Research On The Key Technology For Concurrent Multipath Transfer In Heterogeneous Networks

In heterogeneous networks,concurrent multipath transfer(CMT)utilizes multiple networks for transmission simultaneously,which can aggragate the transmission capability of multiple networks,improve the throughput and achieve load balancing effectively.It is one of the important methods to meet the de-mands for network resources and to improve the resource utilization efficiency.The solutions at the transport layer can take advantage of each layer's charac-teristics and achieve flow control,congestion control and friendliness to other flows at the same time.These solutions play important roles in current and future network architectures.In the thesis,we focus on CMT based on multi-path transfer protocols at the transport layer.A bandwidth scheduling scheme,a congestion control scheme and a handover optimization scheme are proposed to improve the spectral efficiency,the energy efficieny of terminals and the transmission performance of mobile users,respectively.Although the proposed schemes are based on Multipath TCP(MPTCP),the ideas can be also applied for other multipath transfer protocols.The main contributions of this thesis are summarized as follows.(1)To improve the spectral efficiency of the MPTCP connection,a novel bandwidth scheduling scheme is proposed.Firstly,analyze the influence of end-to-end link parameters on the transfer performance of an MPTCP con-nection by simulations.The simulation results show that continuous increase in bandwidth on physical layer cannot lead to significantly higher throughput when the value of bandwidth exceeds a certain value.Secondly,use validity bandwidth to indicate the required bandwidth resources of each subflow in the MPTCP connection for achievable throughput.According to Mathis Model,this term can be formulated as a function of delay and path loss rate.To correct the error caused by the time-varying and diverse wireless channels,Kalman Fil-ter is adopted for more accurate estimation of round trip time and path loss rate.Finally,updates the occupied wireless spectral r-esources of a subflow according to the obtained value of validity bandwidth.The simulation results validate the effectiveness of the proposed scheme which improves the spectral efficiency greatly with minor decrease in throughput.(2)To improve the energy efficiency of mobile multi-mode terminals,a congestion control scheme for MPTCP is proposed.Firstly,define energy ef-ficiency as the number of bits received per joule energy,the ratio of total data size to the sum energy consumption.With the constrains of sum transmission rate,path capacity and available power,the terminal energy efficiency maxi-mization problem is formulated as a constrained optimization problem.In the problem,the variables to be solved are the values of receiving rate at each net-work interface.Secondly,genetic algorithm is adopted to obtain a near-optimal solution for the problem and the obtained rate terms are normalized as a traffic distribution vector for output.Finally,the acquired vector is applied into the window-based congestion control algorithm by means of fluid model to adjust the increment of each subflow in congestion avoidance phase for each ACK re-ceipt.Simulation results demonstrate the effectiveness of the proposed scheme which shifts part of traffic from the higher energy consumption paths to the lower ones and improves the energy efficiency of multi-mode terminals.(3)To improve the transmission performance of mobile users,a handover optimization scheme is proposed.Firstly,schedule the packets proactively ac-cording to the user's mobility pattern and the position of the access point of the current connected network.When the mobile user moves to the edge of the current connected network,the time for getting out of the coverage can be es-timated.If the duration is not more than the path's round trip time,setting the path via this network as a candidate path over which no more data is allowed to be transmitted.Secondly,select the best network for handover in multiple candidate networks according to the end-to-end link information of paths via candidate networks and the moving pattern of the mobile user.In the selection process,path quality and dwelling time are taken into account jointly by the sim-ple additive weighting method and the best one is selected out as the handover network.Finally,the handover optimization scheme is proposed based on the above schemes.Through the process for signaling exchanging,remove the old subflow and start a new subflow with the new IP address provided by the han-dover network.Simulation results evaluate the effectiveness of the proposed scheme which can achieve relative higher throughput with less handovers,thus improving the transmission performance of mobile users when making use of MPTCP based CMT for traffic transfer.