Research On Queue Management Algorithm Based On Self-similar Traffic Prediction In AOS

In order to adapt to the rapid development of aerospace technology,the Consultative Committee for Space Data Systems(CCSDS)has introduced a flexible and diverse data processing service named advanced orbiting system AOS provides a variety of transmission schemes,has a variety of user data format,and uses packet channel multiplexing and virtual channel multiplexing mechanism,so that a variety of data can share the scarce space link resources.Nevertheless,with the increasing of spatial data and the self-similarity of data stream,the congestion of space network becomes inevitable.In order to alleviate the congestion of network,this thesis proposes an ARED queue management algorithm based on self-similar traffic prediction to realize the effective control of congestion.The main contents of this thesis include:Firstly,the self-similarity of network traffic and the method of self-similar traffic prediction are introduced in detail.Network traffic has self-similarity,which means that the traditional Poisson model is not suitable for self-similar traffic.In this thesis,we propose two business models: ON/OFF model and FBM model,which are suitable for self-similarity,and then model and simulate both of them.This thesis introduces the basic ideas of self similar traffic prediction,and puts forward two kinds of prediction models: fractal model and statistical prediction model.We choose statistical prediction model to predict the traffic generated by the ON/OFF model,and verify the correctness of the simulation results.Then,the ARED algorithm based on self-similar traffic prediction is realized.The traffic flow is introduced into the ARED algorithm,and the data packet is discarded according to the dropping probability,which can reduce the data transmission rate,ease the network congestion and improve the utilization ratio of the channel.Combined the queue management with the virtual channel multiplexing,the AOS queue management and virtual channel scheduling system model is proposed.Then selects the source model of manned spacecraft for the whole system modeling simulation,uses MATLAB software to build simulation platform for experiments.At the same time,every performance index of the simulation results is evaluated,the average delay of each VC,the channel utilization,and the maximum time delay of the system are verified and compared.The conclusion is that the ARED queue management algorithm based on self similar traffic prediction can greatly improve the performance of the system.