On the Potential of Modern TCP Congestion Control Algorithms in Information-Centric Networking

Information-Centric Networking (ICN) is a network architecture that prioritizes the information being transferred over the hosts transferring the information, as in conventional TCP/IP networks. Although ICN has gained significant attention in recent years, transport design, especially congestion control, in ICN is still in its early stages of development. ICN adopts a communication paradigm that is vastly different from that of conventional TCP/IP. While a number of TCP congestion control algorithms have been considered as transport protocols for ICN, the impact of ICN caching on these algorithms is still unknown. Currently, only AIMD-type congestion control algorithms have been investigated. In this study, we implement various fluid models of TCP congestion control algorithms (namely CUBIC, CTCP, HSTCP, Reno, Veno, and Illinois) used in TCP/IP networks, in the flow-level ICN simulator FICNSIM. We comparatively investigate the performance of modern TCP congestion control algorithms when used as congestion control algorithms for ICN transport protocols through numerical simulations. Additionally, we quantitatively demonstrate how the existence of content caching at intermediate routers affects the transient and steady-state performance of ICN transport protocols with different congestion control algorithms.