Research On Network Congestion Detection And Scheduling Algorithms Under Unstable Transmission

Since the introduction of packet switching technology in the 1980s,computer net-works have been developing rapidly,and the number of network users has also risen sharply.Consequential problems such as network congestion and network resource scheduling have arisen.Network congestion is caused by too many packets arriving in a certain period of time.The router cannot process these packets in time,and thus accu-mulates in the buffer,which causes the network transmission delay to increase.In turn,the delay increases,which in turn reduces the network processing data.That causes a dramatic drop in network transmission efficiency,and deadlock in severe cases.As for network resource scheduling,consider the fact that packets from different users ar-rive at the router and form different queues in the buffer.Because the channel status of each user's link is inconsistent,in order to maximize the use of network resourceswhile ensuring the quality of service for all users,but also to consider certain fairness,a reasonable scheduling strategy needs to be effective for different queues.Consider the fact that under packet switching network,there is inevitable random delay jitter during the transmission of data packets.In particular,the channel interference existing in the wireless network will intensify the delay jitter.In severe cases,the range of delay jitter can reach fraction of a second.We call that as unstable transmission.Unstable transmis-sion interferes with the observation of network parameters such as queue size.However,congestion control algorithms and scheduling strategies all trigger corresponding con-trol mechanisms based on queue size or other queue size based information.Therefore,unstable transmissions will inevitably affect network congestion control and scheduling strategies,and even cannot guarantee the stability of the system.At present,relevant research results are not yet sufficient.Based on this,the main work of this paper is:Under the unstable transmission,the modeling of congestion detection and control al-gorithm is studied,and the robust stability problem related to the scheduling algorithm is discussed.The specific research work is as follows:(1)For the congestion control algorithm in the network layer,unstable transmis-sion will bring a certain degree of interference to the congestion detector,and then affect the performance of the congestion control algorithm.In order to accurately detect the congestion in the network,we use the time series method to analyze the traffic in the network based on the nature of the data packets in the TCP/IP network,establish the relevant models,and use the recursive least squares method to estimate the relevant parameters.Based on these,our congestion detection algorithm is obtained.Our al-gorithm considers the unstable transmission in the network and obtains the quantified relationship between the optimal forgetting factor of the parameters in the algorithm and unstable transmission.In addition,we compare the mainstream congestion detection al-gorithm based on the average queue size,we consider the higher order information of the queue size,and design a congestion detection algorithm based on the average queue size change rate,which is more advantageous in detecting potential congestion.(2)At present,the wireless network resource scheduling mainly adopts a cross-layer design method.According to the queue size and other information,the congestion control problem in network layer and the packet scheduling problem in MAC layer are jointly controlled and optimized to obtain a joint congestion-scheduling algorithm.However,this type of algorithm does not consider the unstable transmission of data packets,that is,it does not consider the effect of unstable transmission on the stability of the system.For this reason,on the basis of the joint congestion-scheduling algorithm,we model the network system reasonably based on the inflow rate and outflow rate of the data packets,apply the Markov random walk model to describe the change in the queue size,and verify the robust stability under unstable transmission.Our analysis method can also be extended to other scheduling algorithms.Finally,we summarize the work of this article and discuss future research direc-tions.