Stochastic Geometry-based Networking Theory And Method For Next-generation Cellular Networks

The international telecommunication union has classified three application scenarios for nextgeneration of mobile communications,i.e.,enhanced mobile broadband(e MBB),massive machine type communication(m MTC),and ultra-reliable low latency communication(URLLC).Among them,e MBB applications such as video transmission and virtual reality usually have high throughput requirements,while URLLC scenarios such as inter-vehicle security information exchange and equipment communication in industrial scenarios require reliable low latency guarantees.To meet the needs of different applications,it is important to explore advanced networking theory and technologies.For e MBB applications,ultra-dense networks(UDNs)technology can improve network throughput effectively,and edge caching can relieve the backhaul pressure caused by densely deployed base stations.For URLLC applications,device-to-device(D2D)communication is an important way to minimize the air interface latency of vehicular communication,and cell-free massive multiple input multiple output(m MIMO)can achieve low latency in industrial scenarios by improving transmission reliability and reducing the number of retransmissions.However,due to the complex communication environment,applying these technologies faces the challenges of heavy interference,low cache hit ratios,difficult power control design,and poor network-edge performance,respectively.This dissertation addresses the above problems and yields several innovative results.For the complex interference introduced by UDN,we propose a dual-domain interference coordination technique to improve the network throughput.Specifically,to ensure the performance of cell-edge users,fractional frequency reuse is applied in the frequency domain.The available spectrum is divided into interior and edge sub-bands.The latter is dedicated to edge users and reused among cells.To enhance uplink performance,spatial cell cooperation is utilized and neighboring cells are grouped into clusters.The transmission directions of cells within the same cluster are aligned.Based on stochastic geometry and queueing theory,we develop a performance analysis framework for mean packet throughput per user and spatially averaged mean packet throughput under the small cell networks of interest.The proposed framework captures the effects of key network factors such as spatial randomness due to the random locations of network nodes,temporal randomness due to queueing dynamics,inter-cell interference,frequency allocation schemes,and cluster size.Simulation results show that by applying the proposed interference coordination technique,the throughput of edge users is increased effectively,and the uplink throughput can be improved significantly at a cost of a slight loss in downlink throughput through inter-cell collaboration and spectrum resource management.To address the low cache hit ratios caused by heterogeneous user preferences when utilizing caching to alleviate the huge video traffic pressure on network capacity,we propose a joint design of caching,user-side recommendation,and D2 D offloading.The proposed scheme includes two phases: content recommendation and content delivery.In the content recommendation phase,two mobile users recommend the already cached contents to each other based on the utility ranking algorithm.Compared to the base station-side recommendation,users’ personalized preferences and relative locations are fully considered for the user-side strategy,which facilitates the D2 D offloading process.In the content delivery phase,an incentive mechanism based on the amount of transmitted data is designed for operators to motivate idle users to assist in content delivery.Further,we derive a performance analysis framework for the distribution of the available D2 D offloading time and the average amount of offloading data based on stochastic geometry.The proposed framework takes into account multiple key network factors such as the random locations of nodes,random requests,user mobility,users’ selfish attributes,users’ maximum latency requirements,heterogeneous user preferences,incentive mechanisms,as well as protection mechanisms for the existing links.Simulation results show that the joint design of caching,recommendation,and offloading can effectively enhance cache hit ratios and protect the interests of operators,content requesters,and D2 D transmitters.To guarantee the freshness of delivered information in D2 D networks,we propose a distributed power control strategy to improve the age of information(Ao I)performance according to the network topology.Firstly,a performance analysis framework for average Ao I and Ao I violation probability is developed based on stochastic geometry and queueing theory,which can portray the effects of random locations of nodes,random arrival and departure of packets,updating patterns of source nodes,retransmission mechanisms,the density of transmission pairs,interference,and the Ao I violation threshold.Secondly,to optimize the Ao I performance,we study a distributed power control problem to minimize the sum of the average Ao I of all the nodes and propose a locally adaptive power control strategy.In this strategy,each node can determine its own transmit power in a closed-form way based on the locations and activity probabilities of its surrounding nodes.Simulation results show that the proposed scheme can significantly improve the Ao I performance of nodes in D2 D networks.A cell-free m MIMO-based network design approach is proposed to address the challenges of poor network-edge performance and differentiated Ao I requirements of heterogeneous traffic in finite-sized networks such as Industrial Internet of Things networks.Firstly,we design a differentiated period-based frame structure.Devices of each priority all allowed to adjust their own transmission periods according to performance requirements without affecting other devices,which can control interference and achieve coordinated transmission.Secondly,we propose a performance analysis framework for the signal-to-interference ratio meta distribution and average Ao I of nodes at arbitrary locations within the finite-sized networks based on stochastic geometry and queueing theory.The proposed framework covers the effect of factors such as node location,interference,and retransmission on Ao I performance.Finally,to balance the Ao I performance of all the devices,we further study a frame parameter optimization problem,whose objective is minimizing the weighted sum of the mean and variance of the average Ao I of all the devices.We also propose a spatial discretization-based decomposition solution.Simulation results show that the cell-free m MIMO-based network design can meet the differentiated information freshness requirements of devices of different priorities,and the average Ao I of network-edge devices can be guaranteed by the reasonable configuration of frame parameters.