Toward a Resource Pooling Fairness: Balancing Congestion in Practical Congestion Control for Named-Data Networking

Resource pooling is an important fairness concept for multi-path congestion control. Practical congestion control (PCON) in Named-Data networking (NDN) is common congestion control in NDN. Previously, we showed that PCON cannot fully achieve resource pooling fairness due to the lack of design policy. In this paper, to improve the fairness of PCON, we introduce our proposed AQM enhancement for PCON and newly propose a suppression method to excessive window reduction caused by simultaneous congestions on multiple paths. Furthermore, we propose a flow-aware traffic steering technique for balancing congestion inspired by a MPTCP's design policy. The combination of these techniques dramatically improves fairness among consumers and quickly brings us achieving closer to the complete resource pooling state. Specifically, the overall throughput among users was improved, and then fairness index became quite high, i.e. 0.998.