An Improvement over Random Early Detection Algorithm: A Self-Tuning Approach

Random Early Detection (RED) is one of the most commonly used Active Queue Management (AQM) algorithms that is recommended by IETF for deployment in the network. Although RED provides low average queuing delay and high throughput at the same time, but effectiveness of RED is highly sensitive to the RED parameters setting. As network condition varies largely, setting RED's parameters with fixed values is not an efficient solution. We propose a new method to dynamically tuning RED's parameters. For this purpose, we compute the rate of which the queue is occupied and consider it as a congestion metric that will be forecasted when the queue is overloaded. This meter is used to dynamically setting RED parameters. The simulation results show the effectiveness of the proposed method. According to the results, we achieve a significantly higher utilization and less packet loss comparing to original RED algorithm in dynamic conditions of the network.