CGAM: A community and geography aware mobility model

Summary The advances of localization‐enabled technologies have led to huge volumes of large‐scale human mobility data collected from Call Data Records (CDR), Global Positioning System (GPS) tracking systems, and Location Based Social networks (LBSN). These location data that encompass mobility patterns could generate an important value for building accurate and realistic mobility models and hence important value for fields of application including context‐aware advertising, city‐wide sensing applications, urban planning, and more. In this paper, we investigate the underlying spatio‐temporal and structural properties for human mobility patterns, and propose the Community and Geography Aware Mobility (CGAM) model, which characterizes user mobility knowledge through several properties such as home location distribution, community members' distribution, and radius of gyration. We validate the CGAM synthetic traces against real‐world GPS traces and against the traces generated by the baseline mobility model SMOOTH and assess that CGAM is accurate in predicting the performance of flooding‐based and community‐based routing protocols. We propose the Community and Geography Aware Mobility (CGAM) model, which characterizes user mobility knowledge such as home location distribution, community members' distribution, and radius of gyration. We validate the CGAM traces against GPS, Bluetooth encounters, and SMOOTH traces to assess that CGAM can reproduce spatiotemporal and structural properties for human mobility. Furthermore, the CGAM model can predict the performance of flooding‐based and community‐based routing protocols, in terms of average delay, delivery ratio, delivery cost, and number of hops.