D-OLIA: A Hybrid MPTCP Congestion Control Algorithm with Network Delay Estimation.

Area of Science:

• Computer Science
• Networking

Background:

• Modern devices utilize multiple communication interfaces, driving the evolution of Multipath Transmission Control Protocol (MPTCP).
• MPTCP congestion control algorithms (MPTCP-CCAs) aim for improved throughput, fairness, and balanced congestion, but current algorithms struggle with dynamic network conditions and short-lived traffic.
• Existing MPTCP-CCAs like OLIA often underutilize network capacity, leading to suboptimal performance.

Purpose of the Study:

• To propose a novel MPTCP-CCA, dynamic OLIA (D-OLIA), that enhances performance by incorporating network delay awareness.
• To improve the ability of MPTCP to fully utilize network capacity, especially in the face of rapidly changing Internet traffic characteristics.

Main Methods:

• Developed D-OLIA, a hybrid MPTCP-CCA that dynamically adjusts the congestion window (CWND) decrease factor based on estimated network delay.
• Estimated network delay (less-congested or congested) by monitoring Round-Trip Time (RTT) changes.
• Implemented D-OLIA in the Linux kernel and validated its performance using the Mininet emulator.

Main Results:

• D-OLIA accurately estimates current network delay conditions.
• Experimental results show D-OLIA achieves approximately 20% increased throughput compared to the original OLIA.
• D-OLIA demonstrates significant improvements in throughput, RTT, packet retransmissions, and fairness over other MPTCP-CCAs.

Conclusions:

• D-OLIA effectively addresses the limitations of current MPTCP-CCAs by adapting to network delay.
• The proposed algorithm offers a promising solution for enhancing MPTCP performance in modern, dynamic network environments.