| As a new generation of broadband mobile communication technology,5th Generation Mobile Communication Technology(5G)has the characteristics of high bandwidth,low latency and large connection.5G is the network infrastructure for realizing human-machine interconnection.Millimeter wave(30-300 GHz electromagnetic wave)communication brings speed and delay advantages for 5G,but also brings challenges to the design of the upper layer transmission protocol of 5G network,especially the design of congestion control mechanism.(1)The TCP protocols widely used in the network currently adopt the slow start and the congestion control mechanism of additive growth and multiplicative reduction,which greatly limits the utilization rate of bandwidth,and will inevitably lead to waste of bandwidth in the high-bandwidth 5G network.Therefore,how to improve the bandwidth utilization in 5G network is a challenge faced by the congestion control mechanisms;(2)The existing congestion control mechanisms use packet loss as a signal of network congestion,A large number of packets are queued in the buffer,which increases the end-to-end transmission delay.Therefore,how to reduce the end-to-end transmission delay to highlight the advantage of the 5G network is the second challenge faced by the congestion control mechanisms;(3)5G millimeter waves are easily affected by the external environment.Therefore the channel quality is extremely unstable,resulting in highly dynamic changes in bandwidth and delay.How to quickly detect changes in bandwidth and delay to adapt to the high dynamics of 5G networks is the third challenge.In this thesis,the above mentioned issues are investigated and the contributions are as follows.1.Adaptability of congestion control protocols in 5G network are analyzed.The current research on the performance of congestion control protocols in 5G networks is not systematic and comprehensive.The existing related research only selects individual congestion control protocols for performance analysis,and the experimental scenarios are relatively simple,so a comprehensive and effective adaptive analysis conclusion cannot be obtained.In view of this,this thesis designs a5 G mm Wave simulation scenario based on the Network Simulator-3(NS-3).8 typical congestion control protocols,including packet loss-based,delay-based and hybrid,are selected to carry out a wide range of experimental studies in three scenarios.The performance parameters such as throughput and round-trip delay of these protocols are analyzed.The results show that BBR has the best adaptability to 5G networks among all protocols.However,it responds slower to network congestion and forms longer queues in the cache,resulting in a significant increase in round-trip delay.In addition,BBR cannot detect changes in bottleneck bandwidth in time in highly dynamic scenarios,resulting in reduced throughput and round-trip delay performance.2.In view of the slow response of BBR protocol to network congestion,this thesis proposes a New BBR protocol,which uses a fast congestion-aware algorithm to improve BBR's congestion detection mechanism.BBR needs to go through three rounds of detection to enter the emptying state,and the fast congestion-aware algorithm uses the round-trip delay and the rate of change of the packet transmission rate to detect the congestion point.increase.Simulation results show that New BBR shows higher throughput and lower round-trip delay in all three scenarios,especially in mobile scenarios,its throughput performance is improved by 5% compared to BBR.3.Aiming at the problem that BBR cannot detect the bottleneck bandwidth change in time in high dynamic scenarios,this thesis proposes a new congestion control protocol Kalman-BR based on Kalman filter.Firstly,the available bandwidth model of the end-to-end link is constructed.Secondly,a dynamic bandwidth prediction algorithm based on Kalman filtering is proposed.The simulation results show that the Kalman-BR protocol achieves higher throughput and lower round-trip delay in all three scenarios,especially in scenarios 2 and 3. |
PDF Full Text Request