System Performance Analysis And Key Caching Technologies Of Ultra-dense Networks

The evolution of existing mobile applications and the emergence of new applications have led to explosive growth of wireless data traffic and diversified service quality requirements for different types of data traffic.The limited frequency spectrum of lower frequency bands such as sub-6GHz can no longer meet the demand for continuous growth of various services and applications.Utilizing higher frequency bands,such as millimeter wave(mm Wave)or Terhertz,has become an inevitable trend.On the other hand,ultra-dense deployment of base stations(BSs)is an other effective approach to increase the network data transmission rate.However,the ultra-dense networks(UDN)have the problems of unbalaced spatio-temporal distribution of traffic,high energy consumption and severe interference.Aiming at these problems,this paper conducts in-depth research on the fine-grained performance and key caching technologies of UDN.Firstly,aiming at the queue interaction in the cell-center/edge areas of the wireless networks in the scenario of dynamic arrival of traffic,a location-dependent meta distribution-based framework is proposed.Based on the distance ratio between the user’s serving BS and the dominant interfering BS,the users are divided into cell center users and cell edge users,and the probability of the cell-center/edge users is derived.The statistical characterization of the conditional successful transmission probabilities(STPs)of the cell-center/edge users are obtained,and then the proportion of non-empty queues in the cell-center/edge area is obtained.The fixed-point equations on the meta-distribution are established for the cell-center/edge users,respectively.The corresponding mean local delays are analyzed.In addition,the relationship between the packet arrival rate and the active probability of the BSs is analyzed.The analysis framework is further generalized to the scenario where each cell is divided into multiple areas according to the service distance of the typical user,and the distribution of link reliability in each area is obtained.Simulation results demonstrate that the mean local delay increases when increasing the active probability for both cell-center and cell-edge users.Given the same arrival rate,the active probability of a cell-edge user queue is larger than that of a CCU queue.Secondly,aiming at the problem of the couping between inter-tier interference and queue status in the mm Wave heterogeneous networks with the dynamic arrival of traffic,a meta distribution-based framework of SINR for the mm Wave heterogeneous networks is proposed.By using stochastic geometry and queuing theory,the statistical characterization of the conditional STP is obtained while the special characteristics of mm Wave communications,including different path loss laws for line-of-sight(LOS)and non-line-of-sight(NLOS)links and directional beamforming,are incorporated into the analysis.A fixed-point equation for the meta distribution is formulated and the numerical solution to the meta distribution is obtained following a recursive approach.In addition,the steady-state analysis of the mm Wave heterogeneous networks is carried out,where the critical value of the packet arrival rate is analyzed and the mean delay of the network is derived.The simulation results reveal that offloading the users to the small BS(SBS)tier by biasing can reduce the performance fluctuation while only causing a slight degradation of the overall network performance.Compared to the mm Wave networks without queueing,increasing the BS density is more beneficial and the deterioration caused by increasing the bias factor is more evident for the performance of the networks with queueing.Moreover,the effect of the packet arrival rate on the active probability of the BSs is more evident under high blockage parameter regime.Then,aiming at the problem of high energy consumption in the dense deployment of BSs,a fine-grained analysis framework for the intelligent reflective surface(IRS)-assisted mm Wave networks is proposed.The probability of providing a reflected link for the typical user is obtained by considering the orientation of the IRSs.Given the location distribution of the BS,the intensity measure of the location distribution of the IRSs is obtained.In addition,the distribution of the distance between the typical user and the serving BS or associated RIS is derived.According to the status of the service link,a flexible beamforming scheme is provided.The moments of the conditional STP,the analytical expressions of the average local delay,area spectral efficiency(ASE)and energy efficiency are obtained.Simulation results demonstrate the increasing the key network parameters such as BS/RIS density and RIS size results in better STP,ASE,and energy efficiency.In addition,the deployment of the RISs can ratchet up the cell-edge rate,hence confirms the effectiveness of deploying RISs in the next generation wireless networks.Afterwards,for the optimization of random caching strategy in mm Wave heterogeneous networks,a general mathematical framework for analyzing mm Wave cache-enabled heterogeneous networks is proposed and a file caching probability optimization problem is solved by utilizing the convex optimization.By using the LOS probability function to model the LOS/NLOS link states,the association probability between a typical user and its serving LOS/NLOS BS in any tier is analyzed,providing a closed-form expression for the association probability with sufficiently small barrier parameters for all tiers.In addition,the Laplace transform of the interference for each tier is derived and the STP of the mm Wave cache-enabled heterogeneous networks is obtained.A closed-form expression for the STP is provided for all tier blockage parameters that are sufficiently small.For the general case,the gradient projection method and the convex-concave procedure(CCCP)algorithm are used to obtain the local optimal cache probability.Under the case where the blockage parameter is sufficiently small,the approximate optimal solution is obtained by using the KKT conditions and the CCCP algorithm.The simulation results show the influence of the key network parameters on the system performance,and verify the superiority of the proposed algorithm compared to the popular baseline algorithms.Finally,for the problem of the interference management in the cache-enabled wireless networks,a meta distribution-based analysis framework is applied to the optimization of the random caching in the cache-enabled networks with random discontinuous transmission(DTX)and a joint BS active probability and file caching probability optimization problem is solved by utilizing the convex optimization and considering the backhaul delay.By using stochastic geometry,a fine-grained general analysis framework of the cache-enabled network is established.The moments of the conditional STP,the SIR meta distribution are derived,respectively.In addition,the mean local delay and jitter in the cache-enabled networks are derived.Given the BS active probability,the critical value of file caching probability is obtained.The numerical simulations show the effect of file caching probability and BS active probability on mean local delay and jitter under different path loss exponents and SIR thresholds as well as the effect of network parameters on the critical value of file caching probability or BS active probability.Moreover,the maximization of STP and the minimization of average system transmission delay are considered,respectively.For the former,a convex optimization problem is formulated to maximize the STP under the BS cache size constraint.For the latter,a non-convex optimization problem is formulated to minimize the average system transmission delay by alternately optimizing the BS active probability and the file cache probability.Finally,a comprehensive simulation and analysis of the impact of key parameters on network performance is carried out: the numerical simulation results show the impact of key network parameters,such as BS active probability,file cache probability and SIR threshold,on the SINR meta distribution and mean local delay.In addition,the effect of backhaul delay,cache size and Zipf exponent on the average system transmission delay under different caching strategies are investigated.The network performance benefits from caching the most popular files when the backhaul delay is small,while benefiting from a larger file diversity when the backhaul delay is large.