Enhancing fairness and congestion control in multipath TCP

Abstract

With the advancement in technology, today most of the end-host devices are equipped with multiple wired/wireless interfaces and capable of using them in parallel. This has led to research in taking advantage of utilizing additional available network resources simultaneously such as multiple paths between the multi-homed end devices to achieve better performance. In this regard, new multipath transport protocols such as Multipath TCP and Concurrent Multipath Transfer SCTP are being designed and developed. On one hand these new protocols need to optimize the usage of the available multiple paths efficiently with a suitable scheduler while on the other hand they should be also fair to the existing transport protocols like TCP and SCTP, especially in a congested environment. This paper illustrates and compares the solutions that have been proposed to ascertain that Multipath TCP can yield an improved performance while being fair to standard TCP. In addition, this paper highlights improvements to two of the proposed congestion control solutionsDynamic Window Coupling and Opportunistic Linked Increases Algorithm. A brief look into hybrid scheduling mechanisms is also provided with a special interest in minimizing the overall reordering delay and to capture the dependency on the receiver buffer size. It is shown that the proposed solutions give good performance in most of the different bottleneck scenarios while staying friendly to legacy transport protocols.

Impact of Path Characteristics and Scheduling Policies on MPTCP Performance

“…The authors of [7] perform a theoretical analysis of the problem, characterizing factors governing responsiveness, TCP-friendliness, and window oscillations of a joint congestion control method for MPTCP. Other work such as [11], [3], and [14] are further examples of work focused on improving MPTCP joint congestion control.…”

Multipath Transmission for the Internet: A Survey

“…Packet reordering It is a hybrid scheduler used to efficiently operate packet scheduling by combining both push and pull style strategies. For example, the hybrid scheduler presented in [182] allocates data segments to active flows (Pull) with dynamic size (Push).…”

“…For example, DWC uses a loss congestion event to trigger an alert while using either the Delay or Loss congestion event to group subflows. Singh et al [182] proposed an extension of the DWC algorithm, denoted as EDWC. This extension uses a delay congestion to trigger an alert while either delay or loss congestion is used to group and couple subflows.…”

“…OLIA also compensates for different RTTs by adapting the window increases as a function of RTTs. However, Singh et al [182] found that OLIA of MPTCP still has performance issues. They presented Adapted Opportunistic Linked Increases Algorithm (AOLIA) to ensure controlled aggressiveness of the MPTCP subflows.…”

MPTCP combining congestion window adaptation and packet scheduling for multi-homed device

“…In fact, stable goodput with minimal variation is preferable for QoS assurance to real-time applications. Author in [6] has studied different congestion control variants for Multipath TCP have been compared. Also author has investigated a couple of relevant hybrid scheduler algorithms that are based on the two implementation strategies, Push and Pull.…”