Stochastic Analysis on Fog Computing Empowered Mobile Crowdsensing with D2D Communications

Abstract

As a new cloud service provision paradigm, fog computing has revolutionized many mobile computing applications, including mobile crowdsensing (MCS). In this study, we focus on the fog computing empowered MCS with device-to-device (D2D) communications, where both the task dissemination and data collection are deployed as services in the fog nodes (e.g., base stations), and the mobile nodes opportunistically forward both the tasks and the sensing data via D2D communications by epidemic routing. One natural question arose in such scenario is on the affection to the quality-of-sensing (e.g., coverage) from the service deployment. To this end, we are motivated to conduct a stochastic analysis on the D2D-based MCS supported by fog computing. In particular, we use ordinary differential equations (ODEs) to describe the task dissemination phase and the data collection phase and derive the achievable coverage as a function of parameters such as the number of fog nodes exploited, the number of mobile nodes, the encounter rates. We also apply our analysis to find out the optimal time allocation for deadline-constrained MCS applications. Through extensive simulation-based evaluations, we verify the correctness of our analysis, with the average error less than 9.6%, and the optimality of our time allocation scheme.