Research On Congestion Control Technology For QKD Network

With the development of quantum computers,computing power increases geometrically,and the security of asymmetric cryptographic key negotiation algorithms such as RSA is severely challenged.However,quantum secure communication networks can provide quantum key distribution services.At present,quantum key distribution technology is in a basic stage of development,and the generation rate of quantum key is low.It needs further development to produce keys on a large scale to support commercial and civil secure communication.In the traditional communication network,when the network load exceeds the processing capacity of the network,the network will be congested,and the network transmission efficiency will decline.The traditional TCP/IP network solves the end-to-end network congestion through the TCP transport layer protocol.The quantum network also has the problem of network congestion.If the encryption keys of the relay key pool in the quantum communication network are exhausted,the quantum secure communication network will not be able to provide network encryption services for users.In addition,if the key generation rate fails to keep pace with the consumption rate,although the key in the key pool is sufficient to support the network encryption service for a period of time,there will always be a time node that causes the key pool to stagnate and the network to collapse.Starting from the traditional network congestion control mechanism,this paper attempts to explore and study the QKD-oriented network congestion control mechanism and scheme.1.Based on the study of the traditional congestion control mechanism and technology,this paper proposes a congestion avoidance strategy based on Cauchy distribution,aiming at the problem that the traditional strategy measures network congestion based on RTT.This strategy mainly samples the RTT,predicts the future network congestion situation through the previous network congestion situation,and makes timely adjustment to avoid congestion.The advantage of this strategy is that don't need the RTT value for precise sampling,just some extremum of RTT sampling,characteristic analysis and to establish statistical model,will be the last time the information transmission congestion window size iterative statistical model function,dynamic update forecast the next transmission congestion window value,make the network can achieve higher throughput in a relatively short time.Secondly,it is applied to QKD network to improve the efficiency of key distribution and performance of QKD network.Finally,NS-3 open source network simulation software is used to carry out simulation experiments,and compared with TCP New Reno network congestion algorithm,the experimental results show that this strategy can effectively improve the throughput.2.For the centralized QKD network based on SDN,a centralized QKD network congestion control mechanism based on Cauchy distribution is proposed.This mechanism detects congestion through intermediate nodes,classifies different network congestion,and reports congestion messages to controller.According to the packet analysis,the controller issues congestion adjustment policy to the source terminal.This strategy can quickly adjust the network congestion according to the link condition,avoid multi-node congestion and prevent congestion upgrade.Combined with the congestion avoidance method,the controller collects the round-trip delay transmitted over the whole link,and uses the statistical model to avoid the congestion by means of prediction.Finally,through NS-3 simulation experiment,compared with the European project SECOQC distributed quantum secure communication network,the experimental results show that the congestion control mechanism CDCCNQKD under the SDN centralized network architecture can increase the proportion of network packet sending and receiving,reduce the flow completion time,increase the network throughput and improve the QKD network performance to a certain extent.