As new congestion control mechanisms are developed, their performance relative to existing mechanisms needs to be understood, in particular over wireless networks. This study aimed to evaluate existing TCP congestion control mechanisms using a comprehensive and reproducible methodology designed to be representative of real world usage of wireless networks. The study sought to investigate whether any existing mechanism could provide significant performance benefits over CUBIC and be recommended for adoption. The findings of this study showed that YeAH demonstrated anincrease in throughput of 3% -- 5% over CUBIC, with no penaltyto latency. While this small improvement may assist applicationsrequiring the highest available performance, it is unlikely that itwill provide a significant improvement over existing congestioncontrol mechanisms. As such, it is the conclusion of this study thatuse of alternate congestion control mechanisms would not providenoticeable improvements in performance in most applications.
展开▼
