Bayesian framework for mobility pattern discovery using mobile network events

Abstract

Abstract-Understanding human mobility patterns is of great importance for planning urban and extra-urban spaces and communication infrastructures. The omnipresence of mobile telephony in today's society opens new avenues of discovering the patterns of human mobility by means of analyzing cellular network data. Of particular interest is analyzing passively collected Network Events (NEs) due to their scalability. However, mobility pattern analysis based on network events is challenging because of the coarse granularity of NEs. In this paper, we propose network event-based Bayesian approaches for mobility pattern recognition and reconstruction, mode of transport recognition and modeling the frequent trajectories.

Transport mode detection based on mobile phone network data: A systematic review

“…A summary of the data used in the eligible papers is presented in Table 1. 2010), Calabrese et al (2011), Xu et al (2011, Horn and Kern (2015), Larijani et al (2015), Holleczek et al (2015), Asgari (2016), Poonawala et al (2016), Yamada et al (2016), Danafar et al (2017), Li et al (2017), Hui et al (2017), Hui (2017), Horn et al (2017) As can be seen from Table 1, 15 out of the 22 studies (about 70%) used network-driven data, while the other 7 studies used event-driven data, particularly CDRs. In terms of location estimation, cell of origin method was often mentioned.…”

“…Among them, six papers only employed spatial proximity for mode detection (Doyle et al, 2011;Holleczek et al, 2015;Horn and Kern, 2015;Phithakkitnukoon et al, 2017;Poonawala et al, 2016;Wu et al, 2013), which were mostly for intercity trips or trips between metro/MRT/train stations. The other 10 papers additionally considered other trip-related attributes, such as trip speed (Danafar et al, 2017;Larijani et al, 2015;Smoreda et al, 2013;Yamada et al, 2016), trip distance (Asgari, 2016;Qu et al, 2015), trip duration (García et al, 2016;Li et al, 2017;Schlaich et al, 2010), and temporal overlap with timetables of public transport (Horn et al, 2017;Yamada et al, 2016).…”

“…The third group of methods uses map matching techniques to map a trip to the underlying network. Two different map matching techniques have been applied: Hidden Markov Model (HMM) (Asgari, 2016;Poonawala et al, 2016), and Bayesian inference (Danafar et al, 2017). For example, Asgari (2016) applied HMM-based map matching to find a likely road path and a likely rail path for the original trip trajectory, and checked which of the two paths has a better match with the original trajectory.…”

Mode Choice Latent Class Estimation on Mobile Network Data

“…Huang et al (2019) reviewed studies classifying modes of trips identified in mobile phone network data. Out of the 22 studies found by Huang et al (2019) only four studies separate bus trips from car trips, where Danafar et al (2017), Kalatian and Shafahi (2016) and Phithakkitnukoon et al (2017) mainly focus on short distance trips based either on proximity to route or travel speed, and Wang et al (2010) only consider an example origin-destination pair (OD pair) where there is a clear difference in travel time between car and mass transit. Yang et al (2022) confirm the difficulty of distinguishing bus and car trips based on geospatial information, even in the case of location-based services data (which derived from a combination of sensors such as Wi-Fi, Bluetooth, cellular tower, and GPS information whenever a mobile application updates the phone's location).…”

Mobile phone data in transportation research: methods for benchmarking against other data sources

“…biking and walking trips (Bachir et al 2019)]. Huang et al (2019) perform a systematic literature review of studies using MPD to detect transport modes and find that only one of the analysed papers differentiates between all available transport modes [which in that specific study include car, bus, tram, train, cycling and walking (Danafar et al 2017)], but note that the study does not provide any measure of accuracy of the proposed detection algorithm.…”