In this paper, we propose a new congestion control scheme for high-speed networks. The basic idea of the scheme is to adopt a game theory called minority game (MG) to realize a selective reduction of the transmission speed of senders. More concretely, upon detecting a congestion, it plays a game among all senders participating in the communication, and reduces the transmission speed of each sender according to the result of the game. MG is a game that has recently attracted considerable attentions, and it is known to have a remarkable property such that the number of winners converges to a half of the number of players in spite of the selfish behavior of the players. By using such property of MG, we can realize a fair reduction of the transmission speed, which is more efficient than previous schemes in which all senders uniformly reduce their transmission speed. The effect of the proposed scheme is evaluated by simulation. The result of simulations indicates that the proposed scheme certainly realizes a selective reduction of the transmission speed, and it is sufficiently fair compared with simple randomized schemes and is sufficiently efficient compared with conventional schemes
[1]
Raj Jain,et al.
Analysis of the Increase and Decrease Algorithms for Congestion Avoidance in Computer Networks
,
1989,
Comput. Networks.
[2]
Vassilis Tsaoussidis,et al.
Bimodal Congestion Avoidance and Control
,
2003
.
[3]
N. Johnson,et al.
Self-Organized Segregation within an Evolving Population
,
1998,
cond-mat/9810142.
[4]
Paczuski,et al.
Self-organized networks of competing boolean agents
,
2000,
Physical review letters.
[5]
Tom Kelly,et al.
Scalable TCP: improving performance in highspeed wide area networks
,
2003,
CCRV.
[6]
Esteban Moro.
The Minority Game: an introductory guide
,
2004
.
[7]
Raj Jain,et al.
A Quantitative Measure Of Fairness And Discrimination For Resource Allocation In Shared Computer Systems
,
1998,
ArXiv.
[8]
Bogdan M. Wilamowski,et al.
The Transmission Control Protocol
,
2005,
The Industrial Information Technology Handbook.