aqm system and method for stochastic early detection algorithm based on generalized pid

Abstract

The invention discloses an AQM system and method based on a generalized PID random early detection algorithm. The system includes a BP neural network module, a generalized PID control module and a random early detection module; the BP neural network can find an optimal parameters under the control law. The generalized PID control module performs proportional, integral, and differential operations on the adjusted parameters and stabilizes the queue length at a corresponding fixed value under different network loads. The random early detection module obtains this fixed value as the maximum discard probability, and generates a The drop probability applicable to the current load, and the controlled object responds accordingly according to the drop probability, so that the queue length in the router can be stabilized to an expected value. In this way, the utilization rate of network resources is improved and the average delay of the network is reduced.

Description

technical field [0001] The invention relates to an AQM system, in particular to an AQM system based on a generalized PID random early detection algorithm. Background technique [0002] An important goal of Active Queue Management (ie: AQM) is to detect network congestion early and send instructions to the system to control congestion, so that the queue length in the router can be stabilized to an expected value. This stability can improve the utilization rate of network resources and reduce the average network delay. The main algorithm of AQM, the random early detection algorithm, has been widely used in IP and asynchronous transfer mode (ie: ATM) routers, and has achieved certain results. However, since the parameter setting of the random early detection algorithm (ie: RED algorithm) has not been well resolved, the change of the network load is likely to cause the instability of the AQM system, so its performance has been greatly affected. From the perspective of RED's wo...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to provide an AQM system based on a generalized PID random early detection algorithm. The system combines the BP neural network and utilizes the generalized PID controller in the control theory to dynamically adjust the maximum discarding probability of the RED algorithm according to the network load status, so that the AQM The system adaptively adjusts its own parameters to improve control performance. This system solves the problem that the parameters of the RED algorithm cannot be set in real time, and the change of network load easily causes the instability of the AQM system.

Images