D-OLIA: The Packet Loss Differentiation Based Opportunistic Linked-Increases Algorithm for MPTCP in Wireless Heterogeneous Network

Abstract

Compared with wired networks, there are random packet loss (RPL) and congestion packet loss (CPL) in wireless heterogeneous network where mobile terminals are located. In the current strategies of congestion control (CC) algorithms, however, they are all universally recognized as packet loss (PL), which is expressed by a sign of congestion. In this way, RPL may be mistaken for CPL. Meanwhile, there is no corresponding congestion window adjustment method for different PL. Blind reduction of congestion window for traffic control will only lead to the deterioration of multipath transmission control protocol (MPTCP) performance. Based on this reality, this paper proposes a PL Differentiation based Opportunistic Linked-Increases Algorithm, called D-OLIA. The D-OLIA can determine the type of PL by combining the eigenvalues of delay jitter and congestion window jitter, so as to make up for the shortage of judging only by delay or congestion window. We conduct simulation experiments in NS-3-DCE platform, the evaluation results show that the D-OLIA can effectively increase the PL differentiation accuracy and the throughput of mobile communication, as well as satisfy fairness and better balance congestion in the case of shared bottleneck based on MPTCP.