On The Capacity Of Heterogeneous Cellular Networks

With the rapid development of the popularization of smart terminals and mobile Internet,there is a 1000 growth in mobile service.To meet the requirement of the ever-increasing data traffic,the 3rd Generation Partnership Project(3GPP)proposes the architecture of heterogeneous cellular networks(HCNs)consisted of a macrocell and a set of low-power base stations(BSs)such as relay BSs,picocells,and femtocells.HCNs improve the network capacity by spectrum reuse.The capacity of HCNs is defined as the maximum transmission rate for a randomly chosen user.The investigation of network capacity provides important insights for system design,parameter settings,and protocol design.The main difference between HCNs and traditional cellular networks lies in the co-channel interference.In addition,the interference is temporally and spatially correlated since it comes from the same set of interferers.This makes the analysis of network capacity for HCNs very challenging.Thus,it is important to investigate the network capacity of HCNs.In view of this,the dissertation focuses on the analysis of network capacity for HCNs and reveals the effect of multi-hop transmission,retransmission,and network topology on network capacity.It is worth noting that both the instantaneous characteristic and temporal correlation of interference are considered in this dissertation.The main contributions of this dissertation are summarized as follows:1.The capacity of HCNs under multi-hop transmissions We have quantified the capacity of HCNs under multi-hop transmissions and pointed that,given SIR threshold,multi-hop transmissions improve the coverage but decrease the capacity gain due to the dedicated resource of multi-hop links.Moreover,we have revealed the effect of system parameters on network capacity,which provide significant insights for system design.Based on the multi-tier independent Poisson network model,we have derived the coverage probability and the capacity of HCNs under multi-hop transmissions via the tool of stochastic geometry.Particularly,the capacity gain achieved by multi-hop relaying was quantified for the first time.We have studied the effect of BS density,transmit power,and signal-to-interference-ratio(SIR)threshold on the capacity of HCNs.More importantly,we have obtained the spectral efficiency enhancement condition under which the increase of BS density and transmit power improve the spectral efficiency,thereby enhancing the capacity.One interesting observation is that at a given SINR threshold,the capacity increases with BS density when all the tiers have the same SINR threshold.Moreover,the capacity of some special networks(i.e.heterogeneous cellular networks,multi-hop cellular networks,and conventional cellular networks)are derived directly by specializing some system parameters in our results.2.The capacity of HCNs under HARQ retransmissionsBased on the multi-tier independent Poisson network model,we have quantified the interference correlation in HCNs in terms of coorelation coefficient and found that it only depends on path loss exponent,channel fading,and BS transmit probability.We have also derived the success probability and the delay-limited throughput for Type-Ⅰhybrid automatic repeat request(HARQ)and Type-Ⅱ HARQ with Chase combining.It was shown that the network performance in multiple time slots is decided by the performance in a single time slot and the interference correlation.Although BS density,transmit power,and cell-selection bias do not change the interference correlation,they severely affect the network performance in a single time slot and thus that in multiple time slots.Finally,we have conducted simulations to validate our analysis and pointed that the analysis without considering temporal interference correlation overestimates the performance of HCNs with HARQ.3.The capacity of HCNs under dependent BSs ModelConsidering the intra-tier dependence(Case 1)and both the inter-tier and intra-tier dependence(Case 2),we modeled the small cell BSs in HCNs as a Matern Cluster Process(MCP)and a Second-order Cluster Process(SOCP),respectively.Based on these two models,we have studied the correlation coefficient of interference when the interferers follow MCP and SOCP and found that they are greater than the interference correlation coefficient in the case of Poisson network.We also pointed that the correlation coefficient for the interferers following MCP and SOCP increase with the increase in the mean number of points in each cluster and decrease with the increase in the radius of each cluster R.Then,we have derived the explicit expressions of joint success probability for HCNs in both cases.We have futher provided the upper and lower bounds for joint success probability.In addition,we have obtained that,for macro users,the joint success probability in Case 2 is greater than that in Case 1.Finally,we conduct simulations to validate our analysis.