9. Conclusions
Previous work on ICN mostly magnifies one specific aspect of the paradigm at a time, like in-network caching or adaptive forwarding, while devoting less attention to the novel communication model as a whole. Overall end-user experience and networkwide resource utilization are often left out of the picture. Also, one may argue that the limited benefits of a single building block taken apart, like in-network caching, can be approached by off-the shelf solutions, so missing the bigger potential of the ICN idea head-totoe. In this paper, we tackle the problem of a joint multipath congestion control and request forwarding in ICN and design scalable, fully distributed and inter-working mechanisms for efficient pointto-multipoint content delivery. More precisely, we formulate and solve a global optimization problem with the twofold objective of maximizing end-user throughput, while minimizing network cost. Our framework explicitly accounts for the interplay between multipath flow control, in-path caching and in-network request scheduling. As a result, we derive: (i) a family of receiver-driven multipath controllers, leveraging a unique congestion window and per-route delay monitoring. (ii) A set of optimal dynamic request forwarding strategies based on the number of pending requests which is the base for the design of a fully distributed algorithm run at ICN nodes. Note that, beyond ICN, the problem of multipath congestion control with a priori unknown and variable sources (due to in-path caching) has never been considered before.
