A new queue management scheme for AIMD based flows with proportional fair scheduling in wireless networks

Abstract

In this paper, we studied the interaction between the proportional fair (PF) scheduling, one of the most popular wireless packet scheduling algorithms, and the additive increase and multiplicative decrease (AIMD) flow control algorithm such as in TCP. The PF scheduler is known to achieve not only efficiency but also fairness in wireless packet networks. Since TCP is one of the most popular transport protocols, understanding the interaction between them is worthwhile. We found out that with the AIMD control, the PF scheduling may not achieve the expected fairness and efficiency when channel conditions of users are not similar. To resolve the issue, we proposed a new queue management scheme, called"proportional queue management scheme". By isolating flows and setting queue thresholds proportional to their channel quality, the efficiency and fairness are both improved significantly. The simulation results with NS-2 confirmed the effectiveness of our algorithm.