Model Integrating Fuzzy Argument with Neural Network Enhancing the Performance of Active Queue Management

Posted by Admin:

Model Integrating Fuzzy Argument with Neural Network Enhancing the Performance of Active Queue Management

Nguyen Kim Quoc1,*, Vo Thanh Tu1 and Nguyen Thuc Hai2

1College of Sciences, Hue University, Vietnam

2Ha Noi University of Science and Technology, Vietnam

Abstract

The bottleneck control by active queue management mechanisms at network nodes is essential. In recent years, some researchers have used fuzzy argument to improve the active queue management mechanisms to enhance the network performance. However, the projects using the fuzzy controller depend heavily on professionals and their parameters cannot be updated according to changes in the network, so the effectiveness of this mechanism is not high. Therefore, we propose a model combining the fuzzy controller with neural network (FNN) to overcome the limitations above. Results of the training of the neural networks will find the optimal parameters for the adaptive fuzzy controller well to changes of the network. This improves the operational efficiency of the active queue management mechanisms at network nodes.

1. Introduction
Along with the advantages that the fuzzy control system has brought to improvements on the active queue management mechanism, there still exists some shortcomings such as the design and optimization of the fuzzy system require some experiences in controlling subjects. On the other hand, for the fuzzy controller to work well, it should have the parameters (the number of fuzzy sets in each language variable , the shape of fuzzy sets, the number of laws and the weight of each law) to be optimal [13][28][29]. To solve this problem, we offer the model integrating fuzzy control with neural network in order to improve the performance of active queue management mechanisms.

Download