Modified Index Coding Increases Capacity of Information-Centric Networks
Mohsen Karimzadeh Kiskani, Hamid R. Sadjadpour · arXiv (Cornell University) · 2014
The Index Coding (IC) problem studies transmission policies when the source node broadcasts encoded data to users with side information. This paper extends the index coding problem to the case when the source node can reach users through multi-hop communications. The new approach is called Modified Index Coding (MIC) which can be applied to both wireless and wired networks. We demonstrate the benefits of our approach by applying MIC to Information-Centric Networks (ICN). We demonstrate that the combination of ICN and MIC requires a hybrid caching scheme that includes caching at the routers and the nodes to support two different goals. The caching is used at the routers to place contents closer to the users when requested (similar to ICN approach) and is used at the users as side information. Our analysis demonstrates that order optimal capacity gains can be achieved in some capacity regions, i.e., capacity gain grows with the number of nodes in the network by using MIC. We have computed the capacity gains by counting either the number of disjoint cycles or cliques in the dependency graph. It was also shown analytically that near edge caching by routers in ICN achieves close to optimal capacity. Simulation results compare the capacity improvement to the traditional ICN architecture.