Research On Caching Strategy And Interference Management In Cache-Enabled Wireless Networks

With the growing popularity of smart mobile terminals and the rise of smart services,such as the virtual reality,augmented reality,and autonomous driving,mobile data traffic has been exploding exponentially.Content-centric services(e.g.,multimedia video transmission)account for the majority of the network traffic.The content transmitted by this type of service usually has a large volume and can be repeatedly requested by different users,which results in a large amount of redundant traffic over the backhaul links.To alleviate the backhaul load,an effective solution is to cache popular files at base stations(BSs),forming a new paradigm of cache-enabled wireless networks(CEWNs).In CEWNs,the repeated request for the same content from the core network is avoided,so that the backhaul load can be significantly alleviated.However,because of the limited cache size at BSs,it is unable to store all the files requested by users in the network.Therefore,the serving BS of a user may not be geographically its nearest BS,which may lead to strong interference at the user.Moreover,the densification of networks also makes co-channel interference a growing problem.To this end,this thesis considers the joint design of cache placement policies and interference management techniques in CEWNs,and the corresponding strategies for single-antenna and multi-antenna networks are respectively investigated.Firstly,the random caching policy design with intra-tier BS cooperation in a singleantenna CEWN is studied.Consider a single-tier CEWN,where a single antenna is equipped at each BS and user.The random caching policy is adopted and a local channel state information based joint transmission(LC-JT)scheme is proposed to improve the desired signal strength received by users.A tractable approximation expression for the successful transmission probability(STP)is derived.Next,with the goal of maximizing the network STP,the cache placement probability for each file is optimized.A globally optimal solution in a special case,as well as a locally optimal solution in general case,are obtained.Furthermore,considering the non-coherent joint transmission(NC-JT)scheme as a comparison benchmark,the proposed caching policy is compared with three baseline policies under both the LC-JT and NC-JT schemes.Simulation results reflect the performance advantages of the proposed policy.Secondly,the random caching policy design with cross-tier BS cooperation in a multi-tier heterogeneous cache-enabled network is studied.Under the scenario where all the BSs in each tier are equipped with cache,a distance-minimum cross-tier cooperation transmission scheme is proposed,and a gradient projection method based algorithm is designed to effectively simplify the complexity of the optimized caching policy design.Specifically,an analytical expression for the STP in general case is derived.Considering the scenario where there are two tiers of BSs in the network,the cache placement probability for each tier is optimized to maximize the STP.An gradient projection method based algorithm is designed to simplify the complexity of solving the optimization problem,and a sub-optimal caching policy is obtained.Simulation results show that the proposed algorithm can significantly decrease the time cost for problem solving.In addition,under the scenario where only small BSs are equipped with cache and macro BSs are only connected to the core networks via backhaul links without cache,a location-aware cross-tier cooperation transmission scheme is also studied.The corresponding expression for the STP is also derived,and a locally optimal caching policy is obtained numerically.Finally,via numerical simulations,the correctness of the theoretical analysis is verified,and the performance advantages of the obtained caching policies are also reflected.Then,the random caching design for multi-user multi-antenna heterogeneous networks with interference nulling(IN)is studied.In a two-tier cache-enabled multi-user multi-antenna network,a user-centric IN scheme is considered to suppress the interference caused by the small base stations to users.Using the stochastic geometry techniques,a tractable expression for the area spectral efficiency(ASE)is derived,and a lower bound with simple form on it is also obtained.By optimizing the cache placement policy and IN coefficient,the lower bound on the ASE is maximized.The resultant problem is a mixed-integer programming problem,by exploring some properties of which,the problem is divided into two sub-problems,i.e.,cache placement optimization and IN coefficient optimization.An alternating optimization algorithm is designed to solve the original problem.Finally,the performance loss of the obtained solution compared with the optimal one is verified by numerical simulations,and several baseline policies are considered for performance comparison.Finally,this thesis focuses on the further fine-grained research on the performance of the cache-enabled multi-antenna wireless networks,proposes a multi-objective caching policy analysis and optimization model,and studies the trade-off between different performance metrics.First of all,a tractable expression for the average STP(ASTP)of the single-tier multi-antenna network with IN scheme is obtained,by using which the effects of the caching parameter and the IN parameters on the ASTP of the network are effectively analyzed.Then,a tight upper bound of the user link STP(LSTP)is derived,and approximate expressions for the LSTP variance and the signal-to-interference ratio(SIR)meta distribution are further obtained.These two approximated expressions can be used to study the effects of the caching parameter and the IN parameters on the LSTP distribution.Finally,based on the obtained theoretical analysis framework and numerical simulations,some system design insights are obtained for three different system design purposes.Moreover,two multi-objective optimization problems are considered to achieve the trade-off between different system design purposes,and the corresponding caching parameter and IN parameters are also designed.