Abstract: Aiming at the calculation and analysis of big data in mobile communication space, the integrated computing platform of ArcGIS and Hadoop is built through Geometry API. COO positioning technology is adopted to collect mobile phone user location data. At the same time, in ArcGIS, Voronoi diagram is used to construct the map information model, and the population density model is built through the density calculation of the circle. Then, the distribution model of the worksite and residence is constructed based on DBSCAN density clustering algorithm. Moreover, the kernel density estimation is used to build the alarm telephone distribution model. The experiment chooses mobile data from Hangzhou Branch of China Mobile Phone between 2017-04 and 2017-06. The result shows that the value of Moran’s I for Hangzhou urban population density is 0.46724. It also evinces that the general characteristics of population distribution is agglomeration. High value agglomeration covers Binjiang, Shangcheng, Xiacheng, as well as parts of Jianggan, Gongshu and Xihu. Thus, the results in this experiment are basically consistent with the results of data analysis of 1% population sampling survey in Hangzhou in 2015. Therefore, the above models are applicable to the spatial and temporal distribution statistics of urban population.

Key words: mobile communication, urban population ;ArcGIS, Hadoop, model, spatial clustering