Estimation of TCP Congestion Control Algorithms Detection of Malicious Giant TCP Flow from Packet Logs

2018 Fiscal Year Final Research Report

• Project/Area Number: 16K06341
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Communication/Network engineering
• Research Institution: The University of Electro-Communications
• Principal Investigator: KATO Toshihiko 電気通信大学, 大学院情報理工学研究科, 教授 (90345421)
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: TCP / 輻輳制御 / 通信ログ

Outline of Final Research Achievements

Recently, various TCP congestion control mechanisms have been introduced. Since the TCP congestion control algorithms affect the performance of the Internet, it is important to analyze which algorithms are used widely. This research focuses on a passive scheme to infer a congestion control algorithm from passively collected packet traces by estimating congestion window at round-trip time (RTT) intervals, and inferring congestion control algorithms by correlating estimated window sizes and their increments. Specifically, two methods are proposed. One is a method for bidirectional packet traces that estimates congestion window sizes by mapping data and ACK segments. The other is a method for unidirectional traces including only data segments. It uses the curve fitting for sequence number vs. time graphs by applying the least squares method with linear through quartic functions, and maps the first-order and second-order differentiations.

Free Research Field

情報ネットワーク

Academic Significance and Societal Importance of the Research Achievements

輻輳制御方式はTCPトラヒックを特徴づけるものである．このため，ネットワーク事業者にとっては，自分の運営するネットワークにおいてどのような輻輳制御方式がどの程度利用されているかを調査することは，重要な意味があると考えられる．しかしこれまでは，パッシブに収集された通信ログから，新たに提案された輻輳制御方式（HighSpeed TCPやCUBIC TCPなど）を推定する方法は，まったく提案されていなかった．本研究では，これまでにない手法により，双方向および片方向の通信ログから輻輳制御方式を推定することを可能とした．