Congestion Control Algorithms For Multimedia Transmission In3G Networks

The widespread use of3G technologies and the beginning of the commercial deployment ofLTE have enhance the performance of mobile networks. Users prefer surfing on the Internet vialaptops or mobile phones. Due to the explosion of network services and user number, the networksare getting more complicated and the problem of congestion is getting worse. As a result, the issueof congestion control has drawn the attention of the research community in recent years.In allusion to the scenario of multimedia transmission in3G networks, this thesis focuses ontwo main solutions to congestion,1) active queue management and2) transmission control protocol.The main contributions of this thesis are listed as follows:The existing queue model with feedback rate-control mechanism sets a strict limitation inservice rate and fails to take feedback delay into consideration. This thesis conducts a deeper studyon this queue model by using discrete time Markov theory. The model parameters, such asmaximum queue length, service rate, feedback delay and rate-control mechanism are taken intoconsideration. The influence of model parameters on system performance (including average queuelength, queue length variation, overflow and downflow probability) is studied by simulation.The existing active queue management algorithms proposed to solve the delay bloat problemlack effectiveness and robustness when encountering non-TCP flows and are sensitive to parameterssettings. This thesis proposes a novel traffic shaping algorithm that the input traffic distribution canbe shaped into the predefined distribution. On the basis of the traffic shaping algorithm and theprementioned queue model, an active queue management algorithm ITDS (Input TrafficDistribution Shaping) is proposed. The performance of ITDS can be analysed by queue theory. Theperformances, including delay, packet drop rate and throughput are verified by NS2simulationunder different service scenarios. The simulation results show that ITDS has lower and more stabledelay, less packet drops and higher throughput than existing technology.Most existing TFRC-based algorithms designed for UMTS networks cannot achieve highperformance in loss type differentiation or bandwidth usage rate. This thesis proposes a novelcross-layer streaming transmission control algorithm WATC (Wireless-loss Aware TransmissionControl) which makes use of the ARQ information to estimate the quality of wireless channels andimprove the measurement of delay with the help of data link layer. WATC distinguishes wirelessloss from congestion loss by the estimation of the quality of wireless channel and adjusts the sending rate according to the quality of wireless links and the tendency of delay variation.Simulation results show that WATC outperforms existing TFRC-like schemes in the terms of lossrate control and throughput, while maintaining high throughput and fairness in multi-flow scenarios.