Research On Self-Optimization Of Cell Parameters For 5G And Small Cell Deployment In Ultra-Dense Networks

Network planning and optimization can effectively improve coverage rate and provide users with a better communication service experience for 5G networks.With the development of wireless communication,some new technologies will gradually be put into use.However,the challenges brought by these new technologies will also become more complex and diverse.In ultra-dense millimeter-wave networks,on the one hand,millimeter waves provide higher transmission rates and lower latency,but it has significant fading characteristics.On the other hand,the higher density of small cells can cause greater interference and energy efficiency problems.Therefore,only by reasonable planning of the number,location,and distribution of access points can the optimal network topology be determined.In addition,digital twin technology has been widely used in the wireless communication field,which can provide real-time network status data.Self-aware networks can use these data for real-time adaptive optimization of network parameters.Therefore,researching faster network parameter optimization methods and improving optimization efficiency has become more necessary.Firstly,based on the electromagnetic wave propagation principle and ray tracing,a vectorized coverage evaluation method was proposed.Compared with the commonly used grid-based method,this method can significantly reduce the computational complexity in coverage quality evaluation.It can be used for cell parameter optimization in 5G networks and small cell deployment problems in ultra-dense networks.The particle swarm algorithm with adaptive inertia weight was used to optimize the antenna azimuth and the electronic down tilt in a real scene of 1500 meters by 1500 meters.Results showed that the vectorized method can shorten the entire optimization process to 12.55%compared to the grid-based method,and the optimization results of the two methods only differ by 0.16%.Secondly,in terms of cell parameter optimization algorithm,a particle swarm-genetic collaborative evolution algorithm was proposed which considers the distribution characteristics and geometric topology between cell locations.And two geometric partitioning methods,uniform partitioning and Voronoi diagram partitioning,were considered.The algorithm iterated inertia weight adaptation among every subregion and introduced crossover with subregions and mutation in the worst subregion strategies.This method was used to conduct cell parameter optimization tests in a large-scale network.The results showed that this method can effectively converge to better solutions,and the Voronoi diagram partitioning method is more advantageous.In addition,in the test of parameter self-optimization,it can improve the average coverage rate of all users by about 2%than static optimization.Finally,two models were proposed to solve the site selection problem in ultra dense networks,which respectively maximized the line-of-sight(LOS)probability and the vectorized coverage rate.Both used the particle swarm-genetic co-evolutionary algorithm.Site selection tests were conducted in a dense urban environment of 500 meters by 500 meters,and the maximum LOS probability that can be achieved by deploying 10 small cells is 62.19%while 71.28%with 15 small cells.The results of vectorized coverage model showed that deploying 12 small cells at 28GHz can achieve a coverage rate over 90%,and an additional one is required at 39GHz.Both models used building polygon data for coverage evaluation,which greatly reduced the time consumed compared to grid-based methods.In the test,the time consumed by both methods is respectively 10.72%and 11.93%of that by the grid-based model.