Simulating Realistic User Mobility for Mobile Crowdsensing Using TACSim: A Performance Study

Mobile Crowdsensing (MCS) has recently taken up an important role in sensor data collection paradigms because of its reduced costs and flexibility. It allows crowdsourcers to recruit a number of mobile users to execute sensing tasks in an area without deploying physical sensors. However, MCS algorithms and policies are very different depending on the application and testing them in the real world is impractical, due to the difficulties in recruiting large crowds of volunteers. For this purpose, the research is mostly oriented to simulations, however, to date, there is no simulation platform that focuses enough on different aspects of MCS, often disregarding some in favor of others. In this paper we focus on TACSim, an extensible simulator and we propose an additional mobility module that fills the gap of accurate road network representation. Participants navigate a real road network offering an improved realism and yielding more accurate results. We also propose a caching system that helps in reducing the processing time of simulations and demonstrate its effectiveness through extensive benchmarks.