Adaptive Hurst-Sensitive Active Queue Management

Abstract

An Active Queue Management (AQM) mechanism, recommended by the Internet Engineering Task Force (IETF), increases the efficiency of network transmission. An example of this type of algorithm can be the Random Early Detection (RED) algorithm. The behavior of the RED algorithm strictly depends on the correct selection of its parameters. This selection may be performed automatically depending on the network conditions. The mechanisms that adjust their parameters to the network conditions are called the adaptive ones. The example can be the Adaptive RED (ARED) mechanism, which adjusts its parameters taking into consideration the traffic intensity. In our paper, we propose to use an additional traffic parameter to adjust the AQM parameters-degree of self-similarity-expressed using the Hurst parameter. In our study, we propose the modifications of the well-known AQM algorithms: ARED and fractional order PIαDβ and the algorithms based on neural networks that are used to automatically adjust the AQM parameters using the traffic intensity and its degree of self-similarity. We use the Fluid Flow approximation and the discrete event simulation to evaluate the behavior of queues controlled by the proposed adaptive AQM mechanisms and compare the results with those obtained with their basic counterparts. In our experiments, we analyzed the average queue occupancies and packet delays in the communication node. The obtained results show that considering the degree of self-similarity of network traffic in the process of AQM parameters determination enabled us to decrease the average queue occupancy and the number of rejected packets, as well as to reduce the transmission latency.

Keywords: PID; adaptive AQM; neural networks; reinforcement learning; self similarity.

Figure 1

Hurst calculation algorithm.

Figure 2

TCP/UDP streams in the adopted Fluid Flow approximation.

Figure 3

Router average queue length values, Fluid Flow approximation, ARED Hurst-insensitive 1 TCP stream (left, top), ARED Hurst-sensitive 1 TCP stream (right, top) and ANRED Hurst-insensitive 1 TCP stream (left, bottom), ANRED with Hurst-sensitive 1 TCP stream (right, bottom).

Figure 4

Router average queue length values, Fluid Flow approximation, $ANP{I}^{\alpha }$ Hurst-insensitive 1 TCP stream (left, top), $ANP{I}^{\alpha }$ Hurst-sensitive 1 TCP stream (right, top) and $ANP{I}^{\alpha }{D}^{\beta }$ Hurst-insensitive 1 TCP stream (left, bottom), $ANP{I}^{\alpha }{D}^{\beta }$ Hurst-sensitive 1 TCP stream (right, bottom).

Figure 5

Network node topology in the adopted simulation method.

Figure 6

Router queue length values, $\mu$ = 0.25, ARED Hurst-insensitive algorithm, $\alpha =0.5$, H = 0.5 (left, top), H = 0.9 (right, top) and ARED Hurst-sensitive algorithm, $\alpha =0.5$, H = 0.5 (left, bottom), H = 0.9 (right, bottom).

Figure 7

Router queue length values, $\mu$ = 0.25, ANRED Hurst-insensitive algorithm, $\alpha =0.5$, H = 0.5 (left, top), H = 0.9 (right, top) and ANRED Hurst-sensitive algorithm, $\alpha =0.5$, $\mu$ = 0.25, H = 0.5 (left, bottom), H = 0.9 (right, bottom).

Figure 8

Queue lengths, $\mu$ = 0.25, $P{I}^{\alpha }$ Hurst-insensitive algorithm, $\alpha =0.5$, H = 0.5 (left, top), H = 0.9 (right, top) and $P{I}^{\alpha }$ Hurst-sensitive algorithm, $\alpha =0.5$, H = 0.5 (left, bottom), H = 0.9 (right, bottom).

Figure 9

Parameters evolution, $\mu$ = 0.25, $P{I}^{\alpha }$ Hurst-insensitive algorithm, $\alpha =0.5$, H = 0.5 (left, top), H=0.9 (right, top) and $P{I}^{\alpha }$ Hurst-sensitive algorithm, $\alpha =0.5$, H = 0.5 (left, bottom), H = 0.9 (right, bottom).

Figure 10

Queue length values, $\mu$ = 0.25, $P{I}^{\alpha }{D}^{\beta }$ Hurst-insensitive algorithm, $\alpha =0.5$, H = 0.5 (left, top), H = 0.9 (right, top) and $P{I}^{\alpha }{D}^{\beta }$ Hurst-sensitive algorithm, $\alpha =0.5$, H = 0.5 (left, bottom), H = 0.9 (right, bottom).