802.11 hardware shows that one XOR-only mechanism nearly doubles the network throughput [Kat05].
Another reduced complexity network coding is proposed for ad hoc networks. The links are divided into two types: (a) entering relay nodes and (b) entering targets. The same capacity can be achieved by applying network coding only on the type (a) nodes and keeping the traditional routing style at all the type (b) nodes [WuY05].
The minimum energy per bit can be achieved by network coding for mobile and ad hoc networks with linear program [WuY04]. Finally, a simple distributed method is proposed for exchange independent information in two wireless nodes using only XOR for network coding [WuY05].
Share with your friends: |