The Spatial Modeling Of Base Station Location And Traffic Distribution Based On Real Data Of Cellular Networks

With the rapid development of wireless communication and network technology, more mobile service types are emerging and the traffic of wireless networks is exploding. As the performance requirements (e.g. latency, bandwidth) are critical to the various upper-layer services, the networking operators are driven to accelerate the deployment of the base stations to fulfill the coverage and the capacity demand. It is a topic worthy of rethinking a more practical spatial model. Moreover, the spatial distribution of network traffic requires an advanced investigation and careful analyses.The master’s thesis begins with the introduction of the point process theory in stochastic geometry. The complete random Poisson point process, as well as the representative Gibbs point processes and the Neyman-Scott point processes are covered in details. In addition, the fitting methods and the evaluation metrics of the goodness-of-fit in stochastic geometry are listed, followed by the applications in the past researches of wireless communication.The main part of the thesis analyzes the clustering property of base stations based on massive real deployment data. The second-order statistics including L-function and the nearest neighbor distance function, and the coverage probability are utilized to evaluate the accuracy of the point process models and draw conclusions of the Strauss hard-core process in rural area modeling and the Matern clustered process in urban area modeling. In addition, the clustering properties of macro base stations and micro base stations are also included.The analysis of the variation of the spatial traffic in dense urban areas is then introduced. The HTTP traffic correlation among different cells and its negative correlation with cell areas are elaborated. Moreover, the thesis focuses on the similarity between the heterogeneity of the base station distribution and its traffic distribution, based on the analysis of massive macrocell and microcell data.In the end, the thesis points out the future work of the topic, as well as its application areas in wireless communication with the real insights into the spatial data patterns.