Research On Key Technolgoies Of Multipath Transmission For Mobile Streaming Media

In recent years,with the rapid development of mobile communication technologies represented by 5G and the rapid popularization of hardware such as smart phones,laptops and wearable devices,mobile streaming media applications have experienced enough development.The Mobile streaming media applications,such as online classes/conferences,video calls and real-time games,not only greatly facilitate people’s work,life and entertainment but also bring massive real-time data to the transmission.In the mobile network environment,due to the influence of user mobility and link error,the transport layer faces the problems of path unstability and high packet-loss rate.The traditional single-path transmission protocols represented by TCP are difficult to deal with the complex and changeable mobile network environment.Their transmission rates are limited by the capacity of single network and have poor connection stability,which cannot meet the transmission requirements of mobile streaming media services.In order to solve the above problems,the multipath transmission protocols represented by MPTCP(Multipath Transmission Control Protocol)have attracted many researchers.MPTCP can use the differert network interfaces(Wi-Fi,4G/5G,etc.)to establish multiple sub-paths in one connection.Compared with single-path protocol,multipath transmission can improve the system throughput and reduce the transmission delay by aggregating the idle bandwidth of heterogeneous networks.Despite the above advantages,when serving the mobile streaming media applications,MPTCP still has the following problems:(1)Low efficiency of scheduling strategy.MPTCP adopts the shortest round-trip-time-based scheduling,and retransmits the packets indiscriminately after transmission failure,which will lead to the out-of-order packets and increases the transmission delay;(2)Fixed congestion control strategies.MPTCP adapts fixed congestion control strategies,which is unable to adjust with the various input streaming and reduces the transmission efficiency(3)Network unawareness.MPTCP dose not consider the network influence in transmission,which leads to inefficient transmission control.(4)Lack of energy management.Due to the limited energy of mobile terminals,the lack of energy management in MPTCP will reduce the service life of mobile devices and influence the user experience.Thus,this paper focuses on making breakthroughs in dynamic data scheduling,intelligent congestion control,network environment awareness,energy-efficiency balance,and obtains the following results:(1)To solve the problem of inefficient scheduling,a partially reliable multipath virtual queue scheduling mechanism is proposed.Firstly,a virtual queue model is constructed to break the traditional constraint of sending window and ensure the in-order arrival of packets to the greatest extent.Then,a message-riented partially reliable retransmission algorithm is designed to avoid invalid retransmission.Simulation results show that under different network parameters,the proposed scheme can effectively reduce out-of-order probability and packet delivery delay.(2)To solve the congestion control rigidity problem,a multipath intelligent congestion control mechanism based on input feature analysis is designed.Firstly,the features of the data stream are extracted in the frequency domain.On this basis,reinforcement learning theory is used to model the multipath congestion control,and the control strategy is dynamically adjusted according to the network environment and input flow.The simulation results show that the proposed mechanism can achieve high throughput improvement with little delay cost,and performances better than the compared solutions.(3)To solve the problem of network unawareness,a multipath adaptive transmission scheme based on coupled subflow perception is proposed.Firstly,the influence of network environment on transmission is quantified from space and time by extracting the subflow coupling features of subflow and congestion features.Then,based on the deep Q network method,the multipath transmission system is trained.The simulation results show that the proposed scheme can achieve better system throughput and delay than the compared solutions under various transmission scenarios.(4)To solve the problem of no energy management,an optimization algorithm of energy-efficiency trade-off was designed.Firstly,the transmission utility model is designed to comprehensively analyze the bandwidth,delay and energy consumption that affect the transmission performance.Then,through Q-learning method,the transmission energy consumption is optimized on the basis of considering the transmission efficiency.Simulation results show that the proposed algorithm can reduce transmission energy consumption on the basis of getting similar transmission throughput and delay performance.This paper analyzes the scheduling,congestion,network,energy consumption problems that affect the mobile streaming media transmission.And use statistical analysis,mathematical modeling,artificial intelligence,experimental testing methods to solve the corresponding problems.The research results of this paper can provide some technical support for the Chinese future streaming media services.