Optimal Media Communication in Capacity Constrained Networks

For a network with links of finite capacity, we formulate the problem of optimal multimedia communication in its multicast settings. The goal is to find source coding and network communication schemes to communicate a real valued source to a number of sinks to realize the best reconstruction quality at all the sink nodes. We find that the possible reconstruction distortions at the sinks are entangled in a highly non-trivial way. An intriguing problem is that of finding the set of achievable distortions at the sink nodes. Equivalently, we formulate the problem as that of finding a coding scheme that minimizes a weighted sum of the distortions. We realize that separate source coding and network communication approaches are sub-optimal in general. We then devise a powerful strategy based on optimal routing of balanced multiple description codes, for which we develop polynomial time sequential and iterative algorithms. Finally, the improvement in the overall quality of source reconstruction by using the proposed schemes is verified through simulations.