Population Flow Analysis Based On Mobile Phone Signaling Data

With the development of information technology and the popularization of smart-phones,the massive mobile phone signaling data gathered by base stations play an impor-tant role in some applications such as trajectory data mining,urban planning and popula-tion flow analysis.Population flow analysis is a significant task for municipal planning departments.It is worth because traditional methods have narrower application scope,lower accuracy and higher cost.However,how to acquire the population flow informa-tion accurately becomes a great challenge because of huge data volume,low precision and unstable sampling frequency.In this paper,we propose two algorithms to find the peo-ple entering or leaving the city and analyze the flow of users among districts,namely the Movement Features based Judging Urban Population Flow algorithm(MF-JUPF)and the Base Station State based Finding Districts Migration algorithm(BSS-FDM).To enhance the efficiency and scalability,we design the algorithms based on MapReduce distributed framework to analyze the massive data sets.Extensive experiments upon real data sets il-lustrate the correctness and effectiveness of the proposed algorithms.Main contributions of this paper are listed as followsMobile Phone Signaling Data Processing and Distributed Framework We pro-pose a distributed framework to process the massive mobile phone signaling data.Based on the characteristics of the data,MF-JUPF algorithm and BSS-FDM algo-rithm,we implement the computing framework to proceed data preprocessing and mine the population flow.This framework improves the computing efficiency and scalability.Judging Urban Population Flow We propose the MF-JUPF algorithm to analyze the urban population flow.According to the pattern of entering and leaving the city,we extract some main features from the trajectory including the time of signal disappearance,the probability of appearing in the transportation area,the value of staying in the transportation area,the shortest distance to the border area and the average distance to the residence and workplace.Then various classification models are used to find whether a user comes in or out of the city.Besides,we optimize the query operation from a big table in the distributed scenario.Finding Migration Among Districts The BSS-FDM algorithm is proposed to find the population flow among districts.We design an algorithm to mine users' real destinations based on the spatial distribution and the connection of base stations.In order to deal with complex situations,we offer two strategies to determine which district a stay region belongs to.Eventually,the sequence of districts can be able to reflect the migration.