Generating Synthetic mobile phone datasets using MATSim

Joseph Molloy*, Michael Cik, Martin Fellendorf, Kay W Axhausen

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference paper

Abstract

Passively collected datasets of mobile phone traces are increasingly used for the generation of transportation models. Datasets can contain more than 2000 location events per person per day and can observe hundreds of thousands of participants with no response burden. Hence, such datasets are very attractive for transport modelling, particularly on a regional level. However, privacy regulations make accessing, working with, and sharing such data challenging. We propose an approach for the generation of open, synthetic mobile phone traces, based on a small sample of network traces, information on the location of the network antennas, and activity patterns from a MATSim scenario. Such datasets will allow for better collaboration between researchers on the development of new algorithms for extracting travel plans and other indicators. Previous approaches only generated
synthetic traces for CDR (Call detail record) data, which contains many less data points than traces from 3G and 4G networks. The method accommodates different network types (GSM, 3G, 4G etc), and the introduction of important data artefacts such as pinging and loss of reception. Using the proposed method, a the first steps towards a synthetic network trace dataset for Switzerland calibrated from Austrian network traces is presented
Original languageEnglish
Title of host publication20th Swiss Transport Research Conference
Number of pages21
Publication statusPublished - May 2020
Event20th Swiss Transport Research Conference - Virtuell
Duration: 13 May 202014 May 2020

Conference

Conference20th Swiss Transport Research Conference
Abbreviated titleSTRC 2020
CityVirtuell
Period13/05/2014/05/20

Fields of Expertise

  • Sustainable Systems

Fingerprint Dive into the research topics of 'Generating Synthetic mobile phone datasets using MATSim'. Together they form a unique fingerprint.

Cite this