Multi-interface Communication: Interface Selection Under Statistical Performance Constraints

Abstract

Recent advancements of multipath communications have made a new dimension of resource management accessible focusing on balanced and efficient utilization of available communication alternatives. Dynamic strategies in this context leverage the most up to date information on available interfaces to outperform comparable static versions. In this work, we propose an adaptive strategy that seeks to improve the worst-case performance of a multipath communication system by periodically looking at waiting time bounds of respective subsystems. We compare the performance of the proposed adaptive assignment with established round-robin and join-the-shortest-queue strategies numerically where we observe that our strategy tends to be superior in heterogeneous environments.