Characteristics Of Human Mobility Research Articles

Semantic-Enhanced Graph Convolutional Neural Networks for Multi-Scale Urban Functional-Feature Identification Based on Human Mobility

Precise identification of spatial unit functional features in the city is a pre-condition for urban planning and policy-making. However, inferring unknown attributes of urban spatial units from data mining of spatial interaction remains a challenge in geographic information science. Although neural-network approaches have been widely applied to this field, urban dynamics, spatial semantics, and their relationship with urban functional features have not been deeply discussed. To this end, we proposed semantic-enhanced graph convolutional neural networks (GCNNs) to facilitate the multi-scale embedding of urban spatial units, based on which the identification of urban land use is achieved by leveraging the characteristics of human mobility extracted from the largest mobile phone datasets to date. Given the heterogeneity of multi-modal spatial data, we introduced the combination of a systematic data-alignment method and a generative feature-fusion method for the robust construction of heterogeneous graphs, providing an adaptive solution to improve GCNNs’ performance in node-classification tasks. Our work explicitly examined the scale effect on GCNN backbones, for the first time. The results prove that large-scale tasks are more sensitive to the directionality of spatial interaction, and small-scale tasks are more sensitive to the adjacency of spatial interaction. Quantitative experiments conducted in Shenzhen demonstrate the superior performance of our proposed framework compared to state-of-the-art methods. The best accuracy is achieved by the inductive GraphSAGE model at the scale of 250 m, exceeding the baseline by 25.4%. Furthermore, we innovatively explained the role of spatial-interaction factors in the identification of urban land use through the deep learning method.

Exploring the spatial characteristics of the human mobility network in rural settings of China's Greater Bay Area

Human mobility is an important carrier of material flow, information flow, capital flow, and technology flow between locations. A better understanding of human mobility in rural areas could inform efforts to achieve rural revitalization. However, there have been limited evaluations of the characteristics of human mobility in rural areas. Based on human movements estimated from mobile phone data, we constructed an inter-town human mobility network and applied a complex network analysis method to study the spatial characteristics of the human mobility network in rural areas, with China's Greater Bay Area as a case study. We found that rural towns with high centrality were mostly located at the convergence of developed megacities. Rural towns with most connections were not necessarily the most central in the inter-town human mobility networks, leading to anomalous values of the centrality. We demonstrated that the anomaly of centrality values emerged due to the community structure of the network. We identified nine communities in the network and found that the geographical borders of communities partly matched the urban administrative borders. However, the results also showed that many rural towns on the fringes of a city were dissolved in their adjacent developed cities, independent of the urban administrative boundary effect. The formation of communities cannot be explained only by the impacts of urban administrative boundaries, but also by the growth of large cities and the geographic dispersal of economic activities. Finally, we identified seven roles of rural towns based on their pattern of intra-community and inter-community connections. These findings provide important insights into the pattern of interaction among rural towns.

Epidemic thresholds and human mobility

A comprehensive view of disease epidemics demands a deep understanding of the complex interplay between human behaviour and infectious diseases. Here, we propose a flexible modelling framework that brings conclusions about the influence of human mobility and disease transmission on early epidemic growth, with applicability in outbreak preparedness. We use random matrix theory to compute an epidemic threshold, equivalent to the basic reproduction number R_{0}, for a SIR metapopulation model. The model includes both systematic and random features of human mobility. Variations in disease transmission rates, mobility modes (i.e. commuting and migration), and connectivity strengths determine the threshold value and whether or not a disease may potentially establish in the population, as well as the local incidence distribution.

Enhancing short-term crime prediction with human mobility flows and deep learning architectures

Place-based short-term crime prediction models leverage the spatio-temporal patterns of historical crimes to predict aggregate volumes of crime incidents at specific locations over time. Under the umbrella of the crime opportunity theory, that suggests that human mobility can play a role in crime generation, increasing attention has been paid to the predictive power of human mobility in place-based short-term crime models. Researchers have used call detail records (CDR), data from location-based services such as Foursquare or from social media to characterize human mobility; and have shown that mobility metrics, together with historical crime data, can improve short-term crime prediction accuracy. In this paper, we propose to use a publicly available fine-grained human mobility dataset from a location intelligence company to explore the effects of human mobility features on short-term crime prediction. For that purpose, we conduct a comprehensive evaluation across multiple cities with diverse demographic characteristics, different types of crimes and various deep learning models; and we show that adding human mobility flow features to historical crimes can improve the F1 scores for a variety of neural crime prediction models across cities and types of crimes, with improvements ranging from 2% to 7%. Our analysis also shows that some neural architectures can slightly improve the crime prediction performance when compared to non-neural regression models by at most 2%.

Characterizing pandemic waves: A latent class analysis of COVID-19 spread across US counties.

The spread of the COVID-19 pandemic is observed to follow the shape of “waves” (i.e., the rise and fall of population-adjusted daily new infection cases with time). Different geographic regions of the world have experienced different position and span of these waves over time. The presence and strength of these waves broadly characterize the dynamics of the pandemic spread in a given area, so their characterization is important to draw meaningful intervention and mitigation plans tailored for that area. In this paper, we propose a novel technique to represent the trend of COVID-19 spread as a sequence of a fixed-length text string defined on three symbols: R (rise), S (Steady), and F (fall). These strings, termed as trend strings, enabled us searching for specific patterns in them (such as for waves). After analyzing county-level infection data, we observe that, US counties—despite their wide variation in trend strings—can be grouped into a number of heterogeneous classes each of which might have a representative COVID spread pattern over time (in terms of presence and propensity of waves). To this end, we conduct a latent class analysis to cluster 3142 US counties into four distinct classes based on their wave characteristics for one year pandemic data (January 2020 to January 2021). We observe that counties in each class have distinct socio-demographics, location, and human mobility characteristics. In short summary, counties have differing number of waves (class 1 counties have only one wave and class 3 counties have three) and their positions also vary (class 1 had the wave later in the year whereas class 3 had waves throughout the year). We believe that this way of characterizing pandemic waves would provide better insights in understanding the complex dynamics of COVID-19 spread and its future evolution, and would, therefore, help in taking class-specific policy interventions.

A vector field approach for identifying anomalous human mobility

AbstractAlthough ordinary human mobility has been extensively studied, anomalous human mobility during emergencies or mass events is not sufficiently understood. The recently proposed vector field approach has shed light on human mobility studies. Here, the authors improve it to analyze anomalous human mobility in mass events. Specifically, the authors develop the anomalous field, the source field, and the dispersion field to identify the crowd gathering location, the start time, and the end time of anomalous human mobility. In addition, the authors propose the decay coefficient and the maximum distance to quantify the influence degree and scope of a mass event. The present approach can be used to capture the spatiotemporal characteristics of human mobility in mass events.

Optimizing Electric Vehicle Charging Schedules Based on Probabilistic Forecast of Individual Mobility

Abstract. The number of electric vehicles (EVs) has been rapidly increasing over the last decade, motivated by the effort to decrease greenhouse gas emissions and the fast development of battery technology. This trend challenges distribution grids since EVs will bring significant stress if the charging of many EVs is not coordinated. Among the many strategies to cope with this challenge, next-day EV energy demand forecasting plays a key role. Existing studies have focused on predicting the next-day energy demand of EVs on the aggregated and individual levels. However, these studies have not yet extensively considered individual user mobility behaviors, which exhibit a high level of predictability. In this study, we consider several mobility features of individual users when forecasting the next-day energy demand of individual EVs. Three types of quantile regression models are used to generate probabilistic forecasts of energy demand, particularly the next-day energy consumption and parking duration. Based on the prediction results, two time-shifting smart charging strategies are designed: unidirectional and bidirectional smart charging. These two strategies are compared with an uncontrolled charging baseline to evaluate their financial benefits and peak-shaving effects. Our results show that human mobility features can partially improve the prediction of next-day individual EV energy demand. Additionally, users and distribution grids can benefit from smart charging strategies both financially and technically.

On building human-aware opportunistic communication strategies for cost-effective content delivery

Mobile networks face challenges such as data traffic growth, increasing connected devices, spectrum limitations, and costly infrastructure updates. In this context, opportunistic communication strategies appear to assist in scenarios such as data offloading, expanding the capillarity given by users’ mobility, delivering content in challenging situations (e.g., emergency, rural areas, or crowded places), and fostering innovative applications. State-of-art opportunistic strategies extract human mobility characteristics (e.g., social and spatial) to improve content delivery’s cost-effectiveness. Nevertheless, due to constraints, such as the limited availability of real-world datasets or the lack of a human-centered vision in networking, most previous works relied on traditional or simplified human-aware metrics to reach their purposes. Next-generation proposals need a more in-depth vision of peculiar aspects hidden into mobility datasets while working with more realistic scenarios. While building TOOTS, a novel human-aware opportunistic communication strategy for improved content delivery cost-effectiveness, we learned that several phases play a significant role in reaching strategy’s superior performance. This work guides the reader through the whole process necessary for building the Tactful Opportunistic Communication Strategy (TOOTS). The proposal consists of the following steps: learning human-aspect best practices from state-of-art; characterizing the traces while obtaining strategy’s insights targeting superior performance; using, proposing, and analyzing human-aware metrics; combining such metrics and insights into a complete strategy for cost-effective content delivery in a more realistic scenario, and finally, evaluating the strategy through discussions of its best practices and shortcomings for future work. This work shows that TOOTS can improve the performance of an opportunistic scenario in terms of overhead, delivery rate, and delivery latency by following this proposed process.

Escape Path Obstacle-Based Mobility Model (EPOM) for Campus Delay-Tolerant Network

In Delay-Tolerant Networks (DTNs), humans are the main carriers of mobile devices, signifying that human mobility can be exploited by extracting nodes’ interests, social behavior, and spatiotemporal features for the performance evaluation of DTNs protocols. This paper presents a new mobility model that describes students’ daily activities in a campus environment. Unlike the conventional random walk models, which use a free space environment, our model includes a collision-avoidance technique that generates an escape path upon encountering obstacles of different shapes and sizes that obstruct pedestrian movement. We evaluate the model’s usefulness by comparing the distributions of its synthetic traces with realistic traces in terms of spatial, temporal, and connectivity features of human mobility. Similarly, we analyze the concept of dynamic movement clusters observed on the location-based trajectories of the studied real traces. The model synthetically generates traces with the distribution of the intercluster travel distance, intracluster travel distance, direction of movement, contact duration, intercontact time, and pause time similar to the distribution of real traces.

Based on Temporal and Spatial Traffic Characteristics for Energy-Efficient Wi-Fi Network

With the rapid development of information and communication technology (ICT) industry, the energy consumption caused by the development of mobile communications is increasing exponentially. Huge energy consumption and increasing environmental awareness prompt people to study the dynamic distribution of users and traffic in Wi-Fi networks in order to design an energy-efficient mechanism for Wi-Fi networks. Based on real data, this paper explores the spatial and temporal characteristics of human mobility and data traffic distribution. The results of access point (AP) traffic analysis show a tidal effect and obvious periodicity in time and space dimension, which can be used for the dynamic design of AP sleep or shutdown strategy. At the same time, we design a new service area (SA) zooming and AP cooperation mode. Through the simulation experiment, we find that the energy consumption can be reduced by about 35% and the communication service quality can be guaranteed.

Sensor location design for interdicting mobile travelers with probabilistic space-time trajectories

Most existing sensor location design studies focus on stationary or deterministic objects (e.g., fixed OD demand) yet have not well investigated uncertain mobility of inspected objects. To bridge this gap, this study aims to integrate human mobility characteristics specified by the time geography theory into sensor location design. We first investigate trajectories of moving objects to see how they can reveal detailed information about human travel characteristics and presence probability with high-resolution detail. Then, a space–time network-based modeling framework is proposed to integrate human mobility into network location design problems. We construct a probabilistic network structure to quantify human’s presence probability at different locations and time. Then, a Mixed Integer Nonlinear Programming (MINLP) model is proposed to maximize the spatial and temporal coverage of individual targets. To obtain near optimal solutions for large-scale problems, greedy heuristic, simulated annealing and Lagrangian relaxation algorithms are tested. Theoretical analysis is conducted to show the optimality gap of the greedy algorithm. The proposed algorithms are implemented on hypothetical and real-world numerical examples (some on a high-performance computing platform) to demonstrate the performance and effectiveness of the proposed model on different network sizes and promising results have been obtained.

Do Individuals' Activity Structures Influence Their PM2.5 Exposure Levels? Evidence from Human Trajectory Data in Wuhan City.

Severe air pollution has become a major risk to human health from a global environmental perspective. It has been recognized that human mobility is an essential component in individual exposure assessment. Activity structure reflects the characteristics of human mobility. Thus, a better understanding of the relationship between human activity structure and individual exposure level is of crucial relevance. This study examines this relationship using a large cell-phone GPS dataset in Wuhan, China. The results indicate that there is a strong linear relationship between people’s activity structures and exposures to PM2.5. Inter-group comparisons based on the four activity structure groups obtained with K-means clustering found that groups with different activity structures do experience different levels of PM2.5 exposure. Furthermore, differences in detailed characteristics of activity structure were also found at different exposure levels at the intra-group level. These results show that people’s activity structures do influence their exposure levels. The paper provides a new perspective for understanding individual exposure through human activity structure, which helps move the perspective of research on individual exposure from the semantic of physical location to the semantic of human activity pattern.

Temporal and Spatial Traffic Analysis Based on Human Mobility for Energy Efficient Cellular Network

With the drastic growth of Information and Communication Technology (ICT) industry, global energy consumption is exponentially increased by mobile communications. The huge energy consumption and increased environmental awareness have triggered great interests on the research of dynamic distribution of cell user and traffic, and then designing the energy efficient cellular network. In this paper, we explore the temporal and spatial characteristics of human mobility and traffic distribution using real data set. The analysis results of cell traffic illustrate the tidal effect in temporal and spatial dimensions and obvious periodic characteristics which can be used to design Base Station (BS) dynamic with sleeping or shut-down strategy. At the same time, we designed a new Cell Zooming and BS cooperation mode. Through simulation experiments, we found that running in this mode can save about 35% of energy consumption and guarantee the required quality of service.

Exploring the relationship between functional urban polycentricity and the regional characteristics of human mobility: A multi-view analysis in the Tokyo metropolitan area

Recent years have witnessed the academic interests of researchers in the topic of urban polycentricity. However, most previous studies have failed to the investigate the fine-scale regional characteristics of polycentricity and to examine how they vary with different urban functions at a metropolis scale. To fill this gap, this study measures the functional urban polycentricity of the Tokyo metropolitan area (TMA) at a fine scale and uncovers its relationship with the regional characteristics of human mobility patterns using a series of exploratory statistical analyses. The findings reveal a static dynamic pattern of the functional urban structure within an original multi-view analytical framework. Results prove that the relationship within each functional urban area reflects different spatiotemporal characteristics depending on its functional allocation and development level under TMA's four-level annular concentric layout. Functional complementarity and temporal symmetry can also be observed. Our achievements can not only provide rich information regarding the investigation of functional urban structure patterns but also allow urban planners to rethink the rationality and scientificity of the existing master plan.

GCN2CDD: A Commercial District Discovery Framework via Embedding Space Clustering on Graph Convolution Networks

Modern enterprises attach much attention to the selection of commercial locations. With the rapid development of urban data and machine learning, we can discover the patterns of human mobility with these data and technology to guide commercial district discovery. In this article, we propose an unsupervised commercial district discovery framework via embedding space clustering on graph convolution networks to solve the problem of commercial district discovery. Specifically, the proposed framework aggregates human mobility features according to geographic similarity by graph convolution networks. Based on the graph convolution network embedding space, we apply hierarchical clustering to mine the latent functional regions hidden in different human patterns. Then, with the kernel density estimation, we can obtain the semantic labels for the clustering results to discover commercial districts. Finally, we analyze the multisource data of the Xiaoshan District and Chengdu City, and experiments verify the effectiveness of our framework.

Changing travel patterns in China during the early stages of the COVID-19 pandemic

Understanding changes in human mobility in the early stages of the COVID-19 pandemic is crucial for assessing the impacts of travel restrictions designed to reduce disease spread. Here, relying on data from mainland China, we investigate the spatio-temporal characteristics of human mobility between 1st January and 1st March 2020, and discuss their public health implications. An outbound travel surge from Wuhan before travel restrictions were implemented was also observed across China due to the Lunar New Year, indicating that holiday travel may have played a larger role in mobility changes compared to impending travel restrictions. Holiday travel also shifted healthcare pressure related to COVID-19 towards locations with lower healthcare capacity. Network analyses showed no sign of major changes in the transportation network after Lunar New Year. Changes observed were temporary and did not lead to structural reorganisation of the transportation network during the study period.

From cellular positioning data to trajectories: Steps towards a more accurate mobility exploration

The recent years have witnessed a greater demand for understanding how people move in urban environments. Due to the widespread usage of mobile phones, there have been several trajectory-based studies focusing on extracting the characteristics of human mobility from georeferenced mobile phone data. Mobile positioning data is generally generated as scattered points in CDRs (Call Detail Records). Even though CDR data can be regarded as an inexpensive scalable source of information on human mobility, mobility studies in urban settings based on such data sources still prove to be a research challenge due to the coarseness of CDR spatial granularity. Motivated by the need for transforming large-scale CDRs to movement trajectories, the present study offers a new solution which is made of two principal building blocks: (1) Developing a Bayesian-based induction method through adopting a GIS-based wave propagation model to solve the GSM-based localization problem when methods such as triangulation are not applicable due to the lack of measurements from more than one base station; (2) Reconstruction of movement trajectories from cellular location information using overlapping relations existing between observed cells as well as detection of ping-pong phenomena as auxiliary information. A case study employing CDR and GPS records obtained from an experimental survey on one of the central urban zones of Tehran was conducted, which showed the effectiveness of the proposed methodology in comparison to current approaches with respect to three perspectives, including movement path exploration, individual-oriented movement features extraction, and crowd-movement modelling.

A New Approach to Refining Land Use Types: Predicting Point-of-Interest Categories Using Weibo Check-in Data

The information of land use plays an important role in urban planning and optimizing the allocation of resources. However, traditional land use classification is imprecise. For instance, the type of commercial land is highly filled with the categories of shopping, eating, etc. The number of mixed-use lands is increasingly growing nowadays, and these lands sometimes are too mixed to be well investigated by conventional approaches such as remote sensing technology. To address this issue, we used a new social sensing approach to classify land use according to human mobility and activity patterns. Previous studies used other social sensing approaches to predict land use types at the parcel or the area level, whilst fine-grained point-of-interest (POI)-level land use data are likely to more useful in urban planning. To abridge this research gap, we proposed a new social sensing approach dedicated to classifying land use at a finer scale (i.e., POI-level or building level) according to human mobility and activity patterns reflected by location-based social network (LBSN) data. Specifically, we firstly investigated spatial and temporal patterns of human mobility and activity behavior using check-in data from a popular Chinese LBSN named Sina Weibo and subsequently applied those patterns to predicting the category of POI to refine urban land use classification in Guangzhou, China. In this study, we applied three classification methods (i.e., naive Bayes, support vector machines, and random forest) to recognize category of a certain POI by spatial and temporal features of human mobility and activity behavior as well as POIs’ locational characteristics. Random forest outperformed the other two methods and obtained an overall accuracy of 72.21%. Apart from that, we compared the results of the different rules in filtering check-in samples. The comparison results show that a reasonable rule to select samples is essential for predicting the category of POI. Moreover, the approach proposed in this study can be potentially applied to identifying functions of buildings according to visitors’ mobility and activity behavior and buildings’ locational characteristics.

Research on Collective Human Mobility in Shanghai Based on Cell Phone Data

The high-frequency mobility of a massive population has caused an enormous influence on the urban internal structure, which is unable to be described by traditional data sources. While recent advances in location-based technologies provides new opportunities for researchers to understand daily human movements and the structure as a whole. The article aims to explore human spatial movements and their aggregate distribution in Shanghai using large-scale cell phone data. The trajectory of each individual is extracted from cell phone data after data cleansing. Then, an indicator system which includes mobility intensity, mobility stability, influential range, and temporal variation is developed to describe collective human mobility features in census tracts scale. Finally, spatial elements are extracted using the indicator system and the structure of human mobility in Shanghai is discussed.

Similarity measurement on human mobility data with spatially weighted structural similarity index (SpSSIM)

AbstractUnderstanding diverse characteristics of human mobility provides profound knowledge of urban dynamics and complexity. Human movements are recorded in a variety of data sources and each describes unique mobility characteristics. Revealing similarity and difference in mobility data sources facilitates grasping comprehensive human mobility patterns. This study introduces a new method to measure similarities on two origin–destination (OD) matrices by spatially extending an image‐assessment tool, the structural similarity index (SSIM). The new measurement, spatially weighted SSIM (SpSSIM), utilizes weight matrices to overcome the SSIM sensitivity issue due to the ordering of OD pairs by explicitly defining spatial adjacency. To evaluate SpSSIM, we compared performances between SSIM and SpSSIM with resampling the orders of OD pairs and conducted bootstrapping to test the statistical significance of SpSSIM. As a case study, we compared OD matrices generated from three data sources in San Diego County, CA: U.S. Census‐based Longitudinal Employer–Household Dynamics Origin–Destination employment statistics, Twitter, and Instagram. The case study demonstrated that SpSSIM was able to capture similarities of mobility patterns between datasets that varied by distance. Some regions showed local dissimilarity while the global index indicated they were similar. The results enhance the understanding of complex mobility patterns from various datasets, including social media.