Edge Caching Based Video Transmission Optimization In Heterogenous Networks

In recent years,with the rapid growth of mobile users and a wide variety of mobile devices,wireless communication traffic trends to increase explosively.Among these traffic,mobile video traffic takes up the largest share,which has put great pressure on the existing mobile networks.Heterogeneous networks(HetNets),however,provide some favorable conditions to video transmissions.In HetNets,since small-cell base stations(SBSs)are introduced and supposed to be densely deployed,the distances between users and base stations are greatly reduced.As thus,users can achieve higher downlink data rate to receive high bitrate videos.Nevertheless,there are still(even more serious)backhaul bandwidth bottlenecks in HetNets.This is because these numerous deployed SBSs need much more backhaul bandwidth to transfer data back and forth from the core network.Besides,for the sake of deployment flexibility and economy,wireless backhaul techniques with low bandwidth are adopted widely in practice.Since then,the high rate fronthaul link and the limited backhaul capacity becomes a contradiction,which in return,restricts the improvement on user's video experience.Edge caching is a promising way to solve the problem mentioned above.It refers to the paradigm of equipping SBSs with a certain amount of caching space to prefetch and store some contents.By doing so,these SBSs are empowered the ability to respond to some of the requests without fetching contents through backhaul links.Hence the limited backhaul bandwidth can be saved.The highlight in the field of edge caching is to design appropriate caching strategies(i.e.,to determine which files to be cached in which SBSs).In general,the more appropriate the designed strategies are,the higher cache hit ratio we will achieve.In this case,the limited backhaul links are more un-likely to become bottlenecks,then more users are guaranteed to obtain the contents they want and achieve better experiences.Note that,in this thesis,we consider the video caching scenario,which is greatly different from the file caching scenario.The chal-lenges brought by video caching are:(a)Since videos are always much bigger than reg-ular files,and there may be some structural characteristics available(e.g.SVC videos),it's more challenge for us to fully utilize the limited caching space.(b)Video services usually have certain requirements on users' downlink data rate(i.e.,QoS requirements),and users' actual experiences are nonlinearly related to many factors.In consideration of this,the designed caching strategies may be much more complicated.To meet the aforesaid challenges,in this thesis we carried out the following two researches on video caching in heterogeneous networks.Some interesting conclusions are drawn through numerical simulations.(1)For SVC videos,we take advantage of its distinct layered feature to design joint power allocation and caching strategies.The reason why we do so is that,in the scenario where SBS's transmit power is limited,its power allocation strategy will affect users'video reception capability.Therefore,we need a caching strategy that takes the power allocation results into account to avoid caching those unreceivable video layers,so that the cache hit ratio can be improved.To tackle this issue,we first formulated the joint power allocation and caching problem as a mixed integer programming problem.Then we proposed a simulated annealing based heuristic algorithm to solve it.In order to discuss the necessity of joint optimization and conduct analysis on the optimality of the proposed heuristic algorithm,we adopted stochastic geometry as a handy tool to derive and optimize the successful transmission probability of different quality versions of videos.Last,extensive numerical simulations are carried out.Results obtained in these simulations verified that our proposed method can indeed improve the cache hit ratio and improve user's average QoE.(2)Consider that in HetNets,because SBSs are densely deployed,users are more likely to be covered by multiple SBSs.Therefore,when designing video caching strate-gies,we can further take the user's association into account.Specifically,in this re-search,we assume that a user can be dispatched to some other SBS to enjoy the video he requested.By doing so,the power,the backhaul bandwidth,and most importantly,the cache space of different SBSs can be efficiently utilized in a cooperative manner,so that user's QoE can be improved.To study the problem,we first introduce the software-defined wireless network architecture and formulated the joint user association,power allocation and caching optimization problem as a mixed integer programming problem.Then,noticing its discreteness and monotonicity,we adopted the DBRB algorithm in the context of monotonic optimization.We further customized the DBRB algorithm based on the feature of the formulated problem to improve its efficiency.Last but not least,numerical simulations are carried out to prove the convergence and effectiveness of the proposed algorithm.