Liekun Hu, East China Normal University; Changlong Li, East China Normal University, Jianghuai Advance Technology Center, and MoE Engineering Research Center of Hardware/Software Co-Design Technology and Application
The rapid proliferation of streaming media applications has driven the need for multipath transport on mobile devices. While multipath techniques successfully improve throughput by exploiting multiple network interfaces, our study reveals that path instability leads to excessive end-to-end latency. This paper analyzes the data path of multipath networks and observes that the high latency is always caused by the "last mile" wireless link, instead of the core network. Additionally, unlike traditional scenarios, both reliable and unreliable data are transmitted across these paths. However, existing multipath schedulers did not fully account for the reliability characteristics in the design. To address this gap, this paper proposes STORM, a novel multipath scheduler that aims to ensure low latency under unstable mobile networks.
We integrate STORM with the mobile device's wireless modules (e.g., WiFi and 5G). STORM differentiates between reliable and unreliable traffic. This approach prevents retransmissions from hindering critical data flows. Our evaluation on real devices shows that STORM reduces tail packet delay by 98.2% and improves the frame rate of streaming media by 1.95x under unstable networks, compared to the state-of-the-art.
USENIX ATC '25 Open Access Sponsored by
King Abdullah University of Science and Technology (KAUST)
Open Access Media
USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.
