Research On Distortion Estimation And Power Rate Distortion Optimization For Mobile Multimedia Communications

Due to the rapid development of wireless communication, the enhancement of the wireless channel bandwidth, the advances of audio and video coding techniques, and the popularity of mobile devices, the demand for mobile multimedia applications, in which image, video, voice and data are mixed are increasing. However, mobile devices rely on limited energy supply by batteries and frequent recharging is not a pratical option. Therefore, in order to design efficeient wireless multimedia communication systems, studing how to obtain the best reconstructed quality of multimedia data under the power consumption constraint or minimize the total system power consumption for given the expected reconstructed quality of multimedia data at the receiver becomes very crucial.This thesis aims at optimal system resource allocation and studies joint power and rate-distortion optimization and distortion estimation in multimedia communication system over wireless channel. The reseach contents in this thesis are as follows.We firstly analyze the effects of bit errors among video bitstream on the reconstructed video quality at the receiver, and propose an end-to-end distortion estimation method for video communication system over wireless channel based on considering the distortion caused by source quantization, channel errors, and error concealment. The method that can estimate the reconstructed video quanlity at the receiver recursively from one frame to another according to average bit error rates of channel and bitstream information extracted during coding process. Simulation results show that the averge relative estimation errors are less than 8% and the average absolute estimation errors are within 0.9 dB for most video sequences under the various bit error rates of wireless channel.In order to provide effective transmission distortion estimation method for video communication system over wireless internet, we firstly study the modes of errors occurring in video compressed bitstream due to packe loss in wired networks and bit errors in wireless link. Then we add packet loss in wired networks to the distortion estimation only based on bit errors in wireless channel which has been studied and present a transmission distortion estimation method for video delivery over hybrid channels with bit errors and packet erasures, in which general video packet scheme is integrated. Simulation results indicate that by using the proposed method the average relative estimation errors are less than 5% and the average absolute estimation errors are within 1 dB for most video squeneces under the constant channel conditions. On the other hand, as the delayed time of channel condition information feedbacked from receiver becomes larger, the estimation errors maintain very low under the varying channel conditions.In order to solve the joint power and rate-distortion optimization problems with lower computational complexity, we firstly analyze the optimization problem of total system energy consumption for end-to-end image communication system over wireless channel. And the we propose a fast implementation algorithm for optimization, in which the relations among the system reconfigurable parameters, the optimization object and system constraints are fully utilized. The presented algorithm can reduce the search scope of optimal results and does not rely on the initial values. Simulation results reveal that the presented algorithm can provide effective optimal results and reduce computational operations by the ratio of 87.1% in average compared with the SQP method, which is one of most effective algorithms to solve nonlinear optimaiztion problem with constraints.To optimize the system resource allocation by using various sensitivities of multimedia sources to transmission errors, we study how to combime unequal error protection with unequal transmission power distribution for end-to-end image communication system over wireless channel. Then we propose two resource allocation system schemes based on jont power and rate-distortion optimization. Specially, we firstly present a resource allocation scheme based on the distortion minimization of the region of interest (ROI) in images, in which the transmission protection levels for ROI in images are adjusted in terms of varying channel conditions while maintaining the system energy consumption and total transmission rate constant. Simulation results demonstrate that our proposed ROI distortion-minimized resource allocation scheme can improve the reconstructed quality of ROI in images significantly compared with the equal error protection optimization scheme. On the other hand, we present an end-to-end system energy-minimized resource allocation scheme, in which transmission protection are adaptively assigned in terms of different sensitivities to transmission errors of various compressed bitstream. Simulation results show that applying our proposed scheme to code and transmit different images under the various channel conditions can save total system energy consumption by 20.8% in average compared with the equal error protection optimization scheme.