Research On Video Transmission In Wireless Heterogeneous Networks

Thanks to the development in wireless technology and the proliferation of smart terminals,wireless video applications,including the emerging virtual reality(VR)and argument reality(AR),have experienced continuous but dramatic growth.This results in exponential increase in the video data volume,which puts further pressure on the already limited network bandwidth,and leads to congestion and low transmission efficiency.It is thus necessary to establish new wireless media networks aimed at improving the users' quality of experience(Qo E)for video applications.Among the identified key techniques,the forward error correction(FEC)and resource allocation under heterogeneous networks have gained a lot of attentions in both academia and industry.In this thesis,we carry out our study in the following two aspects:1.Based on fountain code,a new FEC technique is proposed to address the high packet loss rate in wireless channels and improve the decoding efficiency.An unequal error protection scheme is introduced to achieve stable transmissions.2.Wireless channels are bandwidth-limited.We design efficient resource allocation algorithms for heterogeneous networks.In particular,V2 X scenarios with V2 V communication are taken into account and filed of view(Fo V)oriented immersive media transmission is also investigated.The main contributions of the dissertation are as follows:· New FEC methods based on fountain codes are proposed.Since the decoding efficiency decreases with the decoding process in fountain codes,the relationship between the decoding efficiency and the degree distribution together with decoding status is investigated.Bsed on the optimal degree distribution for different numbers of decoding symbols,we design a new fountain code with limited feedback which can improve performance dramatically.In order to adapt to the multicast channels,a fountain code named unequal fountain code is designed.It contains information regarding the level of protection via virtually copying and measuring the degree distribution.The experiments show that with the proposed unequal foundation code,users subject to poor channel conditions can successfully decode the basic information while users with better channels can obtain more information and enjoy better user Qo E.Besides,a two-layer Qo E model for wireless video streaming with fountain codes is proposed.The first layer relates the coding redundancy to three Qo E factors,while the second layer establishes the functional relationship between Qo E factors and user Qo E evaluated using the mean opinion score(MOS).Experiments show that applying the new Qo E model to video on demand(Vo D)systems can improve the total Qo E.The analysis and proposed schemes are helpful to wireless video transmission,and the corresponding code has been adopted by the standard organization MMT.· Resource allocation algorithms for LTE/WLAN heterogeneous networks are proposed.Considering that users can gain access to both LTE network and WLAN network with multiple Wi-Fi access points(APs),we formulate an application-layer scheduling problem for real-time video transmissions over multi-RAT systems with multiple users.It is shown to be a NP-hard problem.We propose a greedy algorithm and a penalty function-based algorithm to reduce the complexity in finding a solution.To be specific,the penalty function-based algorithm adds a regularization term to the objective function and makes the problem convex.It has a low complexity and can get the results more quickly in mobile scenarios.Besides,in most cases,the final results are exactly the globally optimal solutions.Comparing with the current studies,our work considers a more general scenario.It has potentials for being applied into wireless video transmission.Besides,in order to address the Qo E bottleneck due to the increasing number of cars,we propose the system that utilizes V2 V communication to help V2 I communication offloading and develop associated V2 X access control and resource allocation algorithms.Since the connections and rate allocation in V2 V and V2 I communication are coupled,the problem is NP-hard.We propose a heuristic algorithm to solve the problem.It first finds out the cars that degrade the Qo E in V2 I system and should switch to V2 V mode.It then converts the V2 V communication into V2 I communication.Finally,it finds a suboptimal solution with low complexity.The simulations show that the proposed algorithm can improve the quality of video transmission in V2 X system greatly.· Resource allocation algorithms and buffer management schemes for 360-degree/VR video transmissions over wireless heterogeneous networks are proposed.We jointly consider saliency,Fo V prediction and projection distortion,and propose the Qo E model for 360-degree videos.Then we formulate a tile-based 360-degree video transmission problem for wireless heterogeneous networks.More specifically,we consider the channel difference of users,video contents and users' behavior,then develop algorithms to decide which user should be connected to a certain Wi-Fi AP,and choose appropriate transmission rates for each tile of each video,so that the overall system utility can be maximized.Additionally,the traditional buffering scheme for common videos can't be applied to 360-dgeree video transmission since the Fo V prediction is no longer accurate in the long buffer.In this paper,we design a new buffering scheme for 360-degree video transmission.In buffer rate management,we take the throughput of network and current buffer status into consideration,and propose a new rate control scheme with variable buffer length to reduce the initial delay and promote the transmission efficiency.We propose a hierarchical buffering scheme which can achieve a good balance between downloading the new tiles and updating the existed ones with lower representation.The simulation shows that the proposed algorithms can achieve good Qo E.Besides,the new scheme can maintain sufficient content in tiles and provide high Qo E,which outperforms existing methods and can provide efficient supports for immersive media transmission.