An adaptive network congestion control strategy based on the change trend of average queue length

With the rapid growth in the amount of data transmissions over Internet of Things (IoT) networks, a large amount of bursty traffic is more prone to cause serious network congestions, which would degrade network performance by reducing throughput, increasing end-to-end delay and packet loss, etc. Random early detection (RED) has been the most widely known method for network congestion control. However, RED and its most variant algorithms control network congestion only based on the queue length. More importantly, these algorithms ignore the direction and speed of queue changes, thus limiting their capability to adapt to fluctuations in network traffic. In this paper, a new algorithm called queue change trend based adaptive RED (QCT-ARED) is proposed. First, a novel average queue length evaluation model is introduced. This model can adjust the queue weight values according to the queue state adaptively. Besides, the proposed algorithm brings in a new parameter representing the second-order change rate of the average queue length. It helps to update the intermediate threshold according to the information on the average queue length and its first-order change rate. Finally, the QCT-ARED algorithm adopts a combination of cubic function and linear function for the dropping function settings as a way to predict and avoid early congestion. The method is compared with some existing algorithms, the results show that the proposed algorithm has a better system performance in terms of queue length, delay jitter, and throughput.