| The communication industry has witnessed rapid growth,leading to a significant surge in the volume of data and the number of devices connected to mobile wireless networks.This expansion has imposed considerable strain on cellular networks.Authoritative statistical report indicates that most amount of network data traffic is attributable to a relatively small number of popular contents.Such popular contents are frequently requested by various users.By leveraging edge caching technology to store these popular contents in network infrastructure or at edge devices,users can directly download content from the nearest caching node,potentially alleviating core network congestion and minimizing service delay.Moreover,device-to-device(D2D)communication enables proximal user devices(UDs)to communicate directly,bypassing the base station.This capability ensures that content cached on a user’s device benefits not only the individual user but can also be shared with adjacent users via D2 D communication.This dissemination model can further mitigate network traffic,thereby enhancing communication efficiency and user experience.In the existing research,the interference factors of D2 D link in the caching strategy research are not fully considered.Most previous studies only considered the interference between D2 D communication links,ignoring the interference of cellular communication due to the multiplexing of cellular spectrum resources.The interference environment affects the successful delivery of caching content,which affects the design of caching strategy.At the same time,existing researche lacks a caching strategy that can guarantee the delay needs.Heterogeneous base stations are widely deployed in urban environment,but the existing research lacks the collaborative caching strategy of UD and multiple heterogeneous network(Het Net)base stations.Due to the difference of the quality of service(Qo S)and energy consumption of different kinds of caching nodes when providing service,how to optimize the offloading proportion of different kinds of nodes while ensuring the total offloading rate in the design of caching strategy is also an unsolved problem.For emergency communication or scenes with local enhancement requirements,unmanned aerial vehicle(UAV)as auxiliary base station has attracted much attention due to the advantages of fast deployment,high flexibility and line of sight communication.However,there are still some gaps in the research of cooperative caching strategy between aerial caching node and multiple ground caching nodes.At the same time,how to realize fast intelligent path planning for UAV as a resource constrained device also needs to be studied.In view of the above problems,this paper carries out in-depth research on the optimization of caching strategy under various cellular network structures supporting D2 D communication.Caching strategies are designed for single kind and multiple kinds of caching node under the influence of various network performance requirements.Aiming to provide a theoretical basis for the design and optimization of edge caching under different network structures and user requirements.The main work and contributions of this paper are as follows:(1)Aiming at the problems of low content delivery rate and delay demand in D2D-enabled cellular network,a delay minimization caching strategy optimization algorithm based on stochastic geometry theory is proposed,effectively addresses the issues of delayed core network responses to urgent low-latency services in dense user environments and the diminished success rate of content transmission amidst complex interference.Utilizing stochastic geometry,the cumulative interference affecting both D2 D and cellular links is calculated.Subsequent interference analysis facilitates the derivation of the successful transmission probability for various content request methods.On this basis,the service delay expression for obtaining content is defined,and the user device caching strategy optimization problem with minimum delay is established.The user caching strategy is obtained by using the improved particle swarm optimization algorithm.Simulation results show that the proposed algorithm has higher successful offloading rate and lower average delay compared with the representative comparing caching strategy.(2)To address the challenge of network performance in D2D-enabled heterogeneous networks constrained by the task offloading proportion among diverse nodes,an optimization algorithm of transmission cost minimization caching strategy is proposed.The strategy is predicated on the collaborative caching among UDs,small base stations(SBSs),and macro base stations(MBSs),thereby resolving the issue of content offloading allocation in complex caching scenarios involving multiple nodes.Firstly,the algorithm defines the transmission cost of content service based on transmission delay and energy consumption.Considering the randomness of the network and the uncertainty of the signal,the successful transmission probability of each caching node is derived by using stochastic geometry,and the expression of the average transmission cost is derived on this basis.In order to minimize the service transmission cost,the collaborative caching strategy optimization problem of three kinds of caching nodes is established,and the standard gradient method is used to obtain the suboptimal solution.Experimental results show that compared with the high-quality comparison algorithm,the proposed algorithm can significantly reduce the average transmission cost and delay of the network.(3)Aiming at the content delivery problem under multi-channel model in D2D-enabled air to ground network,a caching strategy optimization algorithm based on tiny machine learning(Tiny ML)supporting UDs,SBSs and UAV cooperative caching is proposed,which solves the communication coverage and cooperative cache problem of UAV emergency auxiliary communication.Firstly,the three-dimensional caching network model including UDs,SBSs and UAV is established.Then,the air-to-ground channel model is established,and the cache hit rate and successful transmission probability under different content transmission modes are derived based on the stochastic geometry theory.On this basis,the joint caching strategies of multiple nodes are optimized to maximize the cache hit rate and the successful offloading rate,respectively.In order to ensure the effective coverage of UAV to user nodes,taking into account the limited computing resources and endurance capability of UAV,the UD request probability prediction algorithm is designed based on Tiny ML,and the UAV shortest coverage path planning algorithm is designed according to the location information of users with high request probability,so as to ensure the content acquisition requirements when users initiate requests.The simulation results show that the algorithm of maximizing the successful offloading rate has obvious advantages over the caching strategy of maximizing the cache hit rate without considering the channel quality in terms of successful offloading rate and cache hit rate. |
