Bandwidth-Delay-Product-Based ACK Optimization Strategy for QUIC in Wi-Fi Networks

Abstract

QUIC has drawn extensive attention in supporting low latency and secure Internet of Things (IoT) communications due to its efficient handshake and default end-to-end encryption. However, in Wi-Fi enabled IoT communications with contentions for shared media, QUIC’s inherent acknowledgment (ACK) policy may induce non-negligible control overhead and limited data throughput. To address the problem, this paper designs and implements an ACK frequency optimization scheme for QUIC by exploiting the tailored bandwidth-delay product (BDP) at the receiver, named QUIC-BDP. To accurately estimate real-time BDP, we design an “ACK-PING" strategy to compensate for the accuracy of round-trip timing estimation and utilize exponential averaging and sliding window filtering for stable bandwidth estimation. Experiments results show that our proposed QUIC-BDP balances between the robustness and throughput performance while maintaining stable performance in lossy cases, with a reduced energy cost. Particularly, QUIC-BDP achieves up to a 67% gain in goodput compared to the original QUIC, and it improves goodput by up to 38% and 28% compared to existing solutions MSQUIC and QUIC-1:10, respectively. In addition, QUIC-BDP reduces energy cost by up to 50% compared to the original QUIC.