Title:	EXPERIMENTAL ANALYSIS OF MPTCP CONGESTION CONTROL ALGORITHMS; LIA, OLIA AND BALIA
Author(s):	Toshihiko Kato, Adhikari Diwakar, Ryo Yamamoto, Satoshi Ohzahata and Nobuo Suzuki
ISBN:	978-989-8533-92-0
Editors:	Ajith P. Abraham and Jörg Roth
Year:	2019
Edition:	Single
Keywords:	MPTCP, Congestion Control, LIA, OLIA, BALIA, TCP Reno, CUBIC TCP
Type:	Full Paper
First Page:	135
Last Page:	142
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	In Multipath TCP (MPTCP), the congestion control is realized by individual subflows (conventional TCP connections). However, there is some concern that the total data sending rate becomes too high compared with other single path TCPs if the associated subflows increase their own congestion windows independently. So, a coupled increase scheme of congestion windows, called Linked Increase Adaptation (LIA), is adopted as a standard congestion control algorithm for subflows comprising a MPTCP connection. But this algorithm supposes that TCP connections use Additive Increase and Multiplicative Decrease (AIMD) based congestion control, and if high speed algorithms such as CUBIC TCP are used, the throughput of MPTCP connections might be decreased. In order to increase the throughput of MPTCP connection, two modified congestion control algorithms, Opportunistic Linked-Increase Algorithm (OLIA) and Balanced Linked Adaptation (BALIA) are also introduced. However, they are still based on AIMD and there is some concern that they are weaker than high speed algorithms. This paper analyzes the performance of MPTCP congestion control algorithms through experiments. Specifically, this paper examines to compare LIA, OLIA or BALIA based MPTCP flow and a single TCP flow with TCP Reno or CUBIC TCP. These experiments show that LIA is weaker than both Reno and CUBIC, and OLIA and BALIA are weaker than CUBIC.
	Go Back