Performance Analysis And Optimization Of Cache-enabled Ground And Air-ground Wireless Heterogeneous Networks

Heterogeneous networks(HetNets),including terrestrial HetNets and air-ground HetNets,have been recognized as a promising technology to meet the ultra-high capacity requirement in the next-generation networks.On the one hand,the traditional terrestrial HetNets can keep providing high throughput for those scenarios with long-term stable high data requirements.On the other hand,the air-ground HetNets can meet the high data requirement in those temporary hot-spot areas.However,even though HetNets can bring the aforementioned benefits,it still faces some challenges.One is the capacity restriction problem of the wireless backhaul.Compared to the macro base stations(MBSs)which could be connected to the core network through high-capacity optical backhaul,the small base stations(SBSs)are mostly connected to the core networks through low-speed backhaul.Besides,due to the flexibility and mobility of the unmanned aerial vehicles(UAVs),UAVs could not be connected with high-speed wired backhaul.Therefore,the terrestrial SBSs and the UAVs do not own high-speed backhaul links,and the rate of their serve links will be constricted by the rate of their backhaul links.To resolve the backhaul rate restriction problem,caching has received significant attention from the research community and industry.SBSs and UAVs equipped with low-cost cache storage can pro-store those popular contents at the off-peak period,and then directly fulfill users' requests in the peak period without the participation of the backhaul.Thus,caching technology can effectively alleviate the congestion of the backhaul,reduce the content transmission latency and unlock the potentials for the HetNets.In this thesis,I will adopt the random caching scheme,and study the capacity performance of the terrestrial HetNets and the air-ground HetNets,respectively.I will first drive the theoretical capacity expressions via the stochastic geometry tools.Then I will use the convex optimization theory to find the optimal caching placement in the HetNets.This work will reveal the relationship of various system parameters,and show their impact on the capacity of the HetNets.The analytical results can provide design insights for future cache-enabled HetNets.My first work will focus on the capacity of the terrestrial HetNets.I will use Poisson Point Processes(PPPs)to model the locations of the MBSs and the SBSs,respectively.To achieve better performance,I will jointly consider the random caching and the user-centric inter-cell interference nulling.This work will use the content-centric association scheme,which means that the user will be connected to the nearest file-cached base stations.Then,using the tools from the stochastic geometry and probability theory,I will derive the theoretical expressions for the average user achievable rate.Based on the standard optimization theory,I will find the local optimal caching placement in the general cases and the near optimal caching placement in some special cases.The numerical simulation shows that compared to jointing maximal ratio transmission and random caching,jointly considering interference nulling and random caching can achieve a better capacity performance.Besides,the file diversity gain brought by random caching will increase when the SBSs are equipped with more antennas.My second work will focus on the air-ground HetNets.Considering a temporary finite hot-spot area with ultra-high user density,multiple UAVs are deployed to offload the users from a ground MBS.The MBS is equipped with multiple antennas,and it will use SDMA to serve multiple users simultaneously.The single-antenna UAVs have cache space,and their locations will be modeled by the Binomial Point Process(BPP).To capture the impact of the load balance,I will derive the theoretical expressions of the rate coverage probability and the area spectral efficiency(ASE),using the tools from the stochastic geometry.Then I will use convex optimization theory to find the local optimal caching placement in the general cases and the global optimal solutions in the special case.The numerical simulation shows that the HetNets capacity can be efficiently improved by deploying several UAVs temporarily.Meanwhile,increasing the number of antennas can also boost the ASE.Besides,the random cache scheme will degrade to the most popular cache if the number of the UAVs is small.Large UAVs number can help the random cache scheme bring enough file diversity gain and achieve a better ASE.