Urban Spatio-temporal Interaction Networks Analysis And Human Mobility Modeling Study

Cities are typically dynamic and complex systems,and the strong nonlinear interactions among individuals and between individuals and the external environment and the resulting emergent behaviors are important features of complex urban systems.The proper functioning of cities depends on the efficient interaction of a large number of individuals,and thus revealing the universal laws underlying the spatio-temporal interaction patterns among residents is essential for various urban studies,including epidemic transmission,urban planning,and traffic engineering.However,due to the limitation of data collection in the past,it is difficult to obtain individual trajectory and interaction data with high spatial and temporal accuracy.In recent years,with the development of information and communication technology,cell phones have become individual sensors in some sense,capable of objectively recording users’ spatio-temporal location information.Based on the data of a large number of cell phone users in three Asian and African cities,Beijing,Abidjan,and Dakar,this paper mines the dynamic mobile trajectories of users and further constructs spatio-temporal interaction networks on different time slices based on the principle of individual spatio-temporal co-occurrence.In this paper,we find that although individuals are in almost constant movement in cities and hotspot areas attracting large amounts of traffic are changing throughout the day,the Zipf distribution of dynamic population on each location within each hour is very stable and follows a similar pattern across cities,which also indicates that urban systems have approximate spatio-temporal aggregation patterns,in addition,larger cities have stronger spatial heterogeneity.After aggregating the spatio-temporal interaction networks in consecutive unit time windows,we find that the urban system switches between "active" and "sleeping" states: in the "active" state,the residents are concentrated in fewer large communities,while in the " sleeping " state,they are dispersed among more small communities,which is common across cities.Moreover,the more people a city has,the less time it has to be sleeping.In addition,the spatio-temporal interaction network of individuals has an obvious community structure,and individuals within the community have strong spatio-temporal interaction strength.By further analyzing the spatial distribution of individuals’ residences within different communities,this paper finds that in smaller cities,their residences are closer to each other spatially,while in large cities,they may be very dispersed.This suggests that residence patterns in smaller cities can better reflect spatio-temporal interaction patterns,while the relationship between them is less evident in larger cities.Finally,by integrating heterogeneous group travel probability distributions,this paper proposes a temporal population-weighted opportunity model that can well explain the above individual spatio-temporal interaction patterns and provides some new insights into the mechanisms of human mobility research.