Research On Adaptive Active Queue Management Algorith

In the military field,after the "decoupling of weapon system and dynamic recombination of elements" of air defense and anti-missile system,a large amount of data is gathered to the command and control node through multiple links;in the civil field,the amount of data grows rapidly with the increasing number of users and terminals and converges to network nodes.In both military and civilian fields,when the massive amount of data cannot be processed by the intermediate nodes of the network in time,the data will be discarded quickly,leading to network congestion,which poses a new challenge to the active queue management algorithm.To address the difficulty of existing active queue management algorithms to alleviate network congestion in different environments in both military and civilian domains,this thesis investigates active queue management algorithms in static and dynamic network environments,respectively,and the main work is as follows:(1)To address the problem that existing algorithms are ineffective in congestion control in static network environments because the discard policy does not accurately respond to congestion and is sensitive to parameter settings,an active queue management algorithm(QSM-AQM)based on queue state model is proposed.Based on the AQMRD algorithm and Ag RED algorithm,the algorithm introduces a queue length change rate,designs an intermediate threshold and formulates a discard policy to reflect the network congestion.Experiments conducted with two parameters in two load cases show that the packet loss rate in the QSM-AQM algorithm decreases by 10.74% and 9.78%,the end-to-end average delay decreases by 1.24% and 1.14%,and the delay jitter decreases by 32.36% and 48.98% in the parameter one cases of load one and load two,respectively.In the parameter two cases of load one and load two,the packet loss rate in the QSM-AQM algorithm decreases by 10.53% and7.31%,the end-to-end average delay decreases by 1.10% and 1.45%,and the delay jitter decreases by 29.00% and 22.75%,respectively.(2)To address the problem that the existing algorithms are ineffective in congestion control in dynamic network environments due to their inability to adapt to constant changes in the environment and slow response to congestion,an active queue management algorithm(SAE-AQM)based on policy adaptability evaluation is proposed.The algorithm is based on the QSM-AQM algorithm and the MDPRED algorithm,and qualitatively sets the intermediate threshold,reconfigures the discard policy and accomplishes the dynamic adjustment of the weight parameters,so that the algorithm can quickly adapt to the dynamic changes of the network environment.Experiments conducted in two parameter cases in three dynamic network environments show that the packet loss rate of SAE-AQM algorithm decreases by 7.45%,9.03% and 10.94% in the parameter one case in environment one,environment two and environment three,respectively;the end-to-end average delay decreases by 1.51%,1.41% and 1.43%,respectively;the delay jitter is reduced by 31.25%,42.71% and49.57%,respectively.In the case of parameter two,the packet loss rate of SAE-AQM algorithm is reduced by 9.61%,9.89% and 10.60%,respectively;the end-to-end average delay is reduced by 1.42%,1.63% and 1.88%,respectively;and the delay jitter is reduced by45.00%,40.20% and 44.10%,respectively.