Research On Energy Efficiency Oriented QoE Enhancement In Wireless Networks

With the development of wireless telecommunication, new types of ser-vices are coming out with higher and higher quality demands. Traditional qual-ity of service(QoS) based service guarantee, due to not being able to reflect user’s personal satisfaction, is no longer suitable for being the optimizing crite-rion any more. Due to user-centric, the enhancement of user’s quality of experi-ence(QoE) has become the new object of telecommunication system designing and optimization. Wireless networks are going through a revolution from QoS guarantee to QoE enhancement. For the future services with high data-rate and complex computation, on one hand, it is hard to automatic retrieve user’s per-sonal perception from the network in order to satisfy the user’s expectation, on the other hand, the guarantee of large number of users’ QoE expectation would consume large amount of energy. Reducing the energy consumption in wireless networks, as well as providing user satisfied services, are the important aspects of the sustainable development of ICT industry.In order to face the challenges of QoE enhancement and energy consump-tion which the ever-increasing traffic would bring, this thesis will focus on the optimization of QoE and energy efficiency in wireless networks, including 4 aspects: 1. Video quality evaluation in wireless networks. 2. QoE coverage analysis in heterogeneous networks. 3. Joint optimization of QoE and energy efficiency in OFDMA networks. 4. Energy efficient task offloading in mobile cloud computing with wireless power transmission. Details about the thesis are illustrated as follows:1. Video quality evaluation in wireless networks.Video quality assessment is the efficient method to obtain the users’ QoE.Based on the characteristics of human visual system(HVS), this thesis explores the visual saliency extraction and objective quality assessment in spatial domain and temporal domain respectively. Firstly, this thesis proposes an adaptive vi-sual saliency extraction method based on HVS characteristics. Compared with the state-of-the-art methods, the proposed method performs better in accuracy and adaptivity. By decomposing the original image into multiple channels with Hilbert-Huang transform, the visual saliency figure is achieved by pooling these channels with information entropy criterion. Experiments show that on datasets with different background textures, the method performs better than the state-of-the-art models. Secondly, this thesis proposes an objective video quality assessment methods based on temporal reduced frame set. Compared with the state-of-the-art methods, the proposed method greatly reduce the computated frames. By analyzing the temporal frame quality of distorted video, a method to extract the quality-related frame is proposed. The overall video quality can be predicted by processing these frames, and they are important to monitor the quality variations during the video services. Combined with spatial quality as-sessment, a video quality method is proposed based on the reduced frame set.Evidence shows that the method can achieve a comparable performance while reducing about 1/2-2/3 of the total frames.2. QoE enhancement based on service and network configurationService and network configration have great impact on the users’ QoE.State-of-the-art, seldom work have analyzed the influence of the configura-tion parameters. For the first time, by incorporating the stochastic geometry,the QoE coverage concept of heterogeneous network is proposed in this the-sis. The QoE coverage is defined on the probability distribution that the user’s quality expectation is satisfied. It takes user, network and service parameters into consideration. By performing the subjective quality tests, we rectify the video quality prediction model and derive the estimation equation of the video QoE coverage in multi-tier heterogeneous networks. Results show that the QoE coverage is highly correlated with network and environment factors, such as the number of tiers, density of base station, path loss; networks configurations, such as transmit power, MCS scheme of base stations; and service-related factors,such as the model parameters defined by the video. As a criterion reflecting comprehensive network ability to bear video services, QoE coverage provides a way to analyze the tradeoffs among multiple factors. Besides, it shows the transmit power of base station is highly related to user satisfaction, which pro-vides a theoretic indication to jointly optimize QoE and energy efficiency.3. Energy Efficient QoE enhancement based on wireless resource alloca-tionWireless resource allocation plays an important role in improving the en-ergy efficiency and enhancing the users’ QoE. However, traditional wireless resource allocation methods in OFDMA network, typically are based on QoS guarantee, which is lack of consideration of user satisfaction. In this thesis, a new energy efficiency, defined as the ratio of overall satisfaction of all users to the total consumed power, is proposed. The new criterion takes the char-acteristic of services, multiuser QoE fairness and related power consumption into consideration. By optimizing the new energy efficiency, both the QoE and energy efficiency can be improved. Due to the non-convexity of the model,it’s hard to find an analytic solution. By transforming the original model into sub-channel allocation and power allocation sub-problems, a suboptimum so-lution is achieved. Evidence shows that the proposed method not only achieves a better systematic utility, but also convergences in limited loops.4. Energy Efficient QoE enhancement based on computation resource dis-patchComputation resource is of great importance in the users’ QoE and sys-tematic energy consumption. Currently study on computation resource dispatch mainly focuses on uplink task offloading. This thesis considers both the uplink information transmission and downlink energy transmission. By desining the uplink-downlink division duplex scheme, the model takes the downlink wire-less power transmitting, local computing and remote task offloading into con-sideration. The throughput is maximized by transforming the original problem into 3 sub-problems: the downlink optimal energy transmitting, optimal of-floading scheme, dynamic time division duplex factors searching. Simulation results show that in different channel fading environments, with different num-ber of mobile terminals in the area, the offloading scheme performs better than state-of-the-art methods.