Performance Analysis of Diverse Active Queue Management Algorithms
摘要
Congestion is a problem which can affect a computer’s network performance, typically occurring when the available resources of the network cannot deal with arriving packets, thus deteriorating performance. Many research studies have acknowledged various Active Queue Management (AQM) methods to alleviate congested networks in an attempt to improve resource management and performance. However, the performance of AQM methods differs significantly and, more importantly, can be influenced by the level of congestion (light, moderate, or heavy). This paper empirically compares three different AQM methods, Curvilinear Random Early Detection (CLRED), Three-section Random Early Detection (TRED) based on nonlinear RED and Enhanced Adaptive Gentle Random Early Detection (Enhanced AGRED), to identify which method most effectively manages congestion in a single queue node system and also at a router buffer in a queueing network system. This router buffer may have more arriving packets than the departing packet. These methods are compared based on distinctive criteria including packet arrival probability, maximum packet-dropping probability and packet arrival probability (Alpha 1 or