Efficient Content Dissemination Mechanisms And Strategies For Mobile Communication Networks

With the development of mobile communication technologies,the demand for mobile internet applications and high-speed multimedia services of smart-phones and tablets users over mobile networks has been growing tremendously in recent years.It will lead to an explosion of mobile traffic and pose an enormous challenge to the fifth generation mobile communication system(5G).From the study on the traffic explosion problem,a major portion of mobile traffic is due to duplicate downloads of a few popular content items from remote servers.Thus content caching near users is the most promising method to cope with the exponentially increasing traffic over mobile networks.By using mobile content caching and delivery techniques,popular content items are cached in the intermediate or local servers so that the user demands for the same content can be fulfilled readily without duplicate transmissions from remote servers,so as to save transmission resources.Traditional content distribution networks(CDN)have been well investigated in the context of wired networks,such as the Internet.Unfortunately,traditional CDN-based content distribution techniques cannot be simply applicable to mobile networks since the mobile network architecture is different and the network resources,including wireless link capacity,mobile backhaul,and storage capacity,are constrained in mobile networks.Moreover,the hit rate of the cached content items could be rather low in mobile networks due to the content dynamics,user mobility and limited number of users in a cell.On the other hand,the amount of contents provided by content providers is growing rapidly and it is thus impossible to cache all content items,although storage is becoming much cheaper.Therefore,it is imperative to develop efficient content caching and delivery strategies to maximize the benefits of local content caching for mobile cellular networks.It is well-known that heterogeneous cellular network(HetNet)has been deemed as a promising architectural technique in 5G.In HetNets,small cells such as femtocells and device-to-device(D2D)communications are introduced,with the aim to greatly mitigate the pressure on wireless link capacity.Intuitively,caching popular content items at local femtocells can greatly reduce the downloading delay of users with low-capacity backhaul links from femto base stations(FBSs)to macro base stations(MBSs).Furthermore,with the D2 D communication technique proposed by 3GPP 4G LTE-Advanced,caching at user equipment(UE)and delivering the cached content items via D2 D links can further reduce BS transmissions,alleviate user perceived latency and improve user quality of experience(QoE).On the other hand,the emerging mobile edge computing(MEC)architectural technology is currently being standardized by the European Telecommunications Standards Institute.MEC architecture provides a highly computation and storage capacity within the radio access network(RAN)that can be used to deploy applications and services as well as to cache popularity content items in close proximity to mobile subscribers.Since machine learning algorithms can be used to learn users' content demands dynamically and optimization algorithms have the advantage on solving complex optimization problems,this dissertation is dedicated to the content caching and distribution problem for mobile communication networks,by expoiting machine learning and optimization techniques.The main research content of this dissertation include the following four parts:(1)optimal cooperative content caching and delivery policy for heterogeneous cellular networks;(2)multi-agent reinforcement learning for efficient content caching in mobile D2 D networks;(3)machine learning based cooperative content caching for mobile edge networks;and(4)an actor-critic learning based proactive caching policy for mobile edge cloud.First,this dissertation studies the optimal cooperative content caching and delivery policy for heterogeneous cellular networks.In this research,we develop an optimal cooperative content caching and delivery policy,where FBSs and UEs are involved in local content caching.We formulate the cooperative content caching problem as a linear programming problem.We use hierarchical primal-dual decomposition method to decouple the problem into two level optimizations,which are solved by using the subgradient method.Furthermore,we design an optimal content delivery policy,which is formulated as an unbalanced assignment problem and solved by using Hungarian algorithm.Numerical results show that the proposed cooperative content caching and delivery policy can significantly improve content delivery performance in comparison with known existing caching strategies.Second,this dissertation investigates multi-agent reinforcement learning for efficient content caching in mobile D2 D networks.To address the increase of multimedia traffic dominated by streaming videos,UEs can collaboratively cache and share contents to alleviate the burden of base-stations.In this research,we design D2 D caching strategies using multi-agent reinforcement learning(MARL).Specifically,we model the D2 D caching problem as a multi-agent multi-armed bandit problem.Since the joint action space is too great to use traditional MARL methods,a belief-based modified combinatorial upper confidence bound(MCUCB)algorithm is proposed to technically solve the problem.Simulation results show that the belief-based MCUCB caching scheme outperforms other popular caching schemes in terms of cache byte hit rate and average downloading latency.Next,this dissertation studies machine learning based cooperative content caching for mobile edge networks.To address the drastic increase of multimedia traffic dominated by streaming videos,mobile edge computing can be exploited for reducing redundant data transmissions and improving content delivery performance.Under the MEC architecture,content providers(CPs)can access MEC servers to deploy popular contents to improve users' quality of experience.Designing an efficient caching policy is crucial for CPs due to the content dynamics,unknown spatial-temporal traffic demands and limited cache capacity.In this research,we propose a learning based cooperative caching policy for the MEC architecture,when the user preferences are unknown and only the historical content demands can be observed.We model the cooperative caching problem as a multi-agent multi-armed bandit problem and propose a MARL-based algorithm to solve the problem.Simulation experiments are conducted based on real dataset and numerical results show that the proposed MARL-based caching policy can significantly improve content cache hit rate and reduce downloading latency in comparison with popular caching strategies.Finally,this dissertation investigates an actor-critic learning based proactive caching policy for mobile edge cloud.Mobile edge cloud is implemented based on a virtualized platform that leverages recent advancements in network functions virtualization(NFV).Specifically,NFV enables a single mobile edge cloud to create multiple virtual machines(VMs)to provide elastic resource allocation for CPs.The scalable resource model of the cloud gives rise to the challenge of determining the ideal cache size for the CP to obtain maximum benefit at the lowest cost for resource use.In this research,we design a proactive caching policy for mobile edge cloud.We vary the leased number of VMs to match up with the dynamical users' content demand to minimize the storage cost,while maximize users' QoE.We formulate the traffic load of mobile edge cloud caused by users' content request as a Markov decision process.Afterwards,in order to foresightedly minimize the caching cost of CPs and the expected delivery latency of users,we design a reinforcement actor-critic learning framework based caching scheme.Finally,the simulation results verify the effectiveness of proposed caching policy.