Research On Cross-layer Resource Scheduling Of Video Applications Over LTE Systems

With fast development of wireless communication technologies and constantly upgrade of mobile terminal devices, more and more video applications come to our daily life and they are envisioned to be primary applications in LTE systems. However, there are great contradictions between growing demands for high quality wireless video applications and scarcity of wireless resources and variable of wireless channels, which have brought challenges to development of wireless video applications. Meanwhile, the increase of user numbers and service requirements would raise video coding power and video transmission power of base station systems. It is estimated that the carbon dioxide emission due to wireless applications will further aggravate the greenhouse effect. The issue of reducing power consumption of wireless video applications has raised new demands for next generation wireless communication systems. In order to solve these challenges, cross-layer resource scheduling technologies have been employed to break the independence of different protocol layers and restrain the negative impact brought by time-varying wireless channels. Besides, through joint optimization of parameters from the application layer, the media access layer and the physical layer, cross-layer design resource scheduling scheme could ensure QoS of video applications as well as improve system performance.In this paper, parameters from different layers were adjusted dynamically to achieve the goal of guaranteeing quality of video applications as well as improving system performance, and efficient cross-layer resource scheduling schemes were designed. The main work of this paper includes following aspects. Based on the presented PSNR-MOS mapping model, we proposed a QoE-driven cross-layer design for video applications over LTE systems. The subjective perception of users is the determining factor of success of video applications. Therefore, QoE is much more suitable to be used as measurement criteria than QoS in terms of multimedia applications. First, based on a hyperbolic tangent function, we proposed a novel mapping model between PSNR and MOS. Through jointly optimizing parameters from APP layer, MAC layer and PHY layer, the model could reflect the relation between objective system parameters and subjective perceived quality simply and precisely. Second, on the basis of PSNR-MOS mapping model, we established utility function which is aimed at maximizing overall QoE and ensuring fairness among users. Third, PSO algorithm was adopted to solve the resource allocation problem to reduce computation complexity and guarantee optimal scheduling policy.In order to solve the problem of increasing system power consumption brought by rapid development of wireless video applications, we proposed a green resource scheduling algorithm for video applications over LTE systems. According to the relationship between video distortion and power consumption, we established a cross-layer power-application and system parameters-distortion model in this paper. The model revealed the relation among power consumption, distortion, application and system parameters in terms of video applications. Based on this model, utility function was proposed to achieve minimum power consumption with constraint of video distortion. Through dynamically adjusting values of parameters such as video coding bit rate and MCS mode, the target of power consumption reduction could be reached.We modeled and implemented the simulation platform for LTE cross-layer resource scheduling. In terms of modular design, the platform was divided into four modules, application level simulation module, system level simulation module, link level simulation module and cross-layer optimizer. With this platform, novel technologies and algorithms in research could be estimated quickly and intuitively, which shortened the development cycle of new technologies, and cut down the development cost.