| Recently, video delivery has become the main driver for the explosive growth of wireless traffic. In order to cope well with this trend, wireless video transmission technologies have been widely concerned. By adopting proper schemes, users ' quality of experience can be improved and the cost of wireless systems can be reduced. -Therefore, it is of high value for the research of wire-less video delivery, especially the optimization of wireless video transmission systems and networks.In this thesis, recent activities in real-time high-definition video transmis-sion, wireless video multicasting and video-friendly wireless networks are sur-veyed and analyzed. Then, the thesis focuses on 3 fields of wireless video tech-nologies: uncompressed video transmission, the delivery of video with linear distortion structure, and network algorithms for video users in cache-enabled HetNets. The research is mainly based on optimization theory. Theoretical analysis, problem formulation, algorithm design and simulation verification are all carried out for each work in the thesis.Firstly, a novel uncompressed video transmission framework is proposed,which incorporates adaptive Hadamard transform and transmits different bits with unequal power. Off-line power allocation algorithms are adopted in this framework, which is able to relieve the calculation burden considerably and reduce the end-to-end latency. There is no need to recalculate the powers allo-cated to different streams after building a lookup table. An accurate relationship between the MSE and the BER is established. The significance of bits is not subjectively determined as most UEP strategies did, but through a mathemat-ical analysis of different bits' contribution to the total MSE. An optimization problem is formulated in order to minimize the total MSE constrained by the total transmit power and the limited PAPR. By solving the problem, an effective UPA method with an adaptive Hadamard switch is proposed.Secondly, a generalized cross-layer design is provided to enable efficient communications for the family of video with linear distortion structure. A joint carrier matching and power allocation scheme is proposed which can minimize the total distortion. The optimality does not depend on any specific measures,and the scheme is general and applicable to any video applications with lin-ear distortion structure and non-increasing distortion functions. An enhanced analog video coding scheme (i.e., ECast) is proposed to demonstrate the effec-tiveness of the generalized scheme. ECast is not only an example application,but also a full solution which can cope well with the fluctuating channel con-ditions, the large feedback overhead from multiple users and the fairness of resource allocation. After that, several example applications are introduced to show the generality of the proposed scheme. The joint channel matching and power allocation scheme is applicable for all these applications.Thirdly, scheduling and association strategies for video users in heteroge-neous cache-enabled small cell networks are studied. First, a scenario is con-sidered that a group of multi-homing UEs are served by a set of cache-enabled SCBSs from different operators. A problem arises as how to allocate wireless resources to multi-homing UEs considering both the cache-enabled and multi-homing characteristics of network nodes. A distributed scheduling algorithm is proposed to minimize the total operation cost. The algorithm is based on al-ternating direction method of multipliers (ADMM) and converges to the global optimum after a few iterations. Second, by modeling the locations of SCBSs in each tier and UEs as independent homogeneous Poisson point processes, prob-abilistic analytical approaches of user association are investigated. The average utility of a typical user is derived considering both the limited caching capacity and the backhaul bandwidth in each SCBS. An optimal association strategy is proposed in terms of maximizing the average user utility. |
PDF Full Text Request