Research On Multipath TCP Transmission And Control For VANETs

With the increasing number of Internet users,the amount of daily transmission of data in the network is also increasing and the network users to the network experience requirements are gradually increased.So it is expected that the future high bandwidth,high reliability,low cost and user privacy protection and other new network transmission needs will gradually be raised.With the network equipment and terminals with multiple network interfaces more common,multi-sink transmission technology is proposed by the relevant researchers.With the continuous evolution of multi-homed transmission technology,the IETF has developed MPTCP multiplexing protocol,which further improves the network transmission performance.As an important part of the Internet of Things,vehicular network is a more reliable and safer road access scheme through comprehensive analysis of vehicle location,speed,driving direction and road pedestrian.But at the same time the application needs of the vehicular network is diversified,need to transmit daily entertainment,traffic control,security warnings and other different types of application data.However,because of the safety characteristics of the car network,security information such as safety warning and traffic control will have higher priority transmission needs than the daily entertainment information.Solving the above problems will be one of the key challenges facing future car networking.Multi-homed network is efficient to improve both robustness and scalability of VANETs(Vehicular Ad-hoc Networks)as well as to multiple the accesses to VANETs.MPTCP(Multipath TCP)is a concrete proposal for multipath transport based on multi-homed devices.It creates more than one TCP links for each TCP communication and,therefore,shortens the latency that a traditional TCP requires when switching between different TCP links.To meet the diversified QoS applications demanded and current actual traffic conditions by terminal units of a VANET,especially to ensure the timeliness and reliability when sending the high-priority messages that are related to safety-critical applications,this paper proposes a demand-driven MPTCP congestion control algorithm: PTLIA,which is based on message priority,throughput estimation and traffic density.First,message priority and throughput proportion factor are defined to characterize the weight of each message being sent.Secondly,the method of forward single-step prediction is used to evaluate the path state in MPTCP.Then,the algorithm of PTLIA is accordingly designed to adjust,dynamically,the window ofthe congestion time on demand.Finally,the algorithm of PTLIA,as proved in our simulations and experiments,has shortened the transmission of the high-priority data,under the premise of MPTCP transmission principles.(1)This thesis proposes a forward single-step linear prediction algorithm based on historical information.The algorithm predicts the next data value according to the data of the previous part,and adjusts the preliminary forecast value according to the historical data data,Get the final forecast.The algorithm has low computational complexity and has strong versatility,and can predict all kinds of discrete functions.At the same time,the throughput evaluation model is put forward according to the forecast results,and the application priorities of the vehicle network are divided in detail by using the dynamic priority in order to distinguish the priority of different applications.(2)This thesis presents an MPTCP congestion control algorithm based on application priority and throughput evaluation.The algorithm not only takes into account the transmission efficiency and fairness factors,but also takes into account the transmission requirements of high priority applications in the car network,and dynamically adjusts the network congestion window values on different applications,which further ensures the security of the network communication Throughput assessment of the way to improve the efficiency of network transmission,thereby enhancing the real-time communication and reliability of the car network.