Hiroshi Kanasugi

PFlow: Reconstructing People Flow Recycling Large-Scale Social Survey Data

Understanding people flow on a macroscopic scale requires reconstructing it from various forms of existing fragmentary spatiotemporal data. This article illustrates a process for reconstructing such data using existing person-trip survey data.

Understanding User Attributes from Calling Behavior: Exploring Call Detail Records through Field Observations

Mobile phones are arguably one of the most prolific sources of large-scale human mobility data. The availability of this data has generated a massive body of research focused on understanding the dynamics and patterns of human mobility. However, it is increasingly evident that additional value can be derived from such data. This paper proposes a novel approach for understanding the attributes of mobile users by analyzing calling behavior derived from field survey data, in combination with call detail records (CDRs). Our survey reveals distinctive traits in calling behavior that correspond to user attributes. Analysis results demonstrate that frequent call locations, the variability in call time distributions, and the locations from which calls are made around midday are all keys to distinguishing gender. In addition, the location of calls initiated during the morning hours is a key to analyzing income levels for males.

Lightweight lane positioning of vehicles using a smartphone GPS by monitoring the distance from the center line

This paper presents a simple method for using the separation distance (offset) between a smartphone GPS and the center line on a digital road map to determine the lane position of a car. The method was verified.

Spatiotemporal route estimation consistent with human mobility using cellular network data

Continuous personal position information has been attracting attention in a variety of service and research areas. In recent years, many studies have applied the telecommunication histories of mobile phones (CDRs: call detail records) to position acquisition. Although large-scale and long-term data are accumulated from CDRs through everyday use of mobile phones, the spatial resolution of CDRs is lower than that of existing positioning technologies. Therefore, interpolating spatiotemporal positions of such sparse CDRs in accordance with human behavior models will facilitate services and researches. In this paper, we propose a new method to compensate for CDR drawbacks in tracking positions. We generate as many candidate routes as possible in the spatiotemporal domain using trip patterns interpolated using road and railway networks and select the most likely route from them. Trip patterns are feasible combinations between stay places that are detected from individual location histories in CDRs. The most likely route could be estimated through comparing candidate routes to observed CDRs during a target day. We also show the assessment of our method using CDRs and GPS logs obtained in the experimental survey.

Combination of spatio-temporal correction methods using traffic survey data for reconstruction of people flow

Abstract Data on people flow has become increasingly important in various fields, including marketing and public services. Although mobile phones enable the user’s position to be located with a certain degree of accuracy from a large number of people and become one of the most promising devise, unwillingness to share related with privacy issues still remain. Therefore, it is also important to establish a practical method for reconstructing people flow from various kinds of existing fragmentary spatio-temporal data, such as public traffic survey data, from a view of complementariness with mobile phone data. In this study, we propose a combination of spatio-temporal correction processes to a previously published method, to generate continuous spatio-temporal people flow data sets at chosen intervals in selected cities. The correction methods include temporal smoothing of departure time using kernel density estimation, network data correction in OpenStreetMap data, and spatial smoothing in geocoding with MODIS data. We also compare the reconstruction accuracy by deriving correlation coefficients for different combinations of correction methods. Such reconstructed people flow data can potentially be used as infrastructure data in various fields, including emergency planning and related events in areas where data collection and real-time awareness are weak.

Real-time passenger location estimation using CDRs and train objects generated from crowdsourced timetables

This paper shows that cell phone Call Detail Records (CDRs) and train objects in GIS, generated from crowdsourced timetable information, can be used to estimate the train on which a specific passenger is riding. Passenger train ride likelihood calculation is first conducted using CDRs and targeted train objects. Then, the results obtained are compared with the trains specified in GPS logs. Empirical results obtained contain both good cases such as trains being estimated from CDRs and train objects corresponding to the train specified in GPS logs and bad cases such as the estimated train not corresponding to the actual train on which the passenger is riding. Therefore, analysis of each case is also carried out. The application of train objects and cell phone CDRs can facilitate analysis of how passengers ride trains, identification of origin-destination rail routes, and congestion/delay reduction.

Particle filter for real-time human mobility prediction following unprecedented disaster

Real-time estimation of human mobility following a massive disaster will play a crucial role in disaster relief. Because human mobility in massive disasters is quite different from their usual mobility, real-time human location data is necessary for precise estimation. Due to privacy concerns, real-time data is anonymized and a popular form of anonymization is population distribution. In this paper, we aim to estimate human mobility following an unprecedented disaster using such population distribution data. To overcome technical obstacles including high dimensionality, we propose novel particle filter by devising proposal distribution. Our proposal distribution provides states considering both prediction model and acquired observation. Therefore, particles maintain high likelihood. In the experiments, our methods realized more accurate estimation than the baselines, and its estimated mobility was consistent with the survey researches. The computational cost is significantly low enough for real-time operations. The GPS data collected on the day of the Great East Japan Earthquake is used for the evaluation.

Inferencing Human Spatiotemporal Mobility in Greater Maputo via Mobile Phone Big Data Mining

The mobility patterns and trip behavior of people are usually extracted from data collected by traditional survey methods. However, these methods are generally costly and difficult to implement, especially in developing cities with limited resources. The massive amounts of call detail record (CDR) data passively generated by ubiquitous mobile phone usage provide researchers with the opportunity to innovate alternative methods that are inexpensive and easier and faster to implement than traditional methods. This paper proposes a method based on proven techniques to extract the origin–destination (OD) trips from the raw CDR data of mobile phone users and process the data to capture the mobility of those users. The proposed method was applied to 3.4 million mobile phone users over a 12-day period in Mozambique, and the data processed to capture the mobility of people living in the Greater Maputo metropolitan area in different time frames (weekdays and weekends). Subsequently, trip generation maps, attraction maps, and the OD matrix of the study area, which are all practically usable for urban and transportation planning, were generated. Furthermore, spatiotemporal interpolation was applied to all OD trips to reconstruct the population distribution in the study area on an average weekday and weekend. Comparison of the results obtained with actual survey results from the Japan International Cooperation Agency (JICA) indicate that the proposed method achieves acceptable accuracy. The proposed method and study demonstrate the efficacy of mining big data sources, particularly mobile phone CDR data, to infer the spatiotemporal human mobility of people in a city and understand their flow pattern, which is valuable information for city planning.