Performance simulation of network coding-based on-demand broadcast models

On-demand broadcast can satisfy multiple requests for the same data item, and improve the system performance. Network coding-based on-demand broadcast further helps save bandwidth, and further improves performance. In this research, a generalized encoding framework which incorporates network coding into data scheduling algorithms for on-demand broadcast is used to study the effect of network coding. In this framework data scheduling is formulated as a weighted maximum clique problem in a graph where the objective function of the clique is defined according to the performance objectives of the applications. The coding opportunity of the proposed framework is measured by the expected number of clients that can decode a data item out of an encoded broadcast data item. We set up a simulation environment and conduct experiments to evaluate the performance of broadcast with coding, based on average access time, average stretch and deadline miss ratio. The results show network coding lead to significant performance improvements for all the algorithms. This effect is particularly significant when the cache size is large or when the data access pattern is more skewed.