Dynamic Window Coupling for multipath congestion control

Abstract

The traditional problem of end-hosts efficiently and fairly utilizing end-to-end paths becomes significantly harder when the end-hosts are multihomed. Such is the case, for instance, when an end-host has simultaneous connectivity through several service providers, or when a mobile device is simultaneously connected via both a wireless LAN and a cellular network. A multihoming-aware transport protocol, such as MPTCP or SCTP, that sends data over the multiple resulting end-to-end paths must be fair to other flows in the network while being able to maximize its own throughput. In this paper, we present Dynamic Window Coupling (DWC), a multipath congestion control mechanism that seeks to achieve both these goals. DWC uses loss and delay signals to detect shared bottlenecks, explicitly grouping and sharing congestion control across subflows on paths that have a common bottleneck, while separating congestion control for subflows on paths with distinct bottlenecks. DWC detects shifting bottlenecks in the network and responds by dynamically regrouping subflows. Simulations demonstrate that DWC detects shared bottlenecks under most network topologies and conditions that we considered, regroups subflows correctly as bottlenecks shift, aggregates throughput across distinct bottlenecks, and is fair to other TCP flows at all bottlenecks.

Is multi-path transport suitable for latency sensitive traffic?

“…Each UDP flow generates Pareto onoff traffic with shape 1.5 and scale 0.166667, sending 25 frames per second each of size 5,000 bytes. The aggregate usage of UDP background flows were maintained at 10% of the bottleneck link capacity to be realistic [65]. The UDP flows carry data at an average of 500 Kbit/s each in the WLAN-WLAN scenario and 100 Kbit/s each in the 3G-3G scenario.…”

Enhancing fairness and congestion control in multipath TCP

“…This can result in sub-optimal performance of MPTCP, as will be shown in section IV-B. Dynamic Window Coupling (DWC) [12] is proposed to address this issue and improve the aggregate throughput of an MPTCP connection by detecting distinct bottlenecks and only coupling those MPTCP flows which share a common bottleneck.…”

“…This equal share outcome of NewReno will result in an unfair share of the bandwidth if more than one TCP flow is active for a single MPTCP connection within the bottleneck link, violating the design goal "Do No Harm". Therefore, alternate congestion control algorithms have been proposed for the MPTCP protocol, such as, Fully Coupled congestion control [9], Linked Increases Algorithm (LIA) [9] or also referred to as (semi-) coupled congestion control [10] [11], Dynamic Window Coupling (DWC) [12] and recently Opportunistic Linked Increases Algorithm (OLIA) [13]. All these algorithms do not modify the Slow Start, Fast Retransmit and Fast Recovery phase of TCP NewReno but only the Congestion Avoidance phase.…”

Improvement of SCTP congestion control in the LTE-A network

“…Duke et al (2003) applied SCTP over a land mobile satellite channel and employed TCP in the same environment. Subsequently, Hassayoun et al (2011) introduced dynamic window coupling, which used performance factors such as delay and packet loss to detect and select a favorable path. Wallace and Shami (2014) proposed two modeling techniques: one is based on renewal theory, and the other applies the Markov chain.…”