A Reformed Dropping Function-Based Active Queue Management Mechanism for Network Routers

Abstract

In light of the fact that the rate of traffic generation on the Internet is ever increasing, the need for an effective algorithm that takes a firm stand against the adverse impact of network traffic congestion remains germane. Since its dissemination in the early 1990s, the Random Early Detection - in short RED has gained prominence as a prevailing active queue management (AQM) scheme put to use in functional network routers. However, guaranteeing a better network performance under changing congestion levels, particularly in respect to high congestion cases is considered an important issue in RED and is attributable chiefly to the influence of the selfsame linear function it exerts as a packet discarding probability function; thus there appears to be a renewed and continuous interest in the development of new algorithms that create an improved dropping functions. As a result, a vast diversity of published research articles has arisen. Propelled by this concern, in this paper, a new RED-dependent algorithm, called Trilateral Segment-Random Early Detection (TS-RED) is developed. Very importantly, the new algorithm is carefully designed to espouse the benefits of three dropping functions (i.e., quadratic plus square root together with exponential in this case) deemed as having an increasing growth in the rate of packet dropping. Results from simulation experiments shows that TS-RED is successful at delivering a better stability and keeping a tight rein on the queue occupancy when compared with two other popular RED-like easy-to-implement contemporary AQM algorithms established in the literature.